The more, the better? Learning with feedback and instruction
越多越好?通过反馈和指导学习
教育学TOPESI学科分类:一般社会科学JCI 2.67IF(5) 6.1SCI升级版 教育学1区SSCI Q1IF 4.7CUG 教育心理学T1RUC BSDUFE A2UIBE BSWUFE AKeywords 关键词
1. Introduction 1.导言
In the course of the digital transformation, online learning environments have become paramount in educational settings. Two frequently applied instructional interventions to support students' learning in online environments are strategy instruction and formative feedback in subsequent practice phases. For instance, in the domain of electric circuits troubleshooting, strategy instruction is given by providing a conceptional explanation about Ohm's law together with worked examples of how to solve electric circuit problems to help students gain a robust understanding of a problem-solving strategy with its different solution steps (Renkl, 2014; van Gog et al., 2006, van Gog et al., 2011; van Harsel et al., 2020). Additional practice tasks in combination with feedback should help apply and transfer these strategies to different, yet related problem-situations (e.g., Fyfe & Rittle-Johnson, 2016; Huitt et al., 2009). From a pragmatic stance, it makes sense to combine strategy instruction and feedback to maximize learning which is also reflected in state-of-the-art instructional design models (e.g., Huitt et al., 2009; van Merriënboer et al., 2002). Previous research, which directly examined combinations of strategy instruction and feedback, however, is scarce and, if available, produced mixed findings as the combination of strategy instruction and feedback yielded both additive (e.g., Salden et al., 2010) and reducing effects on learning (e.g., Fyfe & Rittle-Johnson, 2016).
在数字化转型的过程中,在线学习环境在教育环境中变得至关重要。在在线环境中,支持学生学习的两种常用教学干预措施是策略指导和后续练习阶段的形成性反馈。例如,在电路故障排除领域,策略指导的方式是提供有关欧姆定律的概念性解释以及如何解决电路问题的工作示例,以帮助学生牢固理解解决问题的策略及其不同的解决步骤(Renkl, 2014;van Gog et al.,2006, van Gog et al、2011; van Harsel 等人,2020)。额外的练习任务与反馈相结合,应有助于将这些策略应用和迁移到不同但相关的问题情境中(例如Fyfe & Rittle-Johnson, 2016;Huitt 等人,2009)。从实用的角度来看,将策略指导和反馈结合起来以最大限度地提高学习效果是有道理的,这也反映在最先进的教学设计模式中(如Huitt et al.,2009;van Merriënboer et al.,2002)。然而,以前直接研究策略指导与反馈相结合的研究很少,即使有,也是结果不一,因为策略指导与反馈相结合产生的结果都是相加的(如Salden et al、Fyfe & Rittle-Johnson, 2016)。
In this paper, we present data from three well-powered online experiments (N1 = 437, N2 = 310, N3 = 166), in which we systematically investigated effects of combining strategy instruction and feedback on students’ learning. To test the generalizability of our findings, we used different feedback formats (Experiment 1: corrective feedback, Experiment 2: elaborated feedback), and experimentally varied the sequence of strategy instruction and feedback to rule out potential sequence effect of their combinations (Experiment 3). Additionally, in cases of reducing or non-additive effects (i.e., an instruction-by-feedback interaction), we investigated the underlying processes of combining strategy instruction and feedback both from a cognitive, a metacognitive and an affective-motivational perspective (Narciss, 2008).
在本文中,我们展示了来自三项有实力的在线实验的数据(N1 = 437、N2 = 310、N3 = 166)、N3 = 166),其中我们系统地研究了策略指导与反馈相结合对学生学习的影响。为了检验研究结果的可推广性,我们使用了不同的反馈形式(实验 1:纠正性反馈,实验 2:详细反馈),并通过实验改变了策略指导和反馈的顺序,以排除两者结合可能产生的顺序效应(实验 3)。此外,在减少或非相加效应(即指导与反馈的相互作用)的情况下,我们从认知、元认知和情感动机的角度研究了策略指导与反馈相结合的潜在过程(Narciss, 2008)。
1.1. Strategy instruction
1.1.战略指导
Strategy instruction is regarded as an effective means to foster cognitive skill acquisition (Fyfe & Rittle-Johnson, 2016; Goldman, 1989). During strategy instruction the students are explicitly taught the subject matter on how to apply distinct strategies to support problem-solving. To this end, strategy instruction often encompasses the use of worked examples (e.g., Fyfe & Rittle-Johnson, 2016; Goldman, 1989; Kant et al., 2017; Salden et al., 2010). Students first receive a short conceptual introduction that provides the basic conceptual knowledge referring to the underlying principles and concepts of the problem (e.g., Ohm's law in the domain of physics; Renkl, 2005; Wittwer & Renkl, 2008). Afterwards, students are provided with worked examples portraying the different solution steps leading to the final solution whereby their design follows principles of example selection and sequencing (Atkinson et al., 2000; Kant et al., 2017; Renkl, 2014; van Gog & Rummel, 2010).
策略指导被认为是促进认知技能学习的有效手段(Fyfe &;Rittle-Johnson, 2016; Goldman, 1989)。在策略教学过程中,学生会明确学习到如何运用不同的策略来解决问题。为此,策略教学通常包括使用工作实例(如Fyfe & Rittle-Johnson, 2016;Goldman, 1989; Kant et al.,2017;Salden 等人,2010)。学生首先会收到一个简短的概念介绍,提供与问题的基本原理和概念相关的基本概念知识(例如,在 "欧姆定律 "领域中的欧姆定律)、物理领域的欧姆定律;Renkl, 2005;Wittwer & Renkl, 2008)。随后,为学生提供工作示例,描绘最终解决方案的不同解决步骤,学生的设计遵循示例选择和排序原则(Atkinson et al、2000; Kant et al.,2017 年;Renkl, 2014 年;van Gog & Rummel, 2010 年)。
From a cognitive load perspective, providing worked examples should reduce extraneous processing so that cognitive capacity is freed for germane cognitive processes, such as conceptual learning and elaboration (Kant et al., 2017; Renkl, 2014; van Gog & Paas, 2008). The effectiveness of instruction based on example-based learning is well-documented in several empirical studies (van Gog & Paas, 2008; see also Crissman, 2006; Hoogerheide & Roelle, 2020; Kirschner et al., 2006; Magliaro et al., 2005; Renkl, 1997, 2014, for empirical evidence). A recent meta-analysis in the domain of mathematics (Barbieri et al., 2023) also documented the benefits of worked examples compared to no-examples (g = 0.48), based on 181 effect sizes.
从认知负荷的角度来看,提供有效的示例应能减少无关的处理,从而将认知能力释放给重要的认知过程,如概念学习和阐述(Kant et al、2017; Renkl, 2014;van Gog & Paas, 2008)。一些实证研究充分证明了基于范例学习的教学效果(van Gog & Paas, 2008;另见 Crissman, 2006; Hoogerheide &;Roelle, 2020; Kirschner et al.,2006; Magliaro et al、2005;Renkl, 1997, 2014, 为经验证据)。最近在数学领域进行的一项元分析(Barbieri et al.
1.2. Feedback 1.2.反馈意见
In subsequent phases of skill acquisition, such as during practice phases, formative feedback is regarded as further fundamental instructional supplement to deepen students' learning (Graham et al., 2011, 2015; Hattie & Timperley, 2007; Narciss, 2012). Providing learners with feedback has shown to be effective for enhancing their performance (Hattie & Timperley, 2007) in both classrooms (e.g., Bangert-Drowns et al., 1991) and computer-based learning environments (e.g., Mertens et al., 2022; van der Kleij et al., 2015). Feedback can be considered as post-response instructional information about distinct aspects of a student's current performance or learning process to help the student to reduce discrepancies between the actual performance and the targeted performance (Hattie & Timperley, 2007; Narciss, 2008). It can be used to assess (e.g., in form of grades or scales), as well as to diagnose (e.g., whether a learning goal has been achieved), or to evaluate (e.g., the level of students' knowledge) performance regarding various success criteria (Hattie & Timperley, 2007; Kluger & DeNisi, 1996). The most important purpose of formative feedback is to guide the student to regulate and improve the further learning process (Hattie & Timperley, 2007; Narciss, 2008, 2012; Shute, 2008).
在技能学习的后续阶段,如练习阶段,形成性反馈被视为进一步深化学生学习的基本教学补充(Graham et al、2011,2015; Hattie &;Timperley, 2007; Narciss, 2012)。为学习者提供反馈已被证明能有效提高他们的成绩(Hattie & Timperley, 2007)、Bangert-Drowns et al、1991)和基于计算机的学习环境(例如,Mertens et al、2022; van der Kleij 等人,2015)。反馈可被视为学生当前表现或学习过程中不同方面的反应后指导信息,以帮助学生减少实际表现与目标表现之间的差异(Hattie &;Timperley, 2007; Narciss, 2008)。它既可用于评估(例如,以等级或量表的形式),也可用于诊断(例如,在诊断过程中,可以使用 "阈值 "或 "阈限 "等术语)。例如学生的知识水平)的表现(Hattie &;Timperley, 2007; Kluger & DeNisi, 1996)。形成性反馈最重要的目的是引导学生调节和改进进一步的学习过程(Hattie & Timperley, 2007);Narciss, 2008, 2012;Shute, 2008)。
Regarding the feedback information provided by an external source (e.g., computer-based system), previous research investigated different types of feedback attributing different functions of feedback. Narciss (2012) differentiated between outcome-related feedback types, such as knowledge of result (KR), knowledge of correct response (KCR), or answer-until-correct (AUC) feedback (see also corrective feedback, Fyfe & Rittle-Johnson, 2016; Shute, 2008; lower-order feedback, Kuklick & Lindner, 2023; Mertens et al., 2022). These feedback types are related to technical characteristics of feedback to generate more correct responses. In order to improve incorrect responses, further information is often needed. Therefore, more elaborated feedback (EF) types are used that provide more complex feedback contents as compared to corrective or knowledge of result feedback (Kuklick & Lindner, 2023; Mertens et al., 2022; Narciss, 2012; Shute, 2008). The use of elaborated feedback, which in addition to information on the correctness of the answer provides specific guidance on how to close existing knowledge gaps, is regarded to further enhance the effectiveness of feedback (Brooks et al., 2019; Hattie & Timperley, 2007; Mertens et al., 2022; Shute, 2008).
关于外部来源(如基于计算机的系统)提供的反馈信息,以往的研究调查了不同类型的反馈,这些反馈具有不同的功能。Narciss (2012)区分了与结果相关的反馈类型,如结果知识(KR)、或 "答非所问 "反馈(另见矫正反馈,Fyfe &;Rittle-Johnson, 2016; Shute, 2008;低阶反馈,Kuklick & Lindner,2023;Mertens et al.,2022)。这些反馈类型与反馈的技术特点有关,以产生更多正确的回答。为了改进不正确的回答,通常需要更多的信息。因此,与纠正性反馈或结果知识反馈相比,更详细的反馈(EF)类型可提供更复杂的反馈内容(Kuklick &;Lindner, 2023; Mertens et al.,2022;Narciss, 2012;Shute, 2008)。 精心设计的反馈除了提供关于答案正确性的信息外,还提供关于如何弥补现有知识差距的具体指导,这种反馈的使用被认为可以进一步提高反馈的有效性(Brooks et al、2019; Hattie & Timperley, 2007; Mertens et al、2022;Shute, 2008)。
For instance, in their meta-analysis (435 studies, k = 994, N > 61,000 students), Wisniewski et al. (2020) investigated whether the type of feedback (reinforcement/punishment, corrective feedback, or elaborated feedback) moderated the effect of feedback on students’ learning. Overall, the authors found a medium effect (d = 0.48) of providing feedback on learning. This effect was more pronounced for elaborated feedback (d = 0.99, k = 42 studies) than for corrective feedback (d = 0.46, k = 238 studies), or reinforcement/punishment (d = 0.24, k = 39 studies). Relatedly, Mertens et al. (2022) documented a comparable pattern for computer-based feedback. Based on 163 effects from 77 experimental studies they investigated by conducting a network meta-analysis whether different feedback types (KR, KCR, AUC, and EF compared with a no-feedback control condition) had different effects on lower- (i.e., recall/recognition) and higher-order (i.e., transfer) learning performance. Their results show that elaborated feedback is most effective to support learning (on lower-order: g = 0.71, p < .01, as well as on higher-order: g = 0.46, p < .01). Correct response feedback turned out to be least effective (only on lower-order: g = 0.24, p = .026; higher order: g = 0.14, p = .345).
例如,在他们的荟萃分析(435 项研究,k = 994,N > 61,000 名学生)中,Wisniewski et al.(2020) 调查了反馈类型(强化/惩罚、纠正反馈或详细反馈)是否会调节反馈对学生学习的影响。总体而言,作者发现提供反馈对学习的影响为中等(d = 0.48)。与纠正性反馈(d = 0.46, k = 238 项研究)或强化/惩罚(d = 0.24, k = 39 项研究)。与此相关,Mertens 等人(2022) 记录了基于计算机的反馈的类似模式。基于 77 项实验研究中的 163 项效果,他们通过进行网络荟萃分析,研究了不同的反馈类型(KR、KCR、AUC 和 EF,与无反馈对照条件相比)是否对低阶(即回忆/认知)和高阶(即迁移)学习成绩产生不同的影响。他们的研究结果表明,精心设计的反馈能最有效地支持学习(在低阶学习方面):g = 0.71,p < .01,高阶反馈也是如此:g = 0.46,p < .01)。 正确回答反馈的效果最差(仅对低阶反馈有效:g = 0.24,p = .026;高阶:g = 0.14,p = .345)。
1.3. Combining strategy instruction and formative feedback
1.3.策略指导与形成性反馈相结合
Given that strategy instruction and formative feedback separately are both effective to support students’ learning, they are often combined in educational practice (Magliaro et al., 2005; Rosenshine, 2008): Provided with strategy instruction, students should build a consistent mental representation of the problem situation and the corresponding solution strategies. Receiving feedback during practice phases should further support the application and transfer of the previously learned solution strategies. Alternatively, the interactive, two-feedback-loop (ITFL) model by Narciss (2008) describes the implementation of feedback as a reciprocal interplay between internal (learner) and external (feedback) factors (cf. instructional medium) that may affect the effectiveness of feedback (see also Butler & Winne, 1995; Hattie & Timperley, 2007; Hayes, 2012; Panadero & Lipnevich, 2022; Winne & Butler, 1994, for related assumptions). Thus, alternatively, if prior strategy instruction (external factor) is given in advance of a practice phase, students could be inclined to undervalue the particular task requirements in the practice phase (see also illusion of understanding, Renkl, 2014). These subjective representations of task requirements could result in lower levels of processing of the feedback, both on the cognitive, meta-cognitive, or motivational level (Narciss, 2008). Thus, the combination could result in worse performance compared to the single interventions. Whether the combination of strategy instruction and feedback contributes more or even less to learning than a single intervention is still an open question.
鉴于策略指导和形成性反馈都能有效地支持学生的学习,因此在教育实践中经常将两者结合起来(Magliaro et al、2005; Rosenshine, 2008):通过策略指导,学生应建立起对问题情境和相应解决策略的一致心理表征。在练习阶段获得的反馈应能进一步支持先前所学解决策略的应用和迁移。另外,Narciss (2008) 的交互式双反馈回路(ITFL)模型将反馈的实施描述为内部因素(学习者)和外部因素(反馈)之间的相互影响。参见 Butler &;Winne, 1995; Hattie & Timperley, 2007;Hayes, 2012; Panadero &;Lipnevich, 2022;Winne & Butler, 1994,相关假设)。 因此,另一种情况是,如果在练习阶段之前进行事先策略指导(外部因素),学生可能会倾向于低估练习阶段的特定任务要求(另见理解错觉,Renkl, 2014)。这些任务要求的主观表征可能会导致认知、元认知或动机层面的反馈处理水平降低(Narciss, 2008)。因此,与单一干预措施相比,两者结合可能会导致成绩下降。与单一干预相比,策略指导和反馈的组合对学习的贡献是更大还是更小,这仍然是一个未决问题。
Related evidence can be found in the synthesis by Salden et al. (2010). The authors synthesized the effects of eight studies in which strategy instruction (i.e., worked examples) and tutored problem-solving comprising different feedback elements (e.g., prompts, immediate feedback, and context-sensitive hints) were combined. Across those studies, the findings indicated that adding strategy instruction to tutored problem-solving contributed to learning.
相关证据可参见Salden 等人(2010)的综述。作者综合了八项研究的效果,在这些研究中,策略指导(即工作示例)和由不同反馈元素(如提示、即时反馈和上下文相关提示)组成的问题解决辅导相结合。这些研究的结果表明,在辅导解题过程中加入策略指导有助于学习。
However, there are also contradictory findings. To explore the effects of differences of prior strategy instruction and feedback, Fyfe and Rittle-Johnson (2016) provided elementary students with strategy instruction about solving mathematical equivalence problems or with control instruction. Afterwards, in a practice phase, students were asked to solve 12 structurally similar mathematical problems. One half of the elementary students received corrective feedback after each mathematical problem whereas the other half did not receive feedback. The authors found a strategy-by-feedback interaction, as students only profited from strategy instruction when no feedback was provided (see also Fyfe & Rittle-Johnson, 2017, for similar findings). In Experiment 2, Fyfe and Rittle-Johnson (2016) provided all students with strategy instruction. Additionally, they varied whether students received no feedback, immediate feedback, or summative feedback on the practice tasks. Again, no feedback was more beneficial than providing feedback (see Shirah & Sidney, 2023, for related evidence; see also Wischgoll, 2017, for related findings in academic writing).
然而,也有相互矛盾的研究结果。为了探究事先策略指导和反馈的不同所产生的影响,Fyfe 和 Rittle-Johnson(2016)为小学生提供了解决数学等价问题的策略指导或对照指导。之后,在练习阶段,要求学生解决 12 个结构相似的数学问题。一半的小学生在每个数学问题后都收到了纠正反馈,而另一半学生则没有收到反馈。作者发现了策略与反馈之间的交互作用,因为只有在不提供反馈的情况下,学生才能从策略指导中获益(类似研究结果另见Fyfe & Rittle-Johnson, 2017)。在实验 2 中,Fyfe 和 Rittle-Johnson(2016)为所有学生提供了策略指导。此外,他们还改变了学生在练习任务中获得无反馈、即时反馈或总结性反馈的情况。同样,没有反馈比提供反馈更有益(相关证据见 Shirah & Sidney, 2023;另见Wischgoll, 2017,学术写作中的相关发现)。
Together, these findings provided initial evidence that combining strategy instruction and formative feedback may not necessarily contribute to learning. Instead, they suggest that combining strategy instruction and feedback can also reduce learning performance or at least it has no additive effects (see also Kluger & DeNisi, 1996; Narciss, 2008; Shirah & Sidney, 2023).
总之,这些研究结果提供了初步证据,表明将策略指导与形成性反馈相结合并不一定有助于学习。相反,这些研究结果表明,将策略指导和反馈结合起来也会降低学习成绩,或者至少不会产生叠加效应(参见Kluger &;DeNisi, 1996; Narciss, 2008;Shirah & Sidney, 2023)。
1.3.1. Underlying processes when combining strategy instruction and feedback
1.3.1.策略指导与反馈相结合的基本过程
In line with the ITFL-model (Narciss, 2008), Fyfe and Rittle-Johnson (2016) proposed three non-mutually exclusive explanations for potential strategy instruction-by-feedback interactions.
符合 ITFL 模型(Narciss, 2008)、Fyfe 和 Rittle-Johnson (2016)对潜在的策略指导与反馈的相互作用提出了三种互不排斥的解释。
Regarding the cognitive factors, after receiving strategy instruction, learners' subjective representation of the task requirements and in consequence, learners’ processing of feedback might change due to increases of cognitive load, as students have to process redundant information of the strategy instruction and the feedback (Sweller et al., 2011).
关于认知因素,在接受策略指导后,由于学生必须处理策略指导和反馈的冗余信息,学习者对任务要求的主观表征可能会发生变化,因此学习者对反馈的处理也可能会因认知负荷的增加而发生变化(Sweller et al.)
Alternatively, regarding meta-cognitive factors, the provision of well-designed strategy instruction could result in an overestimation of students‘ comprehension having potentially down-side effects on the regulation of learning processes during subsequent practice phases as well as on their performance (see Baars et al., 2013; Finn & Metcalfe, 2014).
另外,关于元认知因素,提供精心设计的策略指导可能会导致高估学生的理解能力,从而对后续练习阶段的学习过程调节以及学生的成绩产生潜在的负面影响(见Baars et al、2013;Finn & Metcalfe, 2014)。
Last, regarding affective-motivational factors, the illusion of “feeling competent” may contribute to a lower level of arousal, and as such may be not beneficial to learning (Eccles & Wigfield, 2002; Grundmann et al., 2021; Kuklick & Lindner, 2023; Raaijmakers et al., 2017). Arousal refers to a state of excitement varying from low (unexcited, relaxed, calm) to high arousal (increased activation of the sympathetic nervous system; Arnsten, 2009; Hoogerheide et al., 2019). Previous research showed that the arousal level depends on the difficulty of the task or the valence of received feedback, that is, positive (i.e., the task was solved correctly) or negative (i.e., students’ answer was incorrect) feedback (Efklides & Dina, 2004; Kuklick & Lindner, 2023). Thus, the affective-motivational perspective may particularly account in cases of negative feedback, as negative feedback may disengage students (through failed improvements) and reduce learning performance (Grundmann et al., 2021).
最后,关于情感-动机因素,"感觉自己有能力 "的错觉可能会导致较低的唤醒水平,因此可能不利于学习(Eccles &;Wigfield, 2002; Grundmann et al.,2021; Kuklick & Lindner, 2023; Raaijmakers et al、2017)。唤醒是指从低唤醒(不激动、放松、平静)到高唤醒(交感神经系统激活增强)的兴奋状态;Arnsten, 2009;Hoogerheide et al.,2019)。以往的研究表明,唤醒水平取决于任务的难度或收到的反馈的价值,即积极的(即任务正确完成)或消极的(即学生的答案不正确)反馈、反馈(Efklides &;Dina, 2004; Kuklick & Lindner, 2023)。因此,情感-动机观点尤其适用于负面反馈的情况,因为负面反馈可能会(通过失败的改进)使学生失去兴趣,并降低学习成绩(Grundmann et al.2021年)。
However, it has to be noted that the studies that found a reducing effect of combining strategy instruction and feedback, predominantly applied lower-order feedback (Fyfe & Rittle-Johnson, 2016; Wischgoll, 2017). Thus, it is an open question whether these findings may generalize to more elaborated types of feedback. Therefore, the instruction-by-feedback interaction could be simply a result of the combination of strategy instruction and lower-order feedback, as lower-order feedback may not pose additional benefits regarding learning gains, and thus result in extraneous processing (see Mitchell & Jolley, 2010, for a critical discussion).
然而,必须指出的是,那些发现策略指导与反馈相结合具有减小效果的研究主要采用了低阶反馈(Fyfe &;Rittle-Johnson, 2016; Wischgoll, 2017)。因此,这些研究结果是否能推广到更详细的反馈类型还是一个未决问题。因此,教学与反馈之间的相互作用可能仅仅是策略教学与低阶反馈相结合的结果,因为低阶反馈可能不会给学习收益带来额外的益处,从而导致不相干的处理(参见Mitchell & Jolley, 2010,以进行批判性讨论)。
1.3.2. Sequence of learning phase and practice phase
1.3.2.学习阶段和实践阶段的顺序
Another aspect of the orchestration which might influence the combination of instruction and feedback is the sequence in which these activities are arranged. Recent research on PS-I (i.e., problem-solving prior to instruction, see Kapur, 2012; Loibl et al., 2017; Sinha & Kapur, 2021, for an overview) argued that preponing practice phases, such as problem-solving activities could enhance learning (see Sinha & Kapur, 2021, for meta-analytical evidence). Besides other mediating factors, the ability to notice potential flaws in students' understanding in an early stage of knowledge acquisition (Loibl & Rummel, 2014), is discussed as one underlying process that could explain the effects of flipping practice phases and instruction. Following Narciss (2008), flipping instruction and practice phases could furthermore enhance students’ subjective representation of the task requirements, and thus result in more elaborated processing of the feedback.
可能影响教学与反馈结合的另一个方面是这些活动的安排顺序。最近关于 PS-I 的研究(即在教学之前解决问题,见Kapur, 2012;Loibl et al、2017; Sinha &;Kapur,2021,综述)认为,预习练习阶段,如问题解决活动,可以提高学习效果(见Sinha & Kapur,2021,元分析证据)。除其他中介因素外,在知识获取的早期阶段注意到学生理解中潜在缺陷的能力(Loibl & Rummel, 2014),也被视为可以解释翻转练习阶段和教学效果的一个潜在过程。Narciss (2008)认为,翻转教学和练习阶段可进一步增强学生对任务要求的主观表征,从而对反馈进行更精细的处理。
Previous research regarding the effectiveness of PS-I, however, produced mixed findings (Kant et al., 2017; Loibl & Rummel, 2014; Sinha & Kapur, 2021; van Gog et al., 2011; van Harsel et al., 2020). On the one hand, there is supporting evidence for flipping instruction and practice phases: For instance, Sinha and Kapur (2021) conducted a meta-analysis of 53 studies including 166 effect sizes investigating the differential impact of the sequences problem-solving followed by instruction (PS-I) versus instruction followed by problem-solving (I-PS) on students’ procedural, conceptual knowledge, and transfer performance in different domains. Their results showed a significant small to medium effect of PS-I (g = 0.36, 95% CI [0.20, 0.51]) regarding conceptual knowledge and transfer. Regarding procedural knowledge, in line with previous research, the authors found no significant effect (g = −0.03, 95% CI [−0.20, 0.15]).
然而,以往关于 PS-I 效果的研究结果不一(Kant et al、2017; Loibl & Rummel,2014;Sinha & Kapur, 2021; van Gog et al.,2011;van Harsel 等人,2020)。一方面,有支持翻转教学和练习阶段的证据:例如,Sinha 和 Kapur (2021)对 53 项研究(包括 166 个效应大小)进行了荟萃分析,调查了先解决问题后教学(PS-I)与先教学后解决问题(I-PS)的顺序对学生在不同领域的程序性知识、概念性知识和迁移表现的不同影响。他们的研究结果表明,在概念知识和迁移方面,PS-I(g = 0.36,95% CI [0.20,0.51])具有明显的中小型影响。在程序性知识方面,与之前的研究一致,作者发现没有显著影响(g = -0.03,95% CI [-0.20,0.15])。
On the other hand, there is contrasting evidence in the domain of worked examples. For instance, van Gog et al. (2011) compared different sequences of instruction and practice phases: The authors presented secondary students (N = 103) either instruction-only (i.e., by providing worked examples), practice phase-only, instruction followed by a practice phase, or a practice phase followed by instruction. The results showed that the instruction-only condition and the instruction followed by a practice phase condition outperformed the practice phase-only condition and the practice phase followed by instruction condition (see also Kant et al., 2017, for related findings). However, whether these findings could be transferred to practice phases incorporating feedback is an open question. It can be assumed that changing the sequence of instruction and feedback may contribute to learning. If students are engaged in a practice phase at the beginning of their studying, they could be more reliant on the additional feedback information to correctly solve the problems in the practice phase. The subsequent instruction phase would then serve as additional scaffold to consolidate the previously acquired strategies during the practice phase (see productive failure approach, Kapur, 2008).
另一方面,在工作实例领域却有相反的证据。例如,van Gog 等人(2011)比较了教学和练习阶段的不同顺序:作者为中学生(N = 103)提供了纯指导(即通过提供工作示例)、纯练习阶段、指导后的练习阶段或练习阶段后的指导。结果显示,纯指导条件和先指导后练习阶段条件的效果优于纯练习阶段条件和先练习阶段后指导条件(相关研究结果另见Kant 等人,2017)。然而,这些研究结果能否应用于包含反馈的练习阶段还是一个未决问题。可以认为,改变教学和反馈的顺序可能有助于学习。如果学生在学习之初就进入练习阶段,他们可能会更加依赖额外的反馈信息来正确解决练习阶段的问题。随后的指导阶段将作为额外的支架,巩固之前在练习阶段获得的策略(见生产性失败方法,Kapur, 2008)。
1.4. Overview of the present study
1.4.本研究概述
The main aim of the present study, that included three online experiments, was to investigate potential (additive or reduced) effects of combining strategy instruction and feedback on students' physics learning in online environments. Throughout the experiments, we used identical learning materials (i.e., strategy instruction, practice tasks with feedback) to be able to compare findings across our experiments. In a multimedia learning environment, the students were provided with strategy instruction based on worked examples for solving electric circuits problems (Ohm's law, see Hoogerheide et al., 2019). In the practice phase, the students answered several practice tasks, in which they applied Ohm's law (by calculating the electric current) to different electric circuits.
本研究包括三个在线实验,其主要目的是研究策略指导与反馈相结合对学生在在线环境中学习物理的潜在(增加或减少)影响。在整个实验过程中,我们使用了相同的学习材料(即策略指导、带反馈的练习任务),以便比较不同实验的结果。在多媒体学习环境中,我们根据解决电路问题(欧姆定律,见 Hoogerheide et al., 2019)的工作示例为学生提供策略指导。在练习阶段,学生们回答了几个练习任务,在这些任务中,他们将欧姆定律(通过计算电流)应用于不同的电路。
The goal of Experiment 1 was to replicate a potential instruction-by-feedback interaction effect (as Fyfe & Rittle-Johnson, 2016, found in the domain of mathematics) in the domain of physics with corrective feedback. We used a 2 × 2-factorial design, crossing the two factors strategy instruction (yes vs. no) and feedback (yes vs. no). Furthermore, based on the ITFL-model by Narciss (2008), we tested for potential explanations regarding the underlying cognitive (i.e., mental effort, subjective difficulty), meta-cognitive (i.e., monitoring accuracy) and affective-motivational processes (i.e., arousal) accounting for additive versus reducing effects of combining strategy instruction and feedback.
实验 1 的目的是在物理领域中复制潜在的教学与反馈交互效应(如Fyfe & Rittle-Johnson, 2016在数学领域中发现的纠正反馈)。我们采用了 2 × 2 因子设计,将策略指导(是与否)和反馈(是与否)这两个因素交叉在一起。此外,基于Narciss (2008)的ITFL模型,我们测试了潜在的认知(即:脑力劳动、主观难度)、认知(即:认知)、主观难度(即:主观难度)、主观难度(即:主观难度)、主观难度(即:主观难度)、主观难度(即:主观难度)等方面的潜在解释、我们对潜在的认知过程(即脑力劳动、主观难度)、元认知过程(即监控准确性)和情感动机过程(即唤醒)进行了测试,以了解策略指导与反馈相结合所产生的相加效应与相减效应。
Given that instruction-by-feedback interaction could depend on the elaboration of feedback (external factor; Narciss, 2008), Experiment 2 aimed to generalize the findings and examine the robustness of the obtained effects in Experiment 1 by using elaborated feedback. In Experiment 3, we varied the sequence in which strategy instruction and feedback were provided, that is a learning phase with strategy instruction followed by a practice phase including feedback versus a preponed practice phase including feedback followed by a learning phase with strategy instruction. This variation allowed us to investigate whether instruction-by-feedback interaction may be a result of different sequences of strategy instruction and feedback. Together, these experiments allow to rigorously examine the underlying mechanisms and generalizability of combination effects of strategy instruction and feedback.
鉴于指导与反馈之间的交互作用可能取决于反馈的详细程度(外部因素;Narciss, 2008),实验 2 的目的是通过使用详细的反馈来推广实验 1 中的发现并检验实验 1 中获得的效果的稳健性。在实验 3 中,我们改变了提供策略指导和反馈的顺序,即先进行策略指导的学习阶段,然后是包含反馈的练习阶段,与先进行包含反馈的预演练习阶段,然后是包含策略指导的学习阶段。这种变化使我们能够研究策略指导和反馈的不同顺序是否会导致指导与反馈之间的相互作用。通过这些实验,我们可以严格研究策略指导和反馈组合效应的内在机制和可推广性。
2. Experiment 1 2.实验 1
2.1. Research questions and hypotheses
2.1.研究问题和假设
We pre-registered our hypotheses on as predicted (see https://aspredicted.org/tn6nx.pdf). In line with previous evidence (Fyfe & Rittle-Johnson, 2016; Wischgoll, 2017), we hypothesized a main effect of strategy instruction on students’ learning (Hypothesis 1a), as strategy instruction should help students construct a solid mental representation of the subject matter which should be conducive to learning. Regarding the effectiveness of corrective feedback (Experiment 1), it was less clear what to expect. Based on previous research, we can assume that different types of feedback should be differently effective (see Experiment 2 for details). However, in previous studies (e.g., Fyfe & Rittle-Johnson, 2016) on combination effects no significant main effect of feedback was obtained (which is in contrast to the common literature on the effectiveness of feedback). Thus, we refrained from making a clear prediction regarding main effects of feedback on learning.
我们按照预测对假设进行了预先登记(见https://aspredicted.org/tn6nx.pdf )。与之前的证据一致(Fyfe & Rittle-Johnson, 2016;Wischgoll, 2017),我们假设策略教学对学生的学习有主效应(假设1a),因为策略教学应有助于学生构建对主题内容的稳固心智表征,从而有利于学习。至于矫正反馈的效果(实验 1),我们的预期就不那么明确了。根据以往的研究,我们可以认为不同类型的反馈应该有不同的效果(详见实验 2)。然而,在以往关于组合效应的研究中(例如,Fyfe & Rittle-Johnson, 2016),没有发现反馈有显著的主效应(这与有关反馈有效性的常见文献相反)。因此,我们没有就反馈对学习的主要影响做出明确的预测。
Additionally, based on Fyfe and Rittle-Johnson (2016), we hypothesized an instruction-by-feedback interaction (Hypothesis 1b). Accordingly, the addition of feedback should impair the effectiveness of strategy instruction. We also explored, whether the findings on transfer performance were robust for the underlying learning outcomes (i.e., near and far transfer).
此外,根据Fyfe 和 Rittle-Johnson(2016),我们假设教学与反馈之间存在相互作用(假设 1b)。因此,增加反馈应该会损害策略教学的有效性。我们还探讨了有关迁移绩效的研究结果是否对基本学习成果(即近迁移和远迁移)具有稳健性。
In case we find an instruction-by-feedback interaction effect, we aimed at exploring the underlying processes of the potential instruction-by-feedback interaction from a cognitive, a meta-cognitive, and an affective-motivational perspective (Narciss, 2008). Following a cognitive perspective, we explored whether increases in cognitive load during the practice phase may mediate the effect of instruction-by-feedback interaction on learning. We therefore used subjective cognitive load ratings (i.e., mental effort, subjective difficulty) as potential mediators. Concretely, we assumed that students who received both instruction and feedback will experience an increase in cognitive load during the practice phase due to the dual support compared to students who received only one type of support (strategy instruction-only or feedback-only; Hypothesis 2a).
如果我们发现了教学与反馈的交互效应,我们将从认知、元认知和情感动机的角度(Narciss, 2008)来探索潜在的教学与反馈交互作用的潜在过程。从认知角度出发,我们探讨了在练习阶段认知负荷的增加是否会介导教学-反馈互动对学习的影响。因此,我们将主观认知负荷评级(即脑力劳动、主观难度)作为潜在的中介因素。具体来说,我们假设,与只接受一种支持(仅策略指导或仅反馈;假设 2a)的学生相比,同时接受指导和反馈的学生在练习阶段会因为双重支持而增加认知负荷。
Following a meta-cognitive perspective, we also investigated whether students' overestimations mediate this effect. We used monitoring accuracy ratings to calculate students’ overestimations: Students who rated their prospective performance as being better than their actual performance (self-estimated test performance > actual test performance) were defined as overestimated judgements. Specifically, we hypothesized that students who first received strategy instruction followed by feedback will perform worse than students who only received feedback due to their overestimation (represented by positive values of monitoring accuracy; Hypothesis 2b).
从元认知的角度出发,我们还研究了学生的高估是否对这一效果起到了中介作用。我们使用监测准确性评级来计算学生的高估:将自己的预期成绩评定为优于实际成绩(自我估计考试成绩 > 实际考试成绩)的学生被定义为高估判断。具体来说,我们假设先接受策略指导再接受反馈的学生会比只接受反馈的学生成绩差,因为他们高估了自己的成绩(以监测准确度的正值表示;假设 2b)。
Last, following an affective-motivational perspective, we assumed that arousal mediates the effect. We used subjective ratings of perceived arousal to test this assumption. Concretely, we hypothesized that students who first received strategy instruction followed by feedback in the practice phase will show lower arousal levels, resulting in poorer performance compared to students who received only feedback without any prior strategy instruction (Hypothesis 2c).
最后,从情感-动机的角度出发,我们假定唤醒对这一效应具有中介作用。我们采用了对觉察到的唤醒程度进行主观评分的方法来验证这一假设。具体来说,我们假设在练习阶段先接受策略指导再接受反馈的学生会表现出较低的唤醒水平,从而导致他们的学习成绩比那些只接受反馈而事先未接受任何策略指导的学生差(假设 2c)。
2.2. Method 2.2.方法
2.2.1. Participants 2.2.1.参与者
In total, 499 students from non-physics study programs of a German university participated in the online experiment (which was approved by the local research ethics committee). The students were recruited via a university mailing list. We excluded students who were enrolled in physics, science and technology, or engineering (n = 29), whose native language was not German (or whose language level in German was below C1; n = 0), or who were not enrolled in any study program (n = 3). In addition, we only included those participants who solved both attention checks (see section 2.2.3. Materials, for details) correctly to be more restrictive (n = 21 failed at least one of the attention checks). Thus, the final sample consisted of N = 437 students.
共有499名来自德国一所大学非物理学专业的学生参加了在线实验(该实验已获得当地研究伦理委员会的批准)。这些学生是通过大学邮件列表招募的。我们排除了物理、科技或工程专业的学生(n = 29)、母语非德语的学生(或德语水平低于 C1 的学生;n = 0)或未参加任何学习项目的学生(n = 3)。此外,我们只将那些同时解决了注意力检查(详见2.2.3 部分。详细内容请参见 "材料 "一栏)的限制性更强(n = 21 人至少有一项注意力检查未通过)。因此,最终样本包括 N = 437 名学生。
The mean age of the students was 22.77 years (SD = 3.80) and 65% were female. The students were on average in their 6.68 semester (SD = 4.74) and most of them were enrolled in humanities (n = 168). The final sample size of 437 students exceeded the required sample size of N = 157, which was based on an a priori power analysis for an ANCOVA with interactions, expecting a medium to large effect size Cohen's η2 = .08 (based on Fyfe & Rittle-Johnson, 2016) and power (1 − β) = 0.95. The higher number of students in our obtained sample resulted from an unexpected interest in our study (499 university students within 5 h).
学生的平均年龄为 22.77 岁(SD = 3.80),65% 为女性。学生的平均学期为 6.68 个学期(SD = 4.74),大部分学生就读于人文学科(n = 168)。437 名学生的最终样本量超过了N = ;157 人,这是根据对具有交互作用的 ANCOVA 的先验功率分析得出的,预期效果大小为中到大 Cohen's η2 = .08(基于Fyfe & Rittle-Johnson, 2016)和功率(1 - β)= 0.95。我们获得的样本中学生人数较多,这是因为我们的研究受到了意想不到的关注(5 小时内有 499 名大学生)。
2.2.2. Design 2.2.2.设计
We conducted a 2 × 2-factorial design with strategy instruction (yes vs. no) and feedback (yes vs. no) as between-participants factors. Participants were randomly assigned to one of four conditions (without strategy instruction and without feedback, n = 107; without strategy instruction but with feedback, n = 111; with strategy instruction but without feedback, n = 112; with instruction and with feedback, n = 107).
我们采用了 2 × 2 因子设计,将策略指导(有与无)和反馈(有与无)作为参与者之间的因素。参与者被随机分配到四个条件之一(无策略指导和无反馈,n = 107;无策略指导但有反馈,n = 111;有策略指导但无反馈,n = 112;有指导和反馈,n = 107)。
We measured students’ performance in the practice tasks, as well as their performance in the transfer tasks as dependent variables. Following the previously mentioned three different (but not mutually exclusive) perspectives on the underlying processes of an instruction-by-feedback interaction, we deliberately assessed subjective measures of cognitive load (mental effort, subjective difficulty), monitoring accuracy, and affect (arousal, pleasure) as potential cognitive, meta-cognitive and affective-motivational mediators. Furthermore, we collected data on prior knowledge, academic self-concept, and intrinsic motivation to better describe our online sample.
我们将学生在练习任务中的表现以及在迁移任务中的表现作为自变量进行测量。根据前面提到的关于教学与反馈互动的基本过程的三种不同(但并不相互排斥)观点,我们特意评估了认知负荷(脑力劳动、主观难度)、监控准确性和情感(唤醒、愉悦)的主观测量,将其作为潜在的认知、元认知和情感动机中介。此外,我们还收集了有关先前知识、学术自我概念和内在动机的数据,以更好地描述我们的在线样本。
2.2.3. Materials 2.2.3.材料
The entire experiment was presented with the online survey tool SoSci Survey (https://www.soscisurvey.de/). Based on previous learning and test materials (Rhöneck, 1986; Urban-Woldron & Hopf, 2012; see also Hoogerheide et al., 2019; Ivanjek et al., 2021), all the materials were carefully adapted by two experienced subject matter experts in physics education.
整个实验使用在线调查工具 SoSci Survey(https://www.soscisurvey.de/ )进行。基于以前的学习和测试材料(Rhöneck, 1986;Urban-Woldron & Hopf, 2012;另见 Hoogerheide et al.,2019;Ivanjek 等人,2021),所有材料均由两位经验丰富的物理教育专家精心改编。
2.2.3.1. Prior knowledge 2.2.3.1.先验知识
We used a pretest comprising eight multiple-choice items (e.g., “What do you know about the total current in a parallel circuit?”; Cronbach's α for Experiment 1–3: α1 = 0.56, α2 = 0.59, α3 = 0.53) where students were asked to select one correct solution per item out of 4 possible options, adapted from commonly used test instruments in physics education to assess students' conceptual understanding of electrical circuits (Rhöneck, 1986; Urban-Woldron & Hopf, 2012; see also Hoogerheide et al., 2019; Ivanjek et al., 2021, for related approaches). For each correct answer, students received one point, yielding a maximum score of eight points. We transformed the sum score to a proportion score (range: 0–1) by dividing the achieved points by the maximum points.
我们使用了由八个选择题组成的前测(例如:"你对并联电路中的总电流了解多少?"实验 1-3 的 Cronbach's α:α1 = 0.56,α2 = 0.59,α3 = 0.53。53),要求学生从 4 个可能的选项中选择一个正确的解决方案,该测试改编自物理教育中常用的测试工具,用于评估学生对电路概念的理解(Rhöneck, 1986;Urban-Woldron & Hopf, 2012;另见 Hoogerheide et al.,2019; Ivanjek et al.)每答对一个问题,学生得一分,最高得分为八分。我们将总分除以最高分,将总分转换为比例分(范围:0-1)。
2.2.3.2. Strategy instruction
2.2.3.2.战略指导
In the learning phase, students were provided with either strategy instruction or no instruction. The main aim of the strategy instruction was to enable students to calculate currents in a parallel circuit on the basis of Ohm's law in the form of with I referring to current, V to voltage, and R to resistance. Students needed to develop a conceptual understanding of this subject matter to solve the subsequent practice tasks.
在学习阶段,学生要么接受策略指导,要么不接受指导。策略指导的主要目的是让学生根据欧姆定律,以 I 表示电流,V 表示电压,R 表示电阻的形式计算并联电路中的电流。学生需要从概念上理解这一主题,才能解决后续的练习任务。
The strategy instruction consisted of a multimedia presentation comprising conceptual information about the physical quantities “current”, “voltage”, and “resistance”, their relationship in electric circuits based on Ohm's law as well as how a parallel circuit is constructed. A graphic representation of an electric circuit served to illustrate these abstract physics concepts. To enhance the effectiveness of the multimedia presentation, we considered general (multimedia) principles, such as signaling (Mautone & Mayer, 2001; Richter et al., 2018), and added process-oriented information to enhance conceptual understanding (van Gog et al., 2006, 2008; see also Lachner et al., 2019; see Fig. 1a). In the worked example, the general solution steps were modeled to provide a strategy, how the current, displayed by the four ammeters in a parallel circuit (see Fig. 1b), can be determined by applying Ohm's law (see Hoogerheide et al., 2019).
教学策略由多媒体演示组成,包括有关物理量 "电流"、"电压 "和 "电阻 "的概念性信息,它们在电路中基于欧姆定律的关系,以及如何构建并联电路。电路的图示可用于说明这些抽象的物理概念。为了提高多媒体演示的效果,我们考虑了一般(多媒体)原则,如信号传递(Mautone & Mayer, 2001;Richter et al.,2018),并添加了面向过程的信息以增强概念理解(van Gog et al、2006、2008 ;另见Lachner et al、2019;见 图 1a)。在工作示例中,对一般求解步骤进行了建模,以提供一种策略,说明并联电路中四个电流表显示的电流(见 Fig.1b),可以通过应用欧姆定律来确定(见 Hoogerheide et al、2019)。
In the no-strategy instruction conditions students received a control group instruction (see Fig. 2; see Fyfe & Rittle-Johnson, 2016, for similar approaches). We included a control instruction instead of comparing the strategy instruction to a plain control condition which did not receive anything, to avoid potential differences regarding time-on-task (Möckel et al., 2015). Differences in time-on-task may affect the overall amount of processing during the cognitive tasks, and potential differences in learning (Boksem et al., 2005, 2006; Langner et al., 2010; Lorist et al., 2000). Thus, holding time-on-task constant across condition can be regarded necessary to avoid potential confounds (Kovanović et al., 2015; Möckel et al., 2015).
在无策略教学条件下,学生接受对照组教学(见图 2;见)。2;类似方法见 Fyfe & Rittle-Johnson, 2016)。我们加入了一个对照指导,而不是将策略指导与不接受任何指导的普通对照条件进行比较,以避免任务完成时间方面的潜在差异(Möckel 等人,2015)。任务时间的差异可能会影响认知任务的整体处理量,以及潜在的学习差异(Boksem et al、2005, 2006; Langner et al、2010; Lorist 等人,2000)。因此,为了避免潜在的混杂因素,在不同条件下保持任务时间不变被认为是必要的(Kovanović et al、2015;Möckel 等人,2015)。
Furthermore, we included a control instruction instead of another active treatment to be able to reach conclusion about the effectiveness of the strategy instruction. When different aspects of the instructions vary, we could not say with certainty whether the strategy instruction is responsible for the effect (Renkl, 2014). That is why it is important to keep the control instruction as constant as possible to the strategy instruction. To control for time-on-task for all conditions and nevertheless to avoid producing another treatment effect, we created a control instruction which would not influence the effect on task performance regarding the content (since the required information to solve the tasks was not included) but kept all other factors as constant as possible to the strategy instruction.
此外,我们还加入了对照指导,而不是另一种积极的治疗方法,以便能够就策略指导的有效性得出结论。当指导的不同方面发生变化时,我们无法确定是否是策略指导产生了效果(Renkl, 2014)。这就是控制指令与策略指令尽可能保持一致的重要原因。为了控制所有条件下的任务完成时间,同时避免产生另一种处理效应,我们创建了一个控制指令,它不会影响任务内容对任务表现的影响(因为不包括解决任务所需的信息),但尽可能保持所有其他因素与策略指令一致。
The control group instruction was comparable to the strategy instruction regarding text structure, schematics to illustrate the described contents, worked example, and terminology. The control group instruction also comprised a multimedia presentation (see Fig. 2a) and a worked example (see Fig. 2b). The contents of the strategy instruction and the control group instruction were divergent as students who received the control instruction were not introduced to Ohm's law or to the calculation of currents in a parallel circuit. Instead, the control instruction provided information about the invention of the battery and the effects of electric current. How to calculate currents in a parallel circuit was not explained in the multimedia presentation or in the worked example. Thus, the control instruction was not target-oriented for the following practice tasks but was only provided due to time-on-task constraints (see also Fyfe & Rittle-Johnson, 2016).
对照组的教学在文本结构、说明内容的示意图、工作示例和术语方面与策略教学相似。对照组的教学还包括多媒体演示(见 图 2a)和工作示例。2a)和一个工作示例(见 图 2b)。策略指导和对照组指导的内容有所不同,因为对照组指导没有向学生介绍欧姆定律或并联电路中电流的计算。相反,对照组教学提供了有关电池发明和电流影响的信息。如何计算并联电路中的电流在多媒体演示和作业示例中都没有解释。因此,在接下来的练习任务中,控制指令并非以目标为导向,而是由于任务时间限制才提供的(另见Fyfe & Rittle-Johnson, 2016)。
2.2.3.3. Materials in the practice phase
2.2.3.3.实践阶段的材料
The practice phase comprised two isomorphic practice tasks and additionally corresponding feedback in the feedback conditions. We have chosen this design because first, previous research showed that a restricted number of practice tasks were sufficient to foster students' learning. For example, Hoogerheide et al. (2019), whose material we adapted, only implemented one practice task. This decision is in line with findings by van Harsel et al. (2020), who showed that more (i.e., eight) tasks do not necessarily lead to more learning than fewer (i.e., four) tasks. Additionally, the use of a restricted amount of closed-ended questions is often used for rapid assessments in online learning environments (cf. Kalyuga, 2006; Kalyuga & Sweller, 2004). Second, we aimed at reducing the drop-out risk (cf. attrition bias, Nunan et al., 2018) caused by the online study setting and the length of the questionnaire (Galesic, 2006; Hoerger, 2010). Thus, we decided to realize an effective and parsimonious implementation of guided practice, and used two practice problems. Students answered these two isomorphic (i.e., analogous problem-solving strategy, different surface features) practice tasks (Cronbach's α for Experiment 1–3: α1 = 0.47, α2 = 0.56, α3 = 0.56) in which they had to determine the currents by applying Ohm's law. Therefore, students had to select the correct solution from four alternatives. The distractors were based on common misconceptions students hold regarding currents (see McDermott & Shaffer, 1992; Schecker et al., 2018, for examples). The performance on the practice tasks was again transformed to proportion scores (range: 0–1).
练习阶段包括两个同构练习任务和反馈条件下的相应反馈。我们之所以选择这种设计,首先是因为以往的研究表明,有限数量的练习任务足以促进学生的学习。例如,Hoogerheide 等人(2019)(我们改编了他们的教材)只实施了一个练习任务。这一决定与van Harsel 等人(2020)的研究结果一致,他们的研究表明,较多(即八个)的任务并不一定比较少(即四个)的任务带来更多的学习效果。此外,在线学习环境中的快速评估通常会使用数量有限的封闭式问题(参见:)。Kalyuga, 2006; Kalyuga & Sweller, 2004)。其次,我们的目标是降低辍学风险(参见自然减员偏差,Nunan et al、2018)造成的(Galesic, 2006;Hoerger, 2010)。因此,我们决定有效而简洁地实施指导性练习,并使用了两个练习题。学生回答这两个同构问题(即实验 1-3 的 Cronbach's α:α1 = 0.47,α2 = 0.56,α3 = 0.56)。因此,学生必须从四个备选方案中选出正确的解决方案。干扰项基于学生对电流的常见误解(见 McDermott &;Shaffer, 1992; Schecker et al.,2018 为例)。练习任务的成绩再次转换为比例分数(范围:0-1)。
We have chosen the multiple-choice test format (i.e., for the practice and transfer tasks) for several reasons: First, previous research showed that multiple-choice questions are at least as effective as other formats in consolidation phases (e.g., short-answer test; Adesope et al., 2017; Yang et al., 2021) as well as in combination with prior instruction (Ozuru et al., 2007).
我们选择多项选择测试形式(即用于练习和迁移任务)有几个原因:首先,先前的研究表明,在巩固阶段,多项选择题至少与其他形式一样有效(例如,简答测试;Adesope et al、2017; Yang et al、2021)以及结合事先指导( data-dl-uid="6">Ozuru 等人,2007)。
Second, multiple-choice questions are frequently used in online learning environments combined with feedback because multiple-choice questions provide an easy as well as time- and resources-saving opportunity to analyze students' answers objectively and automatically by a computer-based system (Epstein et al., 2002; Guo et al., 2014; Kuechler & Simkin, 2003; Shin et al., 2019). That allows to provide immediate formative feedback to all students simultaneously which can increase students’ self-regulation, learning gain, and motivation (Allen et al., 2016; Epstein et al., 2002; Guo et al., 2014; Shute, 2008). Thus, we used this advantage and provided students in the feedback conditions with immediate, corrective feedback by giving them information about whether their chosen answer was correct or not (Hattie & Timperley, 2007). Additionally, they received a brief justification why the selected answer was correct (see Fig. 3a for the positive feedback) or why the chosen answer was incorrect (see Fig. 3b for the negative feedback; KCR, Narciss, 2012). Through this short justification, students had the opportunity to expand or to consolidate their knowledge and to benefit from the feedback, even if they had chosen the correct answer.
其次,在线学习环境中经常使用与反馈相结合的多选题,因为多选题为计算机系统客观、自动地分析学生的答案提供了一个简便、节省时间和资源的机会(Epstein et al、2002; Guo et al、2014; Kuechler & Simkin,2003; Shin et al、2019)。这样就可以同时向所有学生提供即时的形成性反馈,从而提高学生的自我调节能力、学习收获和学习动力(Allen et al、2016; Epstein et al、2014; Shute, 2008)。因此,我们利用这一优势,向反馈条件下的学生提供即时的纠正性反馈,告诉他们所选答案是否正确(Hattie & Timperley, 2007)。此外,他们还收到了所选答案正确的简要理由说明(见 Fig. 3a 表示正面反馈)或所选答案不正确的原因(见 Fig.3b 为负面反馈;KCR,Narciss, 2012)。通过这个简短的理由说明,学生有机会扩展或巩固知识,并从反馈中受益,即使他们选择了正确的答案。
2.2.3.4. Learning outcomes (posttest)
2.2.3.4.学习成果(事后测试)
Students' learning outcomes were assessed with four transfer tasks: two near transfer tasks and two far transfer tasks (McDonald's ω for overall transfer in Experiment 1–3: ω1 = 0.57, ω2 = 0.56, ω3 = 0.49). We adapted the transfer instruments by Hoogerheide et al. (2019). To this end, the near transfer tasks were isomorphic to the problems from the practice tasks (i.e., “Determine the amperage by ammeters A1 to A4 in the electric circuit shown above”; Cronbach's α for Experiment 1–3: α1 = 0.60, α2 = 0.57, α3 = 0.67) but based on different variable values. In the second task the circuit diagram was rotated. In the far transfer tasks, the students had to transfer their knowledge to different circuits with a different number of ammeters or different resistors: The first far transfer task was similar to the second near transfer task (rotated diagram) but with an additional ammeter. The second far transfer task seemed like the practice tasks and the first near transfer task with four ammeters but with a variable resistor. Since these two transfer items covered different facets of far transfer (additional ammeter, different type of resistance), it can be expected that the reliability would be insufficient (Cronbach's α for Experiment 1–3: α < 0.20). Therefore, we conducted multilevel analyses as the items were nested within person (see Section 2.3 Results).
学生的学习成果通过四个迁移任务进行评估:两个近迁移任务和两个远迁移任务(实验 1-3 中总体迁移的 McDonald's ω:ω1 = 0.57,ω2 = 0.56,ω3 = 0.49)。我们改编了Hoogerheide 等(2019)的转移工具。为此,近距离转移任务与练习任务中的问题同构(即:"确定安培力")、"实验 1-3 的 Cronbach's α:α1 = 0.60,α2 = 0.57,α3 = 0.67),但基于不同的变量值。在第二个任务中,电路图被旋转。在远距离迁移任务中,学生必须将所学知识迁移到具有不同数量电流表或不同电阻的不同电路中:第一个远距离迁移任务与第二个近距离迁移任务(旋转电路图)类似,但多了一个电流表。第二项远距离转移任务与第一项近距离转移任务的练习任务和带有四个电流表的任务相似,但带有一个可变电阻器。由于这两个转移项目涵盖了远距离转移的不同方面(附加电流表、不同类型的电阻),因此可以预计信度不够(实验 1-3 的 Cronbach's α:α < 0.20)。因此,我们进行了多层次分析,因为项目是嵌套在人中的(见2.3 结果部分)。
Similar to the practice tasks, for each transfer task, students had to select the correct solution from four alternatives. The distractors were based on common misconceptions students hold regarding currents (see McDermott & Shaffer, 1992; Schecker et al., 2018, for examples). Students received one point per correct answer, resulting in a total score of 4 points for the transfer tasks. Again, we transformed the total score to a proportion score (range: 0–1). To explore, whether our findings were robust among knowledge facets, we also provide separate explorative analyses for the near and far transfer test components.
与练习任务类似,在每个转移任务中,学生必须从四个备选方案中选出正确的解决方案。分散注意力的因素基于学生对电流的常见误解(见 McDermott &;Shaffer, 1992; Schecker et al.,2018 为例)。学生每答对一个问题得一分,因此转移任务的总分为 4 分。我们再次将总分转换为比例分(范围:0-1)。为了探究我们的研究结果在不同知识面之间是否具有稳健性,我们还分别对近距离和远距离迁移测试部分进行了探索性分析。
2.2.3.5. Perceived cognitive load
2.2.3.5.认知负荷
Students' perceived cognitive load was measured by subjective ratings of students‘ invested mental effort (“How much mental effort did you invest in studying the instruction/completing the task?”; Jacob et al., 2020; Paas, 1992), and their subjectively perceived task difficulty (“How difficult was it for you to understand the instruction/solve the task?”; DeLeeuw & Mayer, 2008; Jacob et al., 2020) after the learning phase and after the practice phase on a Likert-scale from 1 (not difficult at all) to 9 (very difficult).
学生的认知负荷是通过对学生投入的脑力的主观评价来衡量的("你在学习指令/完成任务时投入了多少脑力?";Jacob et al、2020;Paas, 1992),以及他们主观感受到的任务难度("您理解指令/完成任务的难度如何?";DeLeeuw & Mayer,2008;Jacob et al、2020),在学习阶段后和练习阶段后,采用李克特量表,从 1(一点也不难)到 9(非常难)。
2.2.3.6. Monitoring accuracy
2.2.3.6.监测精度
To investigate potential differences of students’ monitoring accuracy, students judged their expected performance on tasks about electric circuits (“How confident are you that you can solve tasks on the topic of electric circuits correctly?”; Eitel, 2016; Jacob et al., 2020; Nelson et al., 1994) on a scale from 0% (not at all confident) to 100% (very confident). Students rated their judgement of learning 3 times (i.e., before the pretest, after the learning phase, and after the practice phase). As in previous research, we operationalized monitoring accuracy as the difference between students' actual transfer performance and their prospective judgements of learning they have delivered after the learning phase (Eitel, 2016; Finn & Metcalfe, 2014). An overestimation of students’ judged performance occurred when the values were positive, an underestimation occurred when the values were negative. A value of zero indicated correct assessment.
为了探究学生监测准确性的潜在差异,学生们对自己在电路任务中的预期表现进行了判断("你对自己能够正确解决电路主题任务有多大信心?";Eitel, 2016;Jacob et al、2020;Nelson 等人,1994),评分标准从 0% ( 完全没有信心)到 100% ( 非常有信心)。学生们对自己的学习判断进行了 3 次评分(即在预测试之前、学习阶段之后和练习阶段之后)。与之前的研究一样,我们将监测准确性操作化为学生的实际迁移表现与他们在学习阶段后对所学知识的预期判断之间的差异(Eitel, 2016;Finn & Metcalfe, 2014)。如果数值为正,则高估了学生的评判成绩;如果数值为负,则低估了学生的评判成绩。数值为零表示评估正确。
2.2.3.7. Affective measures
2.2.3.7.情感措施
We used the affective slider by Betella and Verschure (2016) to assess students' perceived arousal on a continuous scale from 1 (sleepy/bored) to 9 (wide awake/concentrated) and their perceived pleasure on a continuous scale from 1 (sad) to 9 (happy) as potential proxies for students’ current affective states1 while learning.
我们使用Betella 和 Verschure (2016)的情感滑块来评估学生的感知唤醒度,连续量表从 1(困倦/无聊)到 9(完全清醒/集中注意力)的连续量表来评估学生的唤醒感,以及从 1(悲伤)到 9(快乐)的连续量表,作为学生当前 情绪状态1 学习时的情绪状态。
2.2.3.8. Additional control measures
2.2.3.8.其他控制措施
2.2.3.8.1. Academic self-concept
2.2.3.8.1.学术自我概念
As students' academic self-concept could impact their performance, we assessed their self-concept in the domain of physics by using five items (e.g., “I am good at physics.”, “Physics is easy for me.”, adapted from Gaspard et al., 2015, 2018; McDonald's ω = 0.93) on a 4-point Likert-scale from 1 (does not apply at all) to 4 (fully applies).
由于学生的学业自我概念会影响他们的学习成绩,我们使用五个项目(如 "我擅长物理"、"物理对我来说很容易",改编自Gaspard et al、2015, 2018; McDonald's ω = 0.在 1(完全不适用)到 4(完全适用)的 4 点李克特量表上,麦当劳的ω = 0.93。)
2.2.3.8.2. Intrinsic motivation
2.2.3.8.2.内在动力
We measured students' intrinsic motivation and interest in physics with five items (e.g., “I enjoy physics.”, “I am interested in physical topics.”, adapted from Gaspard et al., 2015, Gaspard et al., 2018; based on Pekrun et al., 2002; McDonald's ω = 0.89) on a 4-point Likert scale from 1 (does not apply at all) to 4 (fully applies).
我们用五个项目测量学生的内在动机和对物理的兴趣(如 "我喜欢物理"、"我对物理话题感兴趣",改编自Gaspard et al、2015,Gaspard et al、2018;基于Pekrun et al、2002; McDonald's ω = 0.89),采用 4 点 Likert 标度,从 1(完全不适用)到 4(完全适用)。
2.2.3.8.3. Attention checks
2.2.3.8.3.注意检查
In our experiment, we included an attention check before and after the practice phase to control for students' attention while completing the online experiment and increase quality of the obtained data and implementation fidelity (see also Hauser & Schwarz, 2016; Oppenheimer et al., 2009, for related approaches). Students were asked to fill in two missing numbers in a very simple 4 × 4 Sudoku. Instructions about how to solve a Sudoku were provided beforehand.
在我们的实验中,我们在练习阶段之前和之后都进行了注意力检查,以控制学生在完成在线实验时的注意力,提高所获数据的质量和实施的可信度(另见Hauser &;Schwarz, 2016; Oppenheimer et al.,2009,了解相关方法)。学生们被要求在一个非常简单的 4 × 4 数独游戏中填入两个缺失的数字。事先提供了有关如何解数独的说明。
2.2.4. Procedure 2.2.4.程序
At the beginning, we informed the students about the aim of the experiment. After providing their written consent, students answered the demographic questionnaire (i.e., demographics, academic self-concept, intrinsic motivation, current arousal, pleasure, and their first judgement of learning). Afterwards, all students answered the pretest. Then, the students were randomly assigned to one of the four conditions (without strategy instruction and without feedback, without strategy instruction but with feedback, with strategy instruction but without feedback, with both strategy instruction and feedback). In the learning phase, students randomly read one of the two versions of the multimedia instruction (strategy instruction vs. no instruction), and assessed their perceived mental effort, subjective difficulty, second judgement of learning, and their current arousal and pleasure. After the first attention check in the practice phase, students solved the two practice tasks, students in the feedback conditions additionally received feedback (see Fig. 3). After the practice phase, all students again rated their perceived mental effort, subjective difficulty, third judgement of learning, current arousal and current pleasure. After completing the second attention check, all students solved the posttest. The study lasted about 30 min and was rewarded with 6 Euros (≈ 6 $).
实验开始时,我们向学生说明了实验的目的。在获得书面同意后,学生们回答了人口统计学问卷(即人口统计学、学术自我概念、内在动机、当前唤醒、愉悦感和对学习的第一判断)。之后,所有学生都进行了前测。然后,学生被随机分配到四个条件之一(无策略指导和无反馈、无策略指导但有反馈、有策略指导但无反馈、有策略指导和反馈)。在学习阶段,学生随机阅读两个版本的多媒体教学(策略指导与无指导)中的一个,并评估他们的感知脑力劳动、主观难度、对学习的第二次判断以及当前的兴奋性和愉悦感。在练习阶段的第一次注意力检查之后,学生们解决了两个练习任务,反馈条件下的学生还得到了反馈(见图 3)。练习阶段结束后,所有学生再次对其感知的脑力劳动、主观难度、第三次学习判断、当前兴奋度和当前愉悦度进行评分。完成第二次注意力检查后,所有学生都进行了后测。学习持续了约 30 分钟,并获得了 6 欧元(≈ 6 美元)的奖励。
2.3. Results 2.3.结果
As effect size measures, we used partial with qualifying values of = .01, .06, .14 as small, medium, and large effects (Cohen, 2013). Moreover, we used an alpha level of α = 0.05. In cases of computing generalized mixed effect models, as estimation method, we used maximum likelihood (Laplace approximation), logit as the link function, and Akaike information criterion (AIC), Bayesian information criterion (BIC), and Log likelihood (LogLik) as goodness-of-fit method (cf. Bono et al., 2021).
在衡量效应大小时,我们使用了部分 ,以 = .01、.06、.14 作为小、中、大效应的限定值(Cohen, 2013)。此外,我们使用了 α = 0.05 的α水平。在计算广义混合效应模型时,我们使用最大似然法(拉普拉斯近似法)作为估计方法,使用 logit 作为链接函数,使用 Akaike 信息准则 (AIC)、Bayesian 信息准则 (BIC) 和 Log likelihood (LogLik) 作为拟合优度方法(参阅 Akaike 信息准则 (AIC))。Bono et al、2021)。
2.3.1. Preliminary analyses
2.3.1.初步分析
Initial boxplot analyses indicated no extreme outliers. Correlations between prior knowledge, the dependent variables and the mediation variables are presented in Appendix A. Separate analyses of variance (ANOVAs) showed that conditions did not differ regarding academic self-concept, F(3, 433) = 1.36, p = .253, = .01, intrinsic motivation, F < 1, baseline arousal, F(3, 433) = 1.75, p = .155, = .01, baseline pleasure, F(3, 433) = 1.60, p = .189, = .01, baseline monitoring accuracy, F(3, 433) = 1.64, p = .179, = .02, prior knowledge, F < 1, and time-on-task, F(3, 433) = 1.02, p = .385, = .01. The descriptive values of all measurements can be seen in Table 1.
初步方框图分析表明,没有出现极端异常值。先验知识、因变量和中介变量之间的相关性见 附录 A 。分别进行的方差分析(ANOVA)显示,在学业自我概念方面,条件没有差异,F(3, 433) = 1.36,p = .253, = .01, 内在动机, F < 1, 基线唤醒, F(3, 433) = 1.75, p = .155, = .01,基线愉悦度,F(3, 433) = 1.60,p = .189, = .01,基线监测准确度,F(3, 433) = 1.64,p = .179, = .02,先验知识,F < 1,任务时间,F(3, 433) = 1.02,p = .385, = .01。所有测量值的描述性数值见 表 1。
Variables 变量 | Without instruction, without feedback (n = 107) 无指导,无反馈(n = 107) | Without instruction, with feedback (n = 111) 无指令,有反馈(n = 111) | With instruction, without feedback (n = 112) 有指导,无反馈(n = 112) | With instruction, with feedback (n = 107) 有指导,有反馈(n = 107) |
---|---|---|---|---|
Pretest 预测试 | ||||
Prior knowledge (0–1) 先验知识 (0-1) | 0.60 (0.22) | 0.59 (0.21) | 0.59 (0.21) | 0.58 (0.22) |
Practice phase 实践阶段 | ||||
Practice tasks (0–1) 实践任务 (0-1) | 0.61 (0.39) | 0.63 (0.38) | 0.75 (0.37) | 0.69 (0.37) |
Posttest 期后测试 | ||||
Transfer tasks (0–1) 转移任务 (0-1) | 0.56 (0.29) | 0.61 (0.30) | 0.62 (0.28) | 0.65 (0.29) |
Near transfer tasks (0–1) 接近移交任务 (0-1) | 0.74 (0.36) | 0.74 (0.39) | 0.71 (0.37) | 0.81 (0.33) |
Far transfer tasks (0–1) 远距离转移任务 (0-1) | 0.38 (0.32) | 0.47 (0.34) | 0.53 (0.33) | 0.50 (0.36) |
Mediator variables 中介变量 | ||||
Mental effort (after instruction) (1–9) 脑力劳动(指导后)(1-9) | 5.36 (1.69) | 5.50 (1.95) | 5.70 (1.84) | 5.61 (2.06) |
Mental effort (after practice phase) (1–9) 脑力劳动(练习阶段后)(1-9) | 5.78 (1.95) | 5.49 (2.19) | 5.17 (1.88) | 4.96 (2.01) |
Subjective difficulty (after instruction) (1–9) 主观难度(教学后)(1-9) | 3.66 (2.17) | 4.60 (2.39) | 4.68 (2.44) | 4.77 (2.52) |
Subjective difficulty (after practice phase) (1–9) 主观难度(练习阶段后)(1-9) | 6.27 (2.46) | 5.37 (2.61) | 4.93 (2.64) | 4.28 (2.55) |
Monitoring accuracy (baseline) (−100%–100%) 监测准确性(基线)(-100%-100%) | 13.67 (24.96) | 6.36 (24.43) | 9.39 (24.86) | 10.29 (23.77) |
Monitoring accuracy (after instruction) (−100%–100%) 监测准确性(指导后)(-100%-100%) | −15.83 (38.78) -15.83 (38.78) | −26.52 (39.42) -26.52 (39.42) | −36.40 (40.71) -36.40 (40.71) | −28.69 (36.99) -28.69 (36.99) |
Monitoring accuracy (after practice phase) (−100%–100%) 监测准确性(练习阶段后)(-100%-100%) | −11.16 (30.41) -11.16 (30.41) | −24.50 (35.30) -24.50 (35.30) | −23.68 (34.74) -23.68 (34.74) | −24.95 (30.32) -24.95 (30.32) |
Arousal (baseline) (1–9) 唤醒(基线)(1-9) | 5.13 (1.80) | 4.65 (1.84) | 5.09 (1.71) | 4.81 (1.91) |
Arousal (after instruction) (1–9) 唤醒(指示后)(1-9) | 5.10 (1.64) | 4.86 (1.70) | 5.10 (1.90) | 5.02 (1.86) |
Arousal (after practice phase) (1–9) 唤醒(练习阶段后)(1-9) | 5.15 (1.84) | 5.14 (2.01) | 5.09 (2.03) | 5.21 (1.97) |
Pleasure (baseline) (1–9) 愉悦感(基线)(1-9) | 6.05 (1.67) | 5.57 (1.78) | 5.77 (1.71) | 5.64 (1.80) |
Pleasure (after instruction) (1–9) 快感(指示后)(1-9) | 5.95 (1.66) | 5.54 (1.70) | 5.83 (1.65) | 5.63 (1.64) |
Pleasure (after practice phase) (1–9) 愉悦(练习阶段后)(1-9) | 5.81 (1.67) | 5.61 (1.74) | 5.80 (1.75) | 5.69 (1.71) |
Control variables 控制变量 | ||||
Self-concept (1–4) 自我概念 (1-4) | 2.38 (0.76) | 2.20 (0.81) | 2.21 (0.70) | 2.29 (0.69) |
Intrinsic motivation (1–4) 内在动力 (1-4) | 2.55 (0.73) | 2.50 (0.73) | 2.45 (0.73) | 2.48 (0.65) |
Time in total (in seconds) 总时间(秒) | 1214.01 (420.32) | 1321.59 (526.60) | 1273.79 (509.66) | 1240.95 (453.03) |
2.3.2. Performance on the practice tasks
2.3.2.练习任务的成绩
We aimed to explore potential differences in students' performance and to examine potential main effects of strategy instruction or feedback, or an interaction effect of both interventions. We realized generalized mixed effect models applying binary logistic regressions with participants as random factor, instruction, feedback and the interaction of instruction and feedback as fixed factors and students' performance on the practice tasks as the dependent variable nested within participants (and repeated because of two scores on practice tasks). Additionally, we controlled for students’ prior knowledge.
我们的目标是探索学生成绩的潜在差异,研究策略指导或反馈的潜在主效应,或两种干预措施的交互效应。我们采用二元逻辑回归的广义混合效应模型,将参与者作为随机因素,将指导、反馈以及指导与反馈的交互作用作为固定因素,并将学生在练习任务中的表现作为因变量,嵌套在参与者中(并因练习任务中的两次得分而重复)。此外,我们还对学生的已有知识进行了控制。
The multilevel analysis showed a significant main effect of strategy instruction, indicating that students who received strategy instruction solved the practice tasks better than students who did not receive the strategy instruction. There was no main effect of feedback and no significant interaction effect (see Table 2 for the entire test statistics).2
多层次分析表明,策略指导具有显著的主效应,表明接受策略指导的学生比没有接受策略指导的学生更容易解决练习任务。反馈没有主效应,也没有显著的交互效应(整个测试统计见 表 2)。2
Variables 变量 | Estimate 估算 | SEb | p |
---|---|---|---|
Practice tasks 实践任务 | |||
Strategy Instruction 策略指导 | .89 | 0.30 | .003 |
Feedback 反馈意见 | .12 | 0.28 | .677 |
Interaction 互动 | −.45 -.45 | 0.41 | .264 |
Transfer tasks (overall) 转移任务(总体) | |||
Strategy Instruction 策略指导 | .68 | 0.22 | .002 |
Feedback 反馈意见 | .39 | 0.22 | .071 |
Interaction 互动 | −.52 -.52 | 0.31 | .089 |
Near transfer 就近转移 | |||
Strategy Instruction 策略指导 | −.05 -.05 | 0.32 | .872 |
Feedback 反馈意见 | .01 | 0.32 | .974 |
Interaction 互动 | .31 | 0.46 | .502 |
Far transfer 远距离转移 | |||
Strategy Instruction 策略指导 | .73 | 0.21 | .001 |
Feedback 反馈意见 | .46 | 0.21 | .033 |
Interaction 互动 | −.60 -.60 | 0.30 | .044 |
Note. Significant p-values are highlighted in bold letters.
注:重要p值以粗体字标出。
2.3.3. Performance on the transfer tasks
2.3.3.转移任务的成绩
We applied the same multilevel analysis as regarding students' performance on the practice tasks but now with students’ performance on the transfer tasks as dependent variable. In line with Hypothesis 1a, we obtained a significant main effect of strategy instruction. The main effect of feedback, as well as the instruction-by-feedback interaction effect were not significant (see Table 2).
我们采用了与学生在练习任务中的表现相同的多层次分析,但现在将学生在迁移任务中的表现作为因变量。与假设 1a 一致,我们得到了策略指导的显著主效应。反馈的主效应以及教学与反馈的交互效应不显著(见 表 2)。
2.3.4. Additional explorative analyses
2.3.4.其他探索性分析
To explore differences between near and far transfer, we additionally conducted explorative separate analyses for near and far transfer to investigate whether instruction and feedback had differential effects on near and far transfer. Therefore, we followed the same procedure but first with students' performance in the near transfer tasks as the dependent variable of the multilevel analysis and in a second separate multilevel analysis students’ performance in the far transfer tasks as the dependent variable.
为了探索近迁移和远迁移之间的差异,我们还对近迁移和远迁移分别进行了探索性分析,以研究教学和反馈是否对近迁移和远迁移产生了不同的影响。因此,我们采用了相同的程序,但首先将学生在近迁移任务中的表现作为多层次分析的因变量,然后在第二个单独的多层次分析中将学生在远迁移任务中的表现作为因变量。
Regarding students near transfer performance as the outcome variable, the results of the multilevel analysis showed that there were no significant main effects neither for strategy instruction nor for feedback, as well as no significant interaction effect (see Table 2). We did not obtain significant differences among conditions, F(3, 433) = 1.51, p = .212, = .01.
关于作为结果变量的学生接近转学成绩,多层次分析结果显示,策略指导和反馈都没有显著的主效应,也没有显著的交互效应(见 表 2)。我们没有发现不同条件下的显著差异,F(3, 433) = 1.51,p = .212, = .01。
In contrast to near transfer, regarding far transfer, we obtained a significant main effect of strategy instruction (see Table 2), indicating that students who received strategy instruction but no feedback (M = 0.53, SD = 0.33) outperformed students who received neither strategy instruction nor feedback (M = 0.38, SD = 0.32, p = .005, d = 0.46). Furthermore, we found a significant main effect of feedback (see Table 2), indicating that students who received feedback outperformed students who received no feedback. However, the main effect of feedback seemed to be not that robust, as the effect was not significant, when we performed an ANCOVA (see Appendix B).
与近迁移相反,在远迁移方面,我们获得了策略指导的显著主效应(见 表 2),表明接受策略指导但没有反馈的学生(M = 0.53,SD = 0.33)优于既未接受策略指导也未获得反馈的学生(M = 0.38, SD = 0.32, p = .005, d = 0.46)。此外,我们还发现了反馈的显著主效应(见 表 2),表明获得反馈的学生成绩优于未获得反馈的学生。然而,当我们进行方差分析时,反馈的主效应似乎并不那么稳健,因为效应并不显著(见 附录 B)。
Regarding the interaction effect of strategy instruction and feedback, the multilevel analysis obtained a significant instruction-by-feedback interaction effect on students’ far transfer performance (see Table 2).
关于策略指导与反馈的交互效应,多层次分析得出了指导与反馈对学生远距离迁移成绩的显著交互效应(见 表 2)。
To break up the significant interaction, we conducted simple effect analyses regarding far transfer. They revealed that when there was no feedback provided, the presence of strategy instruction significantly improved far transfer test performance (p = .007; receiving strategy instruction: M = 0.53, SD = 0.33; without strategy instruction: M = 0.38, SD = 0.32). By contrast, when participants received feedback, there was no effect of strategy instruction (d = 0.09, p = .861; receiving strategy instruction: M = 0.50, SD = 0.36; without strategy instruction: M = 0.47, SD = 0.34). Together, the findings suggest that strategy instruction had an effect on learning only when no additional feedback was provided. An instruction-by-feedback interaction only emerged for higher-order learning outcomes (i.e., far transfer). The results further suggested that additional corrective feedback had no additive effect on the far transfer performance.
为了打破这种显著的交互作用,我们对远距离迁移进行了简单的效应分析。结果显示,当没有提供反馈时,策略指导的存在能显著提高远迁移测试成绩(p = .007;接受策略指导:M = 0.53,SD = 0.33;没有策略指导:M = 0.38, SD = 0.32)。相比之下,当参与者接受反馈时,策略指导没有影响(d = 0.09,p = .861;接受策略指导:M = 0.50,SD = 0.36;未接受策略指导:M = 0.47, SD = 0.34)。总之,研究结果表明,只有在不提供额外反馈的情况下,策略指导才会对学习产生影响。只有在高阶学习成果(即远距离迁移)方面,才会出现教学与反馈之间的相互作用。结果进一步表明,额外的纠正性反馈对远距离迁移成绩没有附加效应。
To explore the underlying processes which could explain the significant instruction-by-feedback interaction effect on far transfer, based on our pre-registration and the ITFL-model (Narciss, 2008), we explored cognitive (i.e., mental effort, subjective difficulty), meta-cognitive (i.e., monitoring accuracy), and affective-motivational processes (i.e., arousal, pleasure) by applying separate simple mediation analyses with one of the mediator variables each. We represented the instruction-by-feedback interaction by creating a new variable named “interaction” which we included as contrast-coded predictor in the regression model (Rosnow & Rosenthal, 1996; Wiens & Nilsson, 2017). We assigned the following contrast weights of the interaction variable according to our theoretical assumptions: −1 = without instruction without feedback, 1 = without instruction with feedback, 1 = with instruction without feedback, −1 = with instruction with feedback (cf. Lachner et al., 2020). Students’ far transfer performance was the dependent variable. We used the PROCESS macro (Version 4) in SPSS (Hayes, 2018) with bootstrapped 95% confidence intervals (10,000 bootstrap samples). None of the univariate mediation analyses approached significance (Appendix C shows the results of all mediation analyses).
中介分析,分别使用其中一个中介变量,对认知过程(即脑力劳动、主观难度)、元认知过程(即监控准确性)和情感动机过程(即唤醒、愉悦)进行探究。我们通过创建一个名为 "交互作用 "的新变量来表示指令与反馈之间的交互作用,并将其作为回归模型中的对比编码预测因子(Rosnow &;Rosenthal,1996;Wiens & Nilsson,2017)。我们根据理论假设为交互变量分配了以下对比权重:-1=无反馈指导,1=有反馈指导,1=无反馈指导,-1=有反馈指导(参见Lachner 等人,2020)。学生的远距离迁移成绩是因变量。我们使用了SPSS (Hayes, 2018)中的PROCESS宏(第4版),并对95%置信区间进行了引导(10,000个引导样本)。 单变量中介分析均未达到显著性( 附录 C显示了所有中介分析的结果)。
In a next step, we conducted additional moderated mediation analyses including valence as moderator variable, as indicated by students‘ performance on the practice tasks, following theoretical suggestions of Grundmann et al. (2021) that negative feedback in particular (i.e., feedback, when the answer was incorrect) could largely affect the proposed mediators (see also Eccles & Wigfield, 2002; Kalyuga & Sweller, 2014; Kluger & DeNisi, 1996; Raaijmakers et al., 2017) and reduce learning performance due to students’ disengagement.
当答案不正确时的反馈)会在很大程度上影响所提出的中介因素(另见Eccles & Wigfield, 2002;Kalyuga &;Sweller, 2014; Kluger &;DeNisi, 1996; Raaijmakers et al.,2017),并因学生脱离学习而降低学习成绩。
Based on these theoretical assumptions, we performed moderated mediation analyses in which a moderator (i.e., W = valence) moderates the a-path of the mediation model (independent variable to mediator) but not the c’-path (direct effect) based on Model 7 of Hayes' (2018) PROCESS macro for SPSS. Interestingly, we found a significant effect of arousal, b = 0.01, SEb = 0.00 (95%, CI [0.0005, 0.0154]; see Fig. 4), as zero was not included in the confidence interval. More precisely, valence moderated the effect of interaction on arousal, suggesting that depending on the experimental condition (instruction only, feedback only, both, or neither), students' valence differed. This affected student arousal and, subsequently, far transfer. None of the other moderated mediation analyses revealed significance (see Appendix D). In summary, the results indicated that the instruction-by-feedback interaction may result in far transfer tasks due to lower levels of arousal. This effect, however, only held for students who received negative feedback during the practice phase.
根据这些理论假设,我们进行了调节性中介分析,在这种分析中,中介(即:W = 评价)调节中介模型的 a-路径(自变量对中介的影响),但不调节 c-路径(直接影响)、W = valence)调节中介模型的 a-路径(自变量到中介变量),但不调节 c'-路径(直接效应),其依据是 Hayes's (2018) SPSS 的 PROCESS 宏的模型 7。有趣的是,我们发现唤醒有显著影响,b = 0.01,SEb = 0.00(95%,CI [0.0005,0.0154];见 图 4),因为置信区间中不包括零。更确切地说,情绪调节了交互作用对唤醒的影响,这表明根据实验条件(仅指导、仅反馈、两者都有或两者都没有)的不同,学生的情绪也不同。这影响了学生的唤醒,进而影响了远迁移。其他调节中介分析均未显示显著性(见 附录 D)。总之,研究结果表明,由于唤醒水平较低,教学与反馈的交互作用可能会导致较远的转移任务。然而,只有在练习阶段收到负面反馈的学生才会产生这种效果。
2.4. Discussion 2.4.讨论情况
In Experiment 1, in line with Hypothesis 1a, we found a main effect of strategy instruction on students' learning outcome. Overall, we did not obtain a main effect of feedback or an instruction-by-feedback interaction effect on students' transfer performance. More fine-grained exploratively analyses in which we investigated students’ near and far transfer performance revealed a significant instruction-by-feedback interaction for far transfer. The combination of strategy instruction and feedback did not contribute more to learning than the single interventions (cf. Shirah & Sidney, 2023, for related findings).
在实验 1 中,与假设 1a 一致,我们发现策略教学对学生的学习结果有主效应。总体而言,我们没有发现反馈对学生迁移成绩的主效应或教学与反馈的交互效应。我们对学生的近距离和远距离迁移成绩进行了更精细的探索性分析,结果显示,在远距离迁移方面,教学与反馈之间存在显著的交互作用。与单一干预相比,策略指导和反馈的组合对学习的贡献并不大(参见Shirah & Sidney, 2023,相关研究结果)。
Further analyses of the underlying processes of the instruction-by-feedback interaction were rather in line with our affective-motivational assumptions. Students who received strategy instruction and negative feedback perceived lower levels of arousal during the feedback. Lower levels of arousal were associated with lower far transfer performance. This pattern, however, only emerged for students who received negative feedback in the practice phase.
对教学与反馈互动的基本过程的进一步分析与我们的情感-动机假设相当吻合。接受策略指导和消极反馈的学生在反馈过程中的唤醒水平较低。较低的唤醒水平与较低的远迁移成绩有关。然而,只有在练习阶段接受负面反馈的学生才会出现这种模式。
The findings of Experiment 1 therefore suggest that instruction-by-feedback interactions only emerge under distinct conditions, namely when corrective feedback was used, only with regard to high-level learning outcomes (i.e., far transfer), and for students who received negative feedback, as they perceived lower levels of arousal. As in the study by Fyfe and Rittle-Johnson (2016), in Experiment 1, we realized rather parsimonious feedback, as students only received corrective feedback and a short-answer explanation, as to why their solution was (in-)correct.
因此,实验 1 的结果表明,教学与反馈之间的相互作用只有在不同的条件下才会出现,即在使用纠正性反馈时,只针对高层次的学习成果(即远迁移),以及接受负面反馈的学生,因为他们认为自己的唤醒水平较低。与Fyfe 和 Rittle-Johnson(2016)的研究一样,在实验 1 中,我们实现了相当简洁的反馈,因为学生只收到了纠正性反馈和简答解释,说明他们的解决方案(不)正确的原因。
The interaction effect can be explained by students’ arousal as a consequence of negative feedback, as similar to recent research by Kuklick and Lindner (2023), the task performance (i.e., valence) moderated the emotional impact of feedback (i.e., arousal). Grundmann et al. (2021) also showed that negative feedback can lead to motivated disengagement, which hinders performance improvements. Thus, the arousal resulting from the negative feedback could have led to loss of motivation and feedback disengagement, which could be another explanation for the reducing effect.
这种交互效应可以用负面反馈导致的学生唤醒来解释,因为与Kuklick 和 Lindner (2023)最近的研究类似,任务表现(即情绪)调节了反馈的情绪影响(即唤醒)。Grundmann 等人(2021)也表明,负面反馈会导致动机脱离,从而阻碍绩效的提高。因此,负面反馈引起的唤醒可能会导致动机丧失和反馈脱离,这可能是减少效应的另一种解释。
Kuklick and Lindner (2023) found that after correct response feedback, students reported more negative emotions than when, for example, elaborated feedback was provided. Furthermore, their findings showed that elaborated feedback increased the perceived usefulness. It is unclear whether the effects of Experiment 1 would replicate (with more robust effects) in settings in which more elaborated feedback is provided. When students would receive more specific information in the feedback, the perceived utility of the feedback and the associated arousal could increase, which could result in higher levels of test performance (Shute, 2008), and a reversal into an additive combination effect.
Kuklick 和 Lindner(2023)发现,与提供详细反馈等情况相比,学生在获得正确的反馈后会产生更多的负面情绪。此外,他们的研究结果表明,精心设计的反馈增加了感知有用性。目前还不清楚实验 1 的效果是否会在提供更详尽反馈的环境中重现(效果更强)。当学生在反馈中接收到更具体的信息时,反馈的感知效用和相关唤醒可能会增加,这可能会导致更高水平的测试成绩(Shute, 2008),并逆转为加法组合效应。
3. Experiment 2 3.实验 2
In Experiment 2, we aimed at generalizing the previous findings to more elaborated feedback (see Section 3.1.2. Design and Material). We hypothesized a main effect for instruction (Hypothesis 1a) and a main effect for elaborated feedback (Hypothesis 1b). Furthermore, we hypothesized an instruction-by-feedback interaction effect (Hypothesis 2).
在实验 2 中,我们旨在将之前的研究结果推广到更详细的反馈中(见第 3.1.2 节)。设计与材料)。我们假设教学具有主效应(假设 1a),精心设计的反馈具有主效应(假设 1b)。此外,我们还假设教学与反馈之间存在交互效应(假设 2)。
3.1. Method 3.1.方法
3.1.1. Participants 3.1.1.与会者
In total, 417 students from non-physics study programs of a German university participated in Experiment 2. We applied the identical exclusion criteria of Experiment 1 (students who were enrolled in physics, engineering, science and technology, or technics: n = 23; students who failed both attention checks: n = 4; students who were not enrolled in study programs: n = 3). In addition, to obtain robust estimates, we excluded students who had taken physics, technology and science, or technics as a major subject in high school to ensure that the sample consists of novice students (n = 46). We further excluded students who had already participated in the first experiment (n = 18). And we only included those students who solved both attention checks correctly (n = 13 failed at least one of the two attention checks). Thus, the final sample comprised N = 310 students as scheduled. The mean age of the students was 22.95 years (SD = 3.63) and 67% were female. The students were on average in their 7.19 semester (SD = 4.84) and most of them were enrolled in humanities in at least one subject (n = 169). The students were comparably distributed among conditions (without instruction without feedback, n = 81; without instruction with feedback, n = 77; with instruction without feedback, n = 78; with instruction with feedback, n = 74).
共有 417 名来自德国一所大学非物理学专业的学生参与了实验 2。我们采用了与实验 1 相同的排除标准(就读物理、工程、科技或技术专业的学生:n = 23;两次注意力检查均未通过的学生:n = 4;未参加学习计划的学生:n = 3)。此外,为了获得稳健的估计值,我们排除了在高中主修物理、技术和科学或技术的学生,以确保样本由新手学生组成(n = 46)。我们进一步排除了已经参加过第一次实验的学生(n = 18)。我们只纳入了那些两次注意力检查都正确的学生(n = 13 人两次注意力检查中至少有一次没有通过)。因此,最终样本包括 N = 310 名如期完成任务的学生。学生的平均年龄为 22.95 岁(SD = 3.63),67% 为女性。学生的平均学期为 7.19 个学期(SD = 4.84),大多数学生至少选修了一门人文学科(n = 169)。学生在不同条件下的分布情况相当(无反馈指导,n = 81;有反馈指导,n = 77;有反馈指导,n = 78;有反馈指导,n = 74)。
3.1.2. Design and material
3.1.2.设计和材料
We implemented a 2 × 2-factorial design with strategy instruction (yes vs. no) and feedback (yes vs. no) as between-participants factors. Participants were randomly assigned to one of four conditions (without strategy instruction and without feedback, n = 81; without strategy instruction but with feedback, n = 77; with strategy instruction but without feedback, n = 78; with instruction and with feedback, n = 74).
我们采用了 2 × 2 因子设计,将策略指导(有 vs. 无)和反馈(有 vs. 无)作为参与者之间的因素。参与者被随机分配到四个条件之一(无策略指导和无反馈,n = 81;无策略指导但有反馈,n = 77;有策略指导但无反馈,n = 78;有指导和反馈,n = 74)。
We used the same material as in Experiment 1 except for the feedback, which was provided also on the process level (Hattie & Timperley, 2007; Narciss, 2013). Students were provided with remedial explanations to additionally enhance their conceptual understanding: Thus, in the feedback the underlying physics principles (i.e., how voltage, resistances and currents are related in a parallel circuit and how to calculate the different partial currents on the basis of voltage and resistances) were explained. Additionally, we showed the correct calculations to enhance their procedural understanding (see Fig. 5). Thus, the feedback comprised more specific information regarding about which solution steps to perform, and in what order, to solve the task correctly, and thus constitutes one possible form of elaborated feedback (Shute, 2008).
除了反馈之外,我们使用了与实验 1 相同的材料,反馈也是在过程层面上提供的(Hattie &;Timperley, 2007; Narciss, 2013)。此外,还为学生提供了补救性解释,以增强他们对概念的理解:因此,我们在反馈中解释了基本的物理原理(即并联电路中电压、电阻和电流之间的关系,以及如何根据电压和电阻计算不同的部分电流)。此外,我们还展示了正确的计算方法,以加深学生对程序的理解(见 图 5)。因此,反馈包含了更具体的信息,说明要正确解决任务,需要执行哪些解决步骤,以及以何种顺序执行,从而构成了一种可能的详细反馈形式(Shute, 2008)。
3.1.2.1. Perceived utility of the practice phase
3.1.2.1.对实践阶段效用的看法
Another modification was that all students indicated their perceived utility of the practice phase (inclusive feedback). To assess the utility of the practice phase, we used five items (e.g., “I have considered the practice phase to be helpful”; McDonald's ω = 0.81) based on Gaspard et al. (2015) on a scale from 1 (not at all useful) to 9 (very useful).
另一项修改是,所有学生都表明了他们对练习阶段(包括反馈)的效用感知。为了评估实践阶段的效用,我们使用了五个项目(例如,"我认为实践阶段很有帮助";麦当劳ω = 0.81)、"我认为实践阶段很有帮助";McDonald's ω = 0.81),其依据是 Gaspard et al.(2015) ,评分标准从 1(完全没用)到 9(非常有用)。
3.1.3. Procedure 3.1.3.程序
The procedure was the same as in Experiment 1.
实验过程与实验 1 相同。
3.2. Results 3.2.结果
3.2.1. Preliminary analyses
3.2.1.初步分析
Initial boxplot analyses indicated no extreme outliers. Correlations between prior knowledge, the dependent variables and the mediation variables are presented in Appendix E. As in Experiment 1, students were comparable among conditions regarding their academic self-concept, F < 1, intrinsic motivation, F(3, 306) = 1.06, p = .364, = .01, their baseline arousal, F(3, 306) = 1.29, p = .279, = .01, their baseline pleasure, F(3, 306) = 2.27, p = .080, = .02, their baseline monitoring accuracy, F(3, 304) = 1.46, p = .226, = .01, their prior knowledge, F(3, 306) = 1.70, p = .167, = .02, and regarding time-on-task, F < 1. The descriptive values of all measurements can be seen in Table 3.
初步方框图分析表明没有极端异常值。先前知识、依赖变量和中介变量之间的相关性见 附录 E。与实验 1 一样,学生在学习自我概念(F < 1)、内在动机(F(3, 306) = 1.06)、p = .364, = .01, 他们的基线唤醒,F(3, 306) = 1.29, p = .279, = .01, 他们的基线愉悦,F(3, 306) = 2.27, p = .080, = .02, 他们的基线监测准确度,F(3, 304) = 1.46, p = .226, = .01,他们的先验知识,F(3, 306) = 1.70,p = .167, = .02,关于任务时间,F < 1。所有测量值的描述性数值见 表 3。
Variables 变量 | Without instruction, without feedback (n = 81) 无指导,无反馈(n = 81) | Without instruction, with feedback (n = 77) 无指导,有反馈(n = 77) | With instruction, without feedback (n = 78) 有指导,无反馈(n = 78) | With instruction, with feedback (n = 74) 有指导,有反馈(n = 74) |
---|---|---|---|---|
Pretest 预测试 | ||||
Prior knowledge (0–1) 先验知识 (0-1) | 0.28 (0.12) | 0.31 (0.10) | 0.27 (0.14) | 0.29 (0.12) |
Practice phase 实践阶段 | ||||
Practice tasks (0–1) 实践任务 (0-1) | 0.60 (0.43) | 0.69 (0.36) | 0.72 (0.38) | 0.74 (0.35) |
Posttest 期后测试 | ||||
Transfer tasks (0–1) 转移任务 (0-1) | 0.54 (0.32) | 0.67 (0.25) | 0.66 (0.30) | 0.72 (0.28) |
Near transfer tasks (0–1) 接近移交任务 (0-1) | 0.67 (0.39) | 0.82 (0.32) | 0.72 (0.37) | 0.82 (0.31) |
Far transfer tasks (0–1) 远距离转移任务 (0-1) | 0.41 (0.33) | 0.52 (0.31) | 0.59 (0.36) | 0.61 (0.36) |
Mediator variables 中介变量 | ||||
Mental effort (after instruction) (1–9) 脑力劳动(指导后)(1-9) | 5.68 (1.95) | 5.56 (1.90) | 5.82 (1.95) | 5.58 (1.90) |
Mental effort (after practice phase) (1–9) 脑力劳动(练习阶段后)(1-9) | 6.23 (1.95) | 5.73 (2.14) | 5.64 (2.28) | 4.77 (2.00) |
Subjective difficulty (after instruction) (1–9) 主观难度(教学后)(1-9) | 3.79 (2.06) | 4.08 (2.28) | 5.60 (2.40) | 4.81 (2.44) |
Subjective difficulty (after practice phase) (1–9) 主观难度(练习阶段后)(1-9) | 6.86 (2.26) | 4.88 (2.49) | 5.28 (2.47) | 4.09 (2.47) |
Monitoring accuracy (baseline) (−100%–100%) 监测准确性(基线)(-100%-100%) | 7.39 (24.15) | 4.52 (25.33) | 9.00 (25.18) | 6.02 (20.86) |
Monitoring accuracy (after instruction) (−100%–100%) 监测准确性(指导后)(-100%-100%) | −20.21 (43.34) -20.21 (43.34) | −26.01 (37.09) -26.01 (37.09) | −30.35 (38.45) -30.35 (38.45) | −27.77 (32.99) -27.77 (32.99) |
Monitoring accuracy (after practice phase) (−100%–100%) 监测准确性(练习阶段后)(-100%-100%) | −26.26 (30.50) -26.26 (30.50) | −20.38 (31.53) -20.38 (31.53) | −21.29 (32.98) -21.29 (32.98) | −20.73 (28.69) -20.73 (28.69) |
Arousal (baseline) (1–9) 唤醒(基线)(1-9) | 5.26 (1.70) | 4.84 (1.91) | 4.99 (1.95) | 4.70 (1.84) |
Arousal (after instruction) (1–9) 唤醒(指示后)(1-9) | 5.26 (1.74) | 5.36 (1.93) | 5.06 (2.13) | 5.18 (1.97) |
Arousal (after practice phase) (1–9) 唤醒(练习阶段后)(1-9) | 5.30 (1.91) | 5.53 (2.09) | 5.22 (2.14) | 5.32 (1.92) |
Pleasure (baseline) (1–9) 愉悦感(基线)(1-9) | 6.01 (1.61) | 5.71 (1.92) | 5.54 (1.73) | 5.31 (1.66) |
Pleasure (after instruction) (1–9) 快感(指示后)(1-9) | 5.84 (1.46) | 5.69 (1.80) | 5.51 (1.83) | 5.41 (1.53) |
Pleasure (after practice phase) (1–9) 愉悦(练习阶段后)(1-9) | 5.40 (1.77) | 5.86 (1.97) | 5.62 (1.72) | 5.66 (1.58) |
Control variables 控制变量 | ||||
Utility of the learning phase (1–9) 学习阶段的效用(1-9) | 6.37 (1.46) | 6.38 (1.45) | 6.56 (1.63) | 6.50 (1.42) |
Utility of the practice phase (1–9) 实践阶段的效用 (1-9) | 4.55 (1.93) | 6.39 (1.61) | 5.56 (1.67) | 6.05 (1.79) |
Self-concept (1–4) 自我概念 (1-4) | 2.15 (0.74) | 2.24 (0.69) | 2.11 (0.73) | 2.13 (0.74) |
Intrinsic motivation (1–4) 内在动力 (1-4) | 2.48 (0.65) | 2.46 (0.64) | 2.34 (0.75) | 2.32 (0.71) |
Time in total (in seconds) 总时间(秒) | 1392.32 (513.07) | 1317.91 (469.48) | 1339.27 (601.89) | 1290.73 (433.72) |
3.2.2. Performance on the practice tasks
3.2.2.练习任务的成绩
As in Experiment 1, we aimed to explore potential differences in students’ performance and to examine potential main effects of strategy instruction or feedback, or an interaction effect of both interventions. We applied the same analyses as in Experiment 1. The multilevel analyses showed that none of the main and interaction effects were significant (see Table 4 for the entire test statistics).3
与实验 1 一样,我们的目的是探索学生成绩的潜在差异,并研究策略指导或反馈的潜在主效应,或两种干预措施的交互效应。我们采用了与实验 1 相同的分析方法。多层次分析显示,主效应和交互效应都不显著(整个测试统计见 表 4)。3
Variables 变量 | Estimate 估算 | SEb | p |
---|---|---|---|
Practice tasks 实践任务 | |||
Strategy Instruction 策略指导 | .83 | .43 | .055 |
Feedback 反馈意见 | .01 | .43 | .984 |
Interaction 互动 | −.07 -.07 | .61 | .905 |
Transfer tasks (overall) 转移任务(总体) | |||
Strategy Instruction 策略指导 | .54 | .23 | .017 |
Feedback 反馈意见 | .47 | .23 | .039 |
Interaction 互动 | −.12 -.12 | .33 | .707 |
Near transfer 就近转移 | |||
Strategy Instruction 策略指导 | .15 | .32 | .641 |
Feedback 反馈意见 | .98 | .36 | .007 |
Interaction 互动 | −.14 -.14 | .51 | .785 |
Far transfer 远距离转移 | |||
Strategy Instruction 策略指导 | .73 | .25 | .004 |
Feedback 反馈意见 | .37 | .25 | .143 |
Interaction 互动 | −.22 -.22 | .35 | .542 |
Note. Significant p-values are highlighted in bold letters.
注:重要p值以粗体字标出。
3.2.3. Performance on the transfer tasks
3.2.3.转移任务的成绩
We proceeded similarly for the performance on the transfer tasks (learning outcome). The multilevel analysis revealed a significant main effect of strategy instruction (see Table 4), indicating that students who received strategy instruction outperformed students who did not receive strategy instruction in the transfer tasks. Contrarily to Experiment 1, we found a significant main effect of elaborated feedback (see Table 4), indicating that students who received feedback during the practice phase performed better in the following transfer tasks than students who did not receive feedback. The instruction-by-interaction effect did not approach significance (see Table 4), indicating that the combination did not lead to detrimental or differential effects. Instead, the fact that strategy instruction as well as feedback had an effect, indicated that the combination of strategy instruction and feedback led to additive effects.
我们对迁移任务(学习结果)的成绩进行了类似的分析。多层次分析显示,策略指导具有显著的主效应(见表 4),表明在迁移任务中,接受策略指导的学生的表现优于未接受策略指导的学生。与实验 1 不同的是,我们发现精心设计的反馈具有显著的主效应(见 表 4),表明在练习阶段接受反馈的学生在接下来的迁移任务中的表现优于未接受反馈的学生。互动教学效果并不显著(见 表 4),这表明互动教学并没有产生有害或不同的效果。相反,策略指导和反馈都有效果这一事实表明,策略指导和反馈的结合产生了叠加效应。
3.2.4. Additional explorative analyses
3.2.4.其他探索性分析
In line with Experiment 1, we conducted fine-grained explorative analyses to determine whether analyzing near and far transfer performances separately yielded different results. We followed the same procedure of analysis but first with students' performance in the near transfer tasks as the dependent variable of the multilevel analysis and in a second separate multilevel analysis students’ performance in the far transfer tasks as the dependent variable.
与实验 1 一致,我们进行了细粒度的探索性分析,以确定分别分析近距离迁移和远距离迁移成绩是否会产生不同的结果。我们采用了相同的分析程序,但首先将学生在近距离迁移任务中的表现作为多层次分析的因变量,而在第二次单独的多层次分析中,将学生在远距离迁移任务中的表现作为因变量。
Regarding students’ near transfer performance, we found a significant main effect of feedback (see Table 4; without strategy instruction but with feedback: M = 0.82, SD = 0.32; without strategy instruction and without feedback: M = 0.67, SD = 0.39). However, there was no main effect of strategy instruction and no significant interaction effect (see Table 4).
关于学生的近迁移成绩,我们发现反馈具有显著的主效应(见 表 4;无策略指导但有反馈:M = 0.82,SD = 0.32;无策略指导且无反馈:M = 0.67,SD = 0.39)。但是,策略指导没有主效应,也没有显著的交互效应(见 表 4)。
In contrast to the results regarding the near transfer performance, regarding far transfer, results revealed a significant main effect of strategy instruction, (see Table 4; with strategy instruction but without feedback: M = 0.59, SD = 0.36; without strategy instruction and without feedback: M = 0.41, SD = 0.33). The effect of feedback and the instruction-by-feedback interaction effect were not significant (see Table 4). Apparently, instruction rather affected far transfer performance, whereas feedback affected near transfer performance. Contrarily to Experiment 1, the interaction effect on far transfer was no longer significant, likely because we provided more elaborated feedback.
与近距离迁移成绩的结果不同,远距离迁移成绩的结果显示,策略指导具有显著的主效应(见表 4;有策略指导但无反馈:M = 0.59,SD = 0.36;无策略指导且无反馈:M = 0.41,SD = 0.33)。反馈效应和教学与反馈的交互效应不显著(见 表 4)。显然,指导影响了远迁移的成绩,而反馈影响了近迁移的成绩。与实验 1 不同的是,远迁移的交互效应不再显著,这可能是因为我们提供了更详尽的反馈。
As another exploratory analysis, to explore potential differences among conditions regarding students' perceived utility of the practice phase, we performed an ANOVA with strategy instruction and feedback as between-participants factors and students’ perceived utility of the practice phase as dependent variable. Results showed a significant positive main effect for feedback, F(1, 302) = 34.39, p < .001, = .10, and a significant instruction-by-feedback interaction effect, F(1, 302) = 11.22, p = .001, = .04.
作为另一项探索性分析,为了探索不同条件下学生对练习阶段效用感知的潜在差异,我们进行了一项方差分析,将策略指导和反馈作为参与者间因素,将学生对练习阶段效用感知作为因变量。结果显示,反馈具有显著的正主效应,F(1, 302) = 34.39, p < .001, = .10,并且教学与反馈之间存在明显的交互效应,F(1, 302) = 11.22, p = .001, = .04。
To break up the significant effects, we computed simple effect analyses. The analyses revealed that when students received no strategy instruction, they found the feedback (M = 6.39, SD = 1.61) in the practice phase more useful than no feedback (M = 4.55, SD = 1.93, p < .001). This effect disappeared for the strategy instruction conditions, as the practice phases with (M = 6.05, SD = 1.79) and without feedback (M = 5.56, SD = 1.67, p = .308) were rated as comparably useful.
为了分解显著效果,我们计算了简单效应分析。分析结果显示,当学生没有接受策略指导时,他们认为练习阶段的反馈(M = 6.39,SD = 1.M = 4.55,SD = 1.93,p < .001)。在策略指导条件下,这种效应消失了,因为有反馈的练习阶段(M = 6.05, SD = 1.79)和无反馈(M = 5.56,SD = 1.67,p = .308)被评为相当有用。
3.3. Discussion 3.3.讨论情况
In Experiment 2 we partly replicated the findings of Experiment 1: In line with Hypothesis 1a, we again obtained a significant main effect of strategy instruction. In line with Hypothesis 1b, but different from Experiment 1, we found a significant main effect of feedback, likely because we provided more elaborated feedback than in Experiment 1. However, we found no instruction-by-feedback interaction (Hypothesis 2). Together, the findings suggested additive effects, as strategy instruction (rather accounted for far transfer) as well as elaborated feedback (rather accounted for near transfer) improved students’ learning outcome.
在实验 2 中,我们部分复制了实验 1 的结果:与假设 1a 一致,我们再次发现策略指导具有显著的主效应。与假设 1b 一致,但与实验 1 不同的是,我们发现反馈具有显著的主效应,这可能是因为我们提供了比实验 1 更详尽的反馈。然而,我们没有发现教学与反馈之间的交互作用(假设 2)。总之,这些研究结果表明,策略指导(而不是远迁移)和精心设计的反馈(而不是近迁移)都能提高学生的学习成绩,从而产生叠加效应。
Regarding the perceived usefulness, it made no difference whether students received the combination of strategy instruction and feedback or only one of them. For the single support methods, however, we found that providing only strategy instruction without feedback led to an increased perceived usefulness of the learning phase and providing only feedback without prior strategy instruction led to an increased perceived usefulness of the practice phase. One possible explanation for why the interaction effect did not occur could be that the elaborated feedback was perceived as more useful than simple corrective feedback in Experiment 1 (Kuklick & Lindner, 2023), and thus even when students responded incorrectly (negative valence), the additional specific information provided by the elaborated feedback led from motivated feedback disengagement to more motivated feedback engagement (Grundmann et al., 2021).
在感知有用性方面,学生是同时接受策略指导和反馈,还是只接受其中一种,并无区别。然而,对于单一的支持方法,我们发现,只提供策略指导而不提供反馈会提高学习阶段的有用性,而只提供反馈而不事先提供策略指导会提高练习阶段的有用性。互动效应之所以没有出现,一个可能的解释是,在实验 1 中,精心设计的反馈被认为比简单的纠正性反馈更有用(Kuklick &;Lindner,2023),因此,即使学生的回答不正确(负效价),详细反馈所提供的额外具体信息也会使学生从积极的反馈脱离到更积极的反馈参与(Grundmann et al.,2021)。
4. Experiment 3 4.实验 3
Given that Experiment 2 showed additive effects of combining strategy instruction and elaborated feedback, the aim of Experiment 3 was to test whether the findings could be replicated when strategy instruction and feedback were provided in a different sequence or whether the additive effects of the combination occurred only when students first go through the learning phase followed by the practice phase. Related studies which examined effects of instruction and practice phases showed that findings might depend on the sequence in which they are provided (see for related evidence: Kant et al., 2017; Kapur, 2008; Loibl et al., 2017; Sinha & Kapur, 2021; van Gog et al., 2008, 2011; van Harsel et al., 2019). In Experiment 3, we therefore contrasted the regular sequence of Experiment 1 and 2 (learning phase with strategy instruction followed by practice phase including feedback) with a flipped sequence (practice phase including feedback followed by learning phase with strategy instruction).
鉴于实验 2 显示了策略指导和详细反馈相结合的叠加效应,实验 3 的目的是检验当策略指导和反馈以不同的顺序提供时,研究结果是否可以重复,或者是否只有当学生首先经历学习阶段,然后是练习阶段时,策略指导和反馈相结合才会产生叠加效应。对指导和练习阶段的效果进行研究的相关研究表明,研究结果可能取决于提供指导和练习的顺序(相关证据见Kant et al、2017;Kapur, 2008;Loibl et al、2017; Sinha & Kapur, 2021; van Gog et al、2008, 2011; van Harsel et al、2019)。因此,在实验 3 中,我们将实验 1 和 2 的常规顺序(先进行策略指导的学习阶段,再进行包括反馈在内的练习阶段)与翻转顺序(先进行包括反馈在内的练习阶段,再进行策略指导的学习阶段)进行了对比。
We stated the following hypotheses: First, we hypothesized that students in both sequences (flipped and regular) would outperform students in a control condition in which neither strategy instruction nor feedback was provided (Hypothesis 1). Second, we hypothesized that the flipped sequence (practice phase with feedback → learning phase with instruction) would result in higher learning outcomes than the regular sequence (learning phase with instruction → practice phase with feedback; Hypothesis 2).
我们提出了以下假设:首先,我们假设两种序列(翻转序列和常规序列)中的学生成绩都会优于对照组中既不提供策略指导也不提供反馈的学生(假设 1)。其次,我们假设翻转序列(有反馈的练习阶段→有指导的学习阶段)比常规序列(有指导的学习阶段→有反馈的练习阶段;假设 2)的学习效果更高。
4.1. Method 4.1.方法
4.1.1. Participants 4.1.1.与会者
We conducted an a priori power analysis for contrast analysis using G*Power, expecting small to medium effects around f = 0.27 (based on Darabi et al., 2018; and related to the findings on problem-solving followed by instruction by Sinha and Kapur (2021); and research on sequencing effects in the context of worked examples by van Gog et al. (2008); see pre-registration for more details, https://aspredicted.org/399mp.pdf). With a set power (1 – β) of 80% (α = 0.05, number of covariates = 1), we required a sample sizes of N = 199. Considering potential dropouts, we aimed to collect data from 240 students.
我们使用 G*Power 对对比分析进行了先验功率分析,预期在 f = 0.27 左右会产生中小型效应(基于 Darabi et al、2018;以及与Sinha 和 Kapur (2021)关于问题解决后再进行指导的研究结果有关;以及 van Gog 等人(2008)在工作示例中对排序效果的研究。(2008 年);更多详情请参见预先登记,https://aspredicted.org/399mp.pdf )。在设定功率 (1 - β) 为 80% 时(α = 0.05,协变量数 = 1),我们需要的样本量为 N = 199。考虑到可能出现的辍学情况,我们的目标是收集 240 名学生的数据。
In total, 244 students from non-physics study programs of a German university participated in Experiment 3. We applied the identical exclusion criteria of Experiment 2. We excluded students who were enrolled in physics, engineering, science and technology, or technology (n = 22), had taken physics, technology and science, or technology as a major subject in high school (n = 26), all whose native language (or language level C1) was not German (n = 0), or who already participated in the previous experiments (n = 9). Furthermore, following the same procedure as in Experiment 1 and 2, to obtain robust estimates, we only included those participants who solved both attention checks correctly (n = 18 failed at least one of the attention checks). Since the dropout was larger than expected, the final sample comprised N = 166 students.
共有 244 名来自德国一所大学非物理学专业的学生参加了实验 3。我们采用了与实验 2 相同的排除标准。我们排除了以下学生:物理、工程、科技或技术专业的学生(n = 22),在高中时以物理、科技或技术为主修科目的学生(n = 26)、母语(或 C1 级语言)均非德语(n = 0),或已参加过之前的实验(n = 9)。此外,为了获得稳健的估计值,我们采用了与实验 1 和实验 2 相同的程序,只包括那些两次注意力检查都正确完成的参与者(n = 18 人至少有一次注意力检查失败)。由于辍学率高于预期,最终样本包括N = 166名学生。
With the acquired sample size, we would be able to detect medium effects of f = 0.22, which was still within the range of effect sizes reported in previous related research (see pre-registration, https://aspredicted.org/399mp.pdf. Thus, we considered the acquired sample size to be acceptable.
根据所获得的样本量,我们将能够检测到f = 0.22的中等效应,这仍在以往相关研究报告的效应大小范围之内(见预登记,https://aspredicted.org/399mp.pdf )。因此,我们认为获得的样本量是可以接受的。
The mean age of the students was 23.42 years (SD = 4.62) and 72% were female. The students were on average in their 7.10 semester (SD = 4.48) and most of them were enrolled in humanities (n = 97).
学生的平均年龄为 23.42 岁(SD = 4.62),72% 为女性。学生的平均年龄为 7.10 个学期(SD = 4.48),大部分学生就读于人文学科(n = 97)。
4.1.2. Design and materials
4.1.2.设计和材料
In contrast to the previous experiments, we used a one-factorial design, and the participants were randomly assigned to one of the three conditions: regular sequence (learning phase with instruction → practice phase with feedback; n = 60), flipped sequence (practice phase with feedback → learning phase with instruction; n = 50), and a control condition which did not receive instruction and feedback (n = 56).
与之前的实验不同,我们采用了单因子设计,参与者被随机分配到三种条件之一:常规序列(有指导的学习阶段→有反馈的练习阶段;n = 60)、翻转序列(有反馈的练习阶段→有指导的学习阶段;n = 50)和不接受指导和反馈的对照条件(n = 56)。
We used the same material as in Experiment 2.
我们使用了与实验 2 相同的材料。
4.1.3. Procedure 4.1.3.程序
We followed the same procedure as in Experiment 1 and 2. As an exception, students in the flipped condition first answered the practice tasks and received feedback on their performance. Afterwards, they were provided with the strategy instruction. The regular sequence was comparable to the combined sequence of Experiment 2.
我们采用了与实验 1 和 2 相同的程序。作为例外,翻转条件下的学生首先回答了练习任务,并获得了成绩反馈。之后,他们再接受策略指导。常规顺序与实验 2 的组合顺序相当。
4.1.4. Data analysis 4.1.4.数据分析
We applied planned contrast analyses to test our hypotheses. To investigate whether there was a general effect of the experimental conditions (H1), we contrasted the experimental conditions (flipped and regular sequence) against the control condition by using the following contrast weights: control condition: −2; regular sequence: 1; flipped sequence: 1. To test whether the flipped sequence additionally resulted in higher learning outcome compared to the regular sequence (H2), we used the following contrast weights: control condition: 0; regular sequence: −1; flipped sequence: 1. Furthermore, we controlled for students’ prior knowledge in both contrast analyses.
我们采用了有计划的对比分析来检验我们的假设。为了研究实验条件是否存在普遍效应(H1),我们使用以下对比权重将实验条件(翻转序列和常规序列)与对照条件进行了对比:对照条件:-2;常规序列:-1;翻转序列:-2:对照组条件:-2;正常序列:1;翻转序列:-2:1;翻转序列:1.为了检验翻转序列是否比常规序列带来更高的学习效果(H2),我们使用了以下对比权重:对照条件:-2;常规序列:1;翻转序列:1:0;常规序列-1;翻转序列:1.此外,我们在两次对比分析中都控制了学生的先验知识。
4.2. Results 4.2.结果
4.2.1. Preliminary analyses
4.2.1.初步分析
Initial boxplot analyses indicated no extreme outliers. Correlations between prior knowledge, the dependent variables and the mediation variables are presented in Appendix G. Students were comparable regarding their academic self-concept, F(2, 163) = 1.88, p = .156, = .02, intrinsic motivation, F < 1, baseline arousal, F < 1, baseline pleasure, F < 1, baseline monitoring accuracy, F < 1, prior knowledge, F < 1, and time-on-task, F < 1. Means and standard deviations are shown in Table 5.
初步方框图分析表明,没有出现极端异常值。先验知识、因变量和中介变量之间的相关性见 附录 G。学生在学习自我概念(F(2, 163) = 1.88,p = .156, = .02,内在动机,F <1,基线唤醒,F <1,基线愉悦,F <1、基线监测准确度,F <1,先验知识,F <1,任务时间,F <1。平均值和标准偏差见 表 5。
Variables 变量 | Without instruction, without feedback (n = 56) 无指导,无反馈(n = 56) | Regular sequence (learning phase – practice phase) (n = 60) 常规序列(学习阶段 - 练习阶段)(n = 60) | Flipped sequence (practice phase – learning phase) (n = 50) 翻转序列(练习阶段-学习阶段)(n = 50) |
---|---|---|---|
Pretest 预测试 | |||
Prior knowledge (0–1) 先验知识 (0-1) | 0.61 (0.23) | 0.60 (0.25) | 0.59 (0.24) |
Practice phase 实践阶段 | |||
Practice tasks (0–1) 实践任务 (0-1) | 0.45 (0.43) | 0.58 (0.40) | 0.61 (0.41) |
Posttest 期后测试 | |||
Transfer tasks (0–1) 转移任务 (0-1) | 0.49 (0.31) | 0.59 (0.30) | 0.60 (0.29) |
Near transfer tasks 近距离转移任务 | 0.59 (0.42) | 0.73 (0.41) | 0.70 (0.38) |
Far transfer tasks 远距离转移任务 | 0.39 (0.34) | 0.45 (0.33) | 0.50 (0.32) |
Mediator variables 中介变量 | |||
Mental effort (after instruction) (1–9) 脑力劳动(指导后)(1-9) | 5.71 (1.8) | 5.67 (2.07) | 5.46 (1.96) |
Mental effort (after practice phase) (1–9) 脑力劳动(练习阶段后)(1-9) | 6.18 (1.71) | 5.37 (2.14) | 5.52 (1.95) |
Subjective difficulty (after instruction) (1–9) 主观难度(教学后)(1-9) | 4.61 (2.05) | 5.75 (2.27) | 5.30 (2.58) |
Subjective difficulty (after practice phase) (1–9) 主观难度(练习阶段后)(1-9) | 6.79 (2.09) | 5.63 (2.67) | 6.10 (2.38) |
Monitoring accuracy (baseline) (−100%–100%) 监测准确性(基线)(-100%-100%) | −30.52 (29.72) -30.52 (29.72) | −28.19 (25.95) -28.19 (25.95) | −26.58 (31.30) -26.58 (31.30) |
Monitoring accuracy practice (−100%–100%) 监测准确性实践 (-100%-100%) | −8.55 (41.78) -8.55 (41.78) | −21.77 (35.29) -21.77 (35.29) | −20.92 (36.07) -20.92 (36.07) |
Monitoring accuracy transfer (−100%–100%) 监测精度转移 (-100%-100%) | −17.41 (36.92) -17.41 (36.92) | −18.17 (33.25) -18.17 (33.25) | −27.32 (30.17) -27.32 (30.17) |
Arousal (baseline) (1–9) 唤醒(基线)(1-9) | 4.73 (2.03) | 4.55 (1.73) | 4.78 (1.82) |
Arousal (after instruction) (1–9) 唤醒(指示后)(1-9) | 4.61 (1.91) | 4.68 (1.79) | 4.94 (1.97) |
Arousal (after practice phase) (1–9) 唤醒(练习阶段后)(1-9) | 4.64 (2.06) | 4.90 (2.01) | 5.32 (1.72) |
Pleasure (baseline) (1–9) 愉悦感(基线)(1-9) | 5.91 (1.76) | 5.82 (1.48) | 5.82 (1.61) |
Pleasure (after instruction) (1–9) 快感(指示后)(1-9) | 5.82 (1.61) | 5.52 (1.56) | 5.36 (1.84) |
Pleasure (after practice phase) (1–9) 愉悦(练习阶段后)(1-9) | 5.62 (1.78) | 5.55 (1.86) | 5.46 (1.63) |
Control variables 控制变量 | |||
Utility of the learning phase (1–9) 学习阶段的效用(1-9) | 6.27 (1.42) | 6.23 (1.29) | 5.98 (2.00) |
Utility of the practice phase (1–9) 实践阶段的效用 (1-9) | 4.62 (1.90) | 5.75 (1.71) | 5.18 (1.80) |
Self-concept (1–4) 自我概念(1-4) | 2.12 (0.62) | 1.99 (0.71) | 2.23 (0.63) |
Intrinsic motivation (1–4) 内在动力 (1-4) | 2.26 (0.60) | 2.13 (0.69) | 2.26 (0.73) |
Time in total (in seconds) 总时间(秒) | 980.73 (364.89) | 1033.35 (385.73) | 1031.06 (357.54) |
4.2.2. Performance on the practice tasks
4.2.2.练习任务的成绩
Coherent to Experiment 1 and 2, we ran generalized mixed effect models with participants as random factor, contrasted condition as fixed factor and students' performance on the practice tasks as the dependent variable nested in participants (and repeated because of two scores on practice tasks). In addition, we controlled for students’ prior knowledge.
与实验 1 和 2 一致,我们运行了广义混合效应模型,将参与者作为随机因素,将对比条件作为固定因素,将学生在练习任务中的表现作为因变量,嵌套在参与者中(由于练习任务中有两次得分,因此重复)。此外,我们还控制了学生的先验知识。
Regarding students’ performance in the practice tasks, the multilevel contrast analyses showed that students in the experimental conditions outperformed students in the control condition (see Table 5 for descriptive statistics of all measures; see Table 6 for the entire test statistics). However, there were no significant differences between the flipped sequence and the regular sequence (see Table 6).4
关于学生在练习任务中的表现,多层次对比分析表明,实验条件下的学生表现优于对照条件下的学生(见表 5所有测量的描述性统计;整个测试统计见 表 6)。然而,翻转序列与常规序列之间没有明显差异(见 表 6)。4
Variables 变量 | Estimate 估算 | SEb | p |
---|---|---|---|
Practice tasks 实践任务 | |||
Control vs. experimental condition (H1) 对照与实验条件(H1) | 1.96 | 0.81 | .016 |
Flipped vs. regular condition (H2) 翻转与常规条件(H2) | .24 | 0.45 | .585 |
Transfer tasks (overall) 转移任务(总体) | |||
Control vs. experimental condition (H1) 对照与实验条件(H1) | .68 | 0.22 | .002 |
Flipped vs. regular condition (H2) 翻转与常规条件(H2) | .39 | 0.22 | .071 |
Near transfer 就近转移 | |||
Control vs. experimental condition (H1) 对照与实验条件(H1) | 1.38 | 0.68 | .042 |
Flipped vs. regular condition (H2) 翻转与常规条件(H2) | .05 | 0.43 | .911 |
Far transfer 远距离转移 | |||
Control vs. experimental condition (H1) 对照与实验条件(H1) | .76 | 0.50 | .129 |
Flipped vs. regular condition (H2) 翻转与常规条件(H2) | .23 | 0.29 | .431 |
Note. Significant p-values are highlighted in bold letters. For H1, we used the following contrast weights: control condition: −2; regular sequence: 1; flipped sequence: 1. For H2, we used the following contrast weights: control condition: 0; regular sequence: −1; flipped sequence: 1. Furthermore, we included students' prior knowledge as covariate in both contrast analyses.
注:显著p值以粗体字标出。对于 H1,我们使用了以下对比权重:对照条件:-2;常规序列:1;翻转序列:1.对于 H2,我们使用了以下对比度权重:控制条件:0;常规序列:-1;翻转序列:-1:-1;翻转序列:1.此外,我们还在对比分析中加入了学生的先验知识作为协变量。
4.2.3. Performance on the transfer tasks
4.2.3.转移任务的成绩
We ran again generalized mixed effect models with participants as random factor, contrasted condition as fixed factor and now students’ performance on the transfer tasks as the dependent variable nested in participants (and repeated because of four scores on transfer tasks). Regarding Hypothesis 1, the multilevel contrast analyses showed that students in the experimental conditions outperformed students in the control condition regarding transfer performance (see Table 5 for descriptive statistics; see Table 6 for the entire test statistics). Regarding Hypothesis 2, there were no differences between the flipped and the regular sequence (see Table 6).
我们再次运行了广义混合效应模型,将参与者作为随机因素,对比条件作为固定因素,现在学生的转学任务成绩作为因变量,嵌套在参与者中(由于转学任务有四个分数,所以重复)。关于假设 1,多层次对比分析表明,在迁移成绩方面,实验条件下的学生优于对照条件下的学生(描述性统计见 表 5;整个测试统计见 表 6)。关于假设 2,翻转序列和常规序列之间没有差异(见 表 6)。
4.2.4. Additional explorative analyses
4.2.4.其他探索性分析
Similar to the previous experiments, we conducted additional explorative separate analyses for near and far transfer performance to identify any differences in near and far transfer performance. We followed the same procedure of multilevel contrast analysis but first with students' performance in the near transfer tasks as the dependent variable and in a second separate multilevel contrast analysis students’ performance in the far transfer tasks as the dependent variable.
与之前的实验类似,我们对近迁移和远迁移成绩进行了额外的探索性单独分析,以确定近迁移和远迁移成绩的差异。我们按照相同的程序进行了多层次对比分析,但首先将学生在近距离迁移任务中的表现作为因变量,然后在第二个单独的多层次对比分析中将学生在远距离迁移任务中的表现作为因变量。
Regarding near transfer performance, the multilevel contrast analyses showed that students in the experimental conditions (M = 0.72, SD = 0.39) outperformed students in the control condition (M = 0.59, SD = 0.42; see Table 6). Regarding far transfer performance, results regarding the first contrast revealed no significant effect (see Table 6; experimental conditions: M = 0.47, SD = 0.32; control condition: M = 0.39, SD = 0.34). There were no differences between the flipped and the regular sequence for near transfer (see Table 6; flipped: M = 0.70, SD = 0.38; regular: M = 0.73, SD = 0.41), and far transfer (see Table 6; flipped: M = 0.50, SD = 0.32, regular: M = 0.45, SD = 0.33).
关于近迁移成绩,多层次对比分析表明,实验条件下的学生(M = 0.72,SD = 0.39)优于对照组学生(M = 0.59,SD = 0.42;见 表 6)。关于远距离转移的表现,第一种对比的结果显示没有显著影响(见 表 6;实验条件:M = 0.47,SD = 0.32;对照条件:M = 0.39,SD = 0.34)。M = 0.70,SD = 0.38;常规:M = 0.73,SD = 0.41),以及远距离转移(见表 6;翻转:M = 0.50,SD = 0.32,常规:M = 0.45,SD = 0.33)。
We conducted an additional ANOVA to explore potential differences between the regular and the flipped sequence regarding the valence of feedback (Hypothesis 2). The contrast-coded condition was the independent variable, and the valence of feedback was the dependent variable. Results showed no significant effect, F(1, 108) = 0.21, p = .651, < .01.
我们又进行了一次方差分析,以探讨常规序列和翻转序列在反馈情绪方面的潜在差异(假设 2)。对比编码条件是自变量,反馈的价值是因变量。结果显示没有明显影响,F(1, 108) = 0.21, p = .651, < .01.
4.3. Discussion 4.3.讨论情况
In Experiment 3, we investigated whether the sequence of strategy instruction and feedback would account for reducing versus additive effects on learning. We first hypothesized that both sequences (flipped and regular) led to higher learning outcome than the control condition (Hypothesis 1). The first hypothesis was confirmed as students in the experimental conditions outperformed students in the control condition. This, effect was mainly driven via near transfer performance. Second, we hypothesized that students in the flipped sequence condition outperform students in the regular sequence condition (Hypothesis 2). In contrast to our assumption, our second hypothesis was not confirmed as the flipped sequence did not contribute to higher learning outcome than the regular sequence. This pattern suggests that the obtained additive findings of elaborated feedback and strategy instruction are independent of the sequence of these instructional interventions.
在实验 3 中,我们研究了策略指导和反馈的顺序是否会对学习产生减少效应或增加效应。我们首先假设两种序列(翻转式和常规式)的学习效果都高于对照组(假设 1)。第一个假设得到了证实,实验条件下的学生成绩优于对照条件下的学生。这种效果主要是通过接近迁移的表现产生的。其次,我们假设翻转序列条件下的学生成绩优于常规序列条件下的学生(假设 2)。与我们的假设相反,我们的第二个假设没有得到证实,因为翻转序列并没有使学习成绩高于普通序列。这种模式表明,精心设计的反馈和策略指导所获得的相加结果与这些教学干预的顺序无关。
5. General discussion 5.一般性讨论
The main aim of the present study was to investigate potential additive versus reducing effects of combining strategy instruction and feedback on learning. Furthermore, we aimed at generalizing our findings to different types of feedback (i.e., corrective vs. elaborated feedback; Experiment 1 and 2), and sequences of strategy instruction and feedback (Experiment 3). In addition, we aimed at exploratively investigating the underlying mechanisms of a potential instruction-by-feedback interaction effect.
本研究的主要目的是调查策略指导与反馈相结合对学习的潜在相加效应和相减效应。此外,我们还旨在将研究结果推广到不同类型的反馈(即纠正性反馈与详细反馈;实验 1 和 2)以及策略指导和反馈的序列(实验 3)中。此外,我们还旨在探索潜在的指导与反馈相互作用效应的内在机制。
5.1. Summary of evidence 5.1.证据摘要
Our findings suggest that, whether combining strategy instruction and feedback results in reducing or additive effects depends on the type of feedback (corrective vs. elaborated feedback) and the type of learning goal (near vs. far transfer).
我们的研究结果表明,将策略指导和反馈结合起来,会产生减少效应还是增加效应,取决于反馈的类型(纠正性反馈还是详细反馈)和学习目标的类型(近距离迁移还是远距离迁移)。
A reducing effect (i.e., an instruction-by-feedback interaction) only emerged when corrective feedback was used, and only with regard to higher-order learning outcome of far transfer (Experiment 1; see also Mertens et al., 2022). Our exploratory mediation analyses suggested that this effect was mediated by the level of arousal, but only for students who received negative feedback (moderated mediation; Kuklick & Lindner, 2023; Laudel & Narciss, 2023).
只有在使用纠正性反馈时,才会出现减少效应(即教学与反馈之间的交互作用),而且只有在高阶学习成果远迁移方面才会出现减少效应(实验 1;另见Mertens 等人,2022)。我们的探索性中介分析表明,这种效应是由唤醒水平中介的,但只针对收到负面反馈的学生(调节中介;Kuklick & Lindner, 2023;Laudel & Narciss,2023)。
Contrarily to Experiment 1, when elaborated feedback was provided (Experiment 2), our findings showed that combining strategy instruction with elaborated feedback led to additive effects and supported students’ learning. A potential explanation of how the reducing effect was counteracted could be that a) students found the elaborated feedback to be useful, as indicated by the significant effect of the perceived usefulness of the practice phase in Experiment 2 and b) that the negative emotions (i.e., arousal), when negative feedback was reported, were reduced (Grundmann et al., 2021; Kuklick & Lindner, 2023). Higher levels of usefulness may have turned motivated feedback disengagement into motivated feedback engagement (Grundmann et al., 2021; Kuklick & Lindner, 2023; Laudel & Narciss, 2023; Narciss, 1999). These findings of Experiment 2 highlight the effectiveness of elaborated feedback (Mertens et al., 2022; Shute, 2008; Wisniewski et al., 2020) not only in isolation, but also in orchestration, that is the combination with other instructional interventions, such as strategy instruction.
与实验 1 不同的是,在提供详细反馈的情况下(实验 2),我们的研究结果表明,将策略指导与详细反馈相结合会产生叠加效应,并支持学生的学习。关于如何抵消减少效应的一个可能的解释是:a) 学生认为精心设计的反馈是有用的,正如实验 2 中练习阶段的感知有用性的显著效果所显示的那样;b) 当负面反馈出现时,学生的负面情绪(即:唤醒)会影响他们的学习、唤醒)减少(Grundmann et al、2021; Kuklick & Lindner, 2023)。更高水平的有用性可能会将动机反馈脱离变成动机反馈参与(Grundmann et al、2021; Kuklick & Lindner, 2023;Laudel & Narciss, 2023; Narciss, 1999)。实验 2 的这些发现凸显了精心设计的反馈的有效性(Mertens et al、2022;Shute, 2008;Wisniewski et al.在《2020年》中,"策略教学"(strategy instruction)不仅是孤立的,而且是与其他教学干预措施(如策略教学)相结合的。
We assumed that another possibility to prevent motivated feedback disengagement (Grundmann et al., 2021) and to enhance feedback processing (Narciss, 2008) could be to provide the feedback at the beginning of the learning unit. Students might then be more motivated to process the feedback, perceive it as more useful, and improve their subjective representation of standards, competencies, and task requirements (Narciss, 2013). Similar to the productive failure approach (Kapur, 2008; Sinha & Kapur, 2021), students would first work through the problems themselves and then, in a subsequent learning phase, validate and deepen their knowledge schemes, and close any knowledge gaps. Based on research on sequence effects (e.g., problem-solving prior instruction; Kant et al., 2017; Loibl et al., 2017; van Gog et al., 2011), we reversed the learning and practice phases in Experiment 3. Contrary to our assumptions, there were no differences between the regular and the flipped sequence in terms of motivation, usefulness of the learning and practice phases, or learning outcomes. Experiment 3 suggested that the sequence does not matter as both sequences supported learning similarly well.
我们认为,防止动机反馈脱离(Grundmann et al、2021)和加强反馈处理(Narciss, 2008),可以在学习单元开始时提供反馈。这样,学生可能会更积极地处理反馈,认为反馈更有用,并改进他们对标准、能力和任务要求的主观表述(Narciss, 2013)。与生产性失败方法(Kapur, 2008;Sinha &;Kapur,2021),学生将首先自己解决问题,然后在随后的学习阶段验证和深化他们的知识方案,并弥补任何知识差距。基于对序列效应的研究(例如,解决问题的先行教学;Kant et al、2017;van Gog 等人,2011),我们在实验 3 中颠倒了学习和练习阶段。与我们的假设相反,在学习动机、学习和练习阶段的有用性或学习结果方面,常规序列和翻转序列之间没有差异。 实验 3 表明,顺序并不重要,因为两种顺序都能很好地支持学习。
The findings of our three experiments extend previous studies on combination effects (Fyfe & Rittle-Johnson, 2016; Salden et al., 2010; Wischgoll, 2017), as they provide distinct information about when and why combining strategy instruction and formative feedback may result in additive versus reducing effects, and allow to generalize the findings across several feedback implementations.
我们的三项实验结果扩展了之前关于组合效应的研究(Fyfe & Rittle-Johnson, 2016;Salden et al、2010;Wischgoll, 2017),因为它们提供了关于何时以及为何将策略指导与形成性反馈相结合可能会产生相加效应或相减效应的独特信息,并允许在多种反馈实施中推广研究结果。
5.2. Theoretical implications
5.2.理论影响
What are the broader contributions of our study? The ITFL-model by Narciss (2008, 2013) proposed a reciprocal interplay between internal (cognitive, meta-cognitive, and motivational-affective processes) and external (e.g., feedback source, instructional context) factors that may affect the effectiveness of feedback (see also e.g., Panadero & Lipnevich, 2022; Winne & Butler, 1994, for related assumptions).
我们的研究有哪些更广泛的贡献?Narciss (2008、2013) 提出了内部(认知、元认知和动机-情感过程)和外部(例如,认知、元认知和动机-情感过程)之间的相互影响。g.,反馈源、教学情境)之间的相互作用(另见,例如Panadero & Lipnevich, 2022;Winne & Butler, 1994,相关假设)。
Regarding the internal factors, as a first contribution, we want to highlight the integrative character of our studies. Still to date, a large proportion of instructional research has relied on investigating cognitive mechanisms or motivational processes in isolation, whereas affective-motivational processes have been considered only to a limited extent (see Azevedo, 2015; Braithwaite et al., 2013; D'Mello & Graesser, 2012; Goldin, 2017; Hoogerheide et al., 2019; Narciss et al., 2014, for exceptions). In our study, we integrated different perspectives (i.e., cognitive, meta-cognitive, affective-motivational) in three well-powered online-experimental studies comprising more than 900 students to draw conclusions about potential mechanisms of our intervention. The integration of different cognitive, meta-cognitive and motivational outcomes allowed us to disentangle and model the various processes which may account for potential interactions.
关于内部因素,我们首先要强调的是我们研究的综合性。迄今为止,大部分教学研究仍依赖于孤立地研究认知机制或动机过程,而情感-动机过程只在有限的范围内得到考虑(见Azevedo, 2015;Braithwaite et al.,2013; D'Mello & Graesser, 2012;Goldin, 2017; Hoogerheide et al.,2019; Narciss et al.)在我们的研究中,我们将不同的视角(即认知、元认知、情感-动机)整合到三项动力充足的在线实验研究中,其中包括 900 多名学生,从而就我们干预措施的潜在机制得出结论。将不同的认知、元认知和动机结果整合在一起,使我们能够对可能产生潜在相互作用的各种过程进行分解和建模。
Moreover, we want to note that the tested explanatory mechanisms were derived from theory (Baars et al., 2013; Grundmann et al., 2021; Narciss, 2008; Sweller et al., 2011; see also pre-registration). However, we have to note that our findings rely only on self-reported measures (i.e., cognitive load, monitoring accuracy, and arousal), as we realized three online experiments. Yet it is unclear, whether and how these measures reflect the proposed theoretical constructs (see Pekrun & Bühner, 2014; Tan et al., 2021, for a critical discussion of self-reports). Replications are needed that additionally integrate less obtrusive measures, such as electrodermal activity or eye-tracking as potential behavioral proxies of (meta-)cognitive and affective-motivational measurements (e.g., Hoogerheide et al., 2019; Potter & Bolls, 2011; Schneider et al., 2019).
此外,我们想指出的是,测试的解释机制来自理论(Baars et al、2021; Narciss, 2008; Sweller et al、2011;另见预登记)。然而,我们必须注意到,我们的研究结果仅依赖于自我报告的测量(即认知负荷、监测准确性和唤醒),因为我们实现了三个在线实验。然而,目前还不清楚这些测量是否以及如何反映所提出的理论构架(见Pekrun & Bühner, 2014;Tan et al、2021 ,对自我报告进行了批判性讨论)。此外,还需要进行重复研究,将干扰性较小的测量方法,如皮电活动或眼动跟踪,作为(元)认知和情感动机测量的潜在行为代用指标(例如:皮电活动或眼动跟踪)纳入研究、Hoogerheide et al、2019; Potter & Bolls, 2011; Schneider et al、2019)。
As a further contribution, regarding the external factors, our study showed that the orchestration of feedback with other instructional strategies plays an important role. We refer to the fact that we deliberately investigated combination effects of two different interventions, that is strategy instruction and computer-based feedback. In educational settings and online learning environments, strategy instruction and feedback are often combined (Magliaro et al., 2005; Rosenshine, 2008), however, most studies often relied on investigating only a single instructional intervention and research investigating to what extent both interventions affect each other's effectiveness to enhance students' learning is rare and produced heterogeneous findings (Fyfe & Rittle-Johnson, 2016; Salden et al., 2010; Wischgoll, 2017). Thus, our study extends prior research as we are contributing to the scarce evidence of combining instructional interventions (Fyfe & Rittle-Johnson, 2016; Salden et al., 2010; Wischgoll, 2017; see also e.g., Murphy & Alexander, 2005).
在外部因素方面的另一个贡献是,我们的研究表明,反馈与其他教学策略的协调发挥着重要作用。我们特意研究了两种不同干预措施的组合效果,即策略指导和基于计算机的反馈。在教育环境和在线学习环境中,策略指导和反馈通常是结合在一起的(Magliaro et al、2005;Rosenshine, 2008)、大多数研究往往只依赖于调查单一的教学干预措施,而调查两种干预措施在多大程度上相互影响以提高学生学习效率的研究并不多见,而且结果也不尽相同(Fyfe &;Rittle-Johnson, 2016; Salden et al.,2010; Wischgoll, 2017)。因此,我们的研究扩展了先前的研究,因为我们为教学干预相结合的稀缺证据做出了贡献(Fyfe &;Rittle-Johnson, 2016; Salden et al.,2010;Wischgoll, 2017;另见如,Murphy & Alexander, 2005)。
Furthermore, our findings support prior research by highlighting the role of feedback implementation (e.g., Hattie & Timperley, 2007; Kluger & DeNisi, 1996; Narciss, 2008; Patchan et al., 2016; Shute, 2008; Strobl et al., 2019). The design of feedback, such as the level of elaboration, may not only impact the effectiveness of feedback per se, but also might influence the effectiveness of combining instructional interventions.
此外,我们的研究结果还支持先前的研究,强调了反馈实施的作用(如Hattie & Timperley, 2007; Kluger &;DeNisi, 1996; Narciss, 2008; Patchan et al.,2016; Shute, 2008; Strobl et al、2019)。反馈的设计,如详细说明的程度,不仅可能影响反馈本身的有效性,还可能影响综合教学干预的有效性。
Summarized, our main contribution is that our findings support the theoretical assumption (Narciss, 2008, 2017; Narciss et al., 2014; cf. Butler & Winne, 1995; Hattie & Timperley, 2007; Hayes, 2012; Shirah & Sidney, 2023) that the effectiveness of feedback depends on internal as well as on external factors: The results of our experiments indicate that whether combining strategy instruction and feedback may result in non-additive/even reducing or additive effects, strongly depends on individual student-related (internal) factors (i.e., task performance/valence, arousal) but also on feedback-related (external) factors (design decisions, i.e., elaboration level, or instructional context, i.e., orchestration with strategy instruction), probably explained by utility.
综上所述,我们的主要贡献在于我们的研究结果支持理论假设(Narciss, 2008,2017;Narciss et al.,2014; cf.Butler & Winne, 1995; Hattie & Timperley, 2007;Hayes, 2012; Shirah &;Sidney,2023),反馈的有效性取决于内部和外部因素:我们的实验结果表明,将策略指导和反馈结合起来是否会产生非加成/偶减成或加成效应,在很大程度上取决于与学生个体相关的(内部)因素(即任务表现/价值、唤醒),同时也取决于与反馈相关的(外部)因素(设计决策,即阐述水平,或教学情境,即与策略指导的协调),这可能可以用效用来解释。
5.3. Practical implications
5.3.实际影响
Our findings also have implications for educational practice, as they show that combining strategy instruction and computer-based feedback is only effective under certain conditions and should therefore be used wisely. Our results showed that all types of support (strategy instruction, feedback, combination) improve learners' performance compared to learners without any instructional interventions. Contrary to the prevalent assumption, however, double support (combination of strategy instruction and feedback) does not necessarily increase learners' performance, as demonstrated by Experiment 1. The findings counteract the commonly held belief that combining instructional interventions, which are regarded to be effective strategies in isolation, contributes to learning. Instead, practitioners should implement elaborated feedback. Our experiments have shown that elaborated feedback can have positive or additive effects on learning both in isolation (Mertens et al., 2022; Shute, 2008) and orchestrated in combination with strategy instruction (cf. Salden et al., 2010).
我们的研究结果还对教育实践产生了影响,因为它们表明,将策略指导和计算机反馈结合起来只有在特定条件下才会有效,因此应明智使用。我们的研究结果表明,与没有任何教学干预措施的学习者相比,所有类型的支持(策略指导、反馈、组合)都能提高学习者的成绩。然而,与普遍的假设相反,双重支持(策略指导与反馈相结合)并不一定会提高学习者的成绩,实验 1 就证明了这一点。这些研究结果反驳了人们普遍持有的观点,即把被认为是有效策略的教学干预结合起来有助于学习。相反,实践者应该实施精心设计的反馈。我们的实验表明,精心设计的反馈对单独学习(Mertens et al、2022;Shute, 2008),并与策略指导相结合进行协调(参见:Shute, 2008)。Salden et al、2010)。
5.4. Limitations and future research
5.4.局限性和未来研究
We would like to note that all detected effects in all three experiments were rather small. Regarding corrective feedback, this is in line with previous research which showed that corrective feedback usually produces small effects (e.g., Van der Kleij et al., 2015; Swart et al., 2019; see Mertens et al., 2022 for an overview). However, regarding elaborated feedback, previous research mostly documented medium effect sizes (see Mertens et al., 2022; Wisniewski et al., 2020, for meta-analytic evidence). It is unclear why our implemented elaborated feedback only produced small but not medium effects. One explanation refers to the fact, that we realized our study as an online experiment. Online experiments have the risk of less control and inattentive behavior by our participants, as compared to laboratory experiments (Hauser & Schwarz, 2016; Oppenheimer et al., 2009). The lower levels of control and attention may result in lower effect sizes, as compared to traditional laboratory studies (see Ryan et al., 2013).
我们要指出的是,所有三个实验中检测到的效果都相当小。关于纠正性反馈,这与之前的研究一致,这些研究表明纠正性反馈通常会产生较小的效果(例如,Van der Kleij et al、2015; Swart et al、2019;概况见Mertens 等人,2022)。然而,关于精心设计的反馈,以往的研究大多记录了中等效果(见 Mertens et al、2022; Wisniewski et al.)目前还不清楚为什么我们实施的详细反馈只产生了小效果,而没有产生中等效果。一种解释是,我们的研究是以在线实验的形式进行的。与实验室实验相比,在线实验可能会减少参与者的控制和注意力不集中行为(Hauser &;Schwarz, 2016; Oppenheimer et al.,2009)。与传统的实验室研究相比,较低的控制和注意力水平可能会导致较低的效应大小(见Ryan 等人,2013)。
Another limitation refers to the number of practice tasks. We implemented only two practice tasks which were realized as a multiple-choice question format. Such practice phases can be regarded to be prototypical for online learning phases. Nevertheless, implementing more challenging learning materials with more practice tasks and more transfer tasks (cf. van Harsel et al., 2020) preferably in combination with open-ended questions could help to further generalize the obtained findings.
另一个限制因素是练习任务的数量。我们只实施了两个以选择题形式实现的练习任务。这种练习阶段可视为在线学习阶段的原型。尽管如此,采用更具挑战性的学习材料,包括更多的练习任务和更多的迁移任务(参见van Harsel 等人,2020),最好与开放式问题相结合,这将有助于进一步推广所取得的研究成果。
Given that our study provides a complex picture of combining instructional interventions and the use of online experiments, we see the need to further investigate these combination effects in practice to replicate the experiments in more authentic settings. Since our findings are limited to one specific topic in physics and the instruction phase and the feedback phase were relatively short (approximately 10–15 min each), it is essential to replicate the findings in different study contexts, with different types of interventions, and with longer durations. Thus, further field-oriented work such as field experiments are required to replicate the findings and to increase the ecological validity of our findings (Renkl, 2013).
鉴于我们的研究提供了教学干预与在线实验相结合的复杂情况,我们认为有必要在实践中进一步研究这些组合效果,以便在更真实的环境中复制实验。由于我们的研究结果仅限于物理中的一个特定主题,而且指导阶段和反馈阶段都相对较短(各约 10-15 分钟),因此必须在不同的研究背景下、使用不同类型的干预措施和更长的持续时间来重复研究结果。因此,需要进一步开展以实地为导向的工作,如实地实验,以复制研究结果并提高我们研究结果的生态有效性(Renkl, 2013)。
Our findings cannot be generalized regarding learning in general. However, we achieved our goal to provide deeper insights of the impact of combining strategy instruction with computer-based feedback on learning in an online setting.
我们的研究结果不能推广到一般的学习中。不过,我们实现了自己的目标,即更深入地了解将策略指导与基于计算机的反馈相结合对在线学习的影响。
5.5. Conclusions 5.5.结论
All in all, the experiments presented in this paper provide important evidence on combining strategy instruction and feedback: The findings illustrate that combining strategy instruction and feedback does not necessarily contribute to learning, as the effectiveness depends on the provided type of feedback and the learning goals. Therefore, they provide important information for fostering cognitive skill acquisition, as they show that not only the design of instructional interventions but also their orchestration is critical to contribute to the effectiveness of digital learning environments.
总之,本文介绍的实验为策略指导与反馈相结合提供了重要证据:研究结果表明,策略指导与反馈相结合并不一定有助于学习,因为其有效性取决于所提供的反馈类型和学习目标。因此,它们为促进认知技能的习得提供了重要信息,因为它们表明,要提高数字学习环境的有效性,不仅要设计教学干预措施,还要对其进行协调。
Funding 资金筹措
This work was supported by the German Research Foundation (DFG) under [contract number LA 4009/1-1].
这项工作得到了德国研究基金会(DFG)[合同号 LA 4009/1-1]的支持。
CRediT authorship contribution statement
CRediT 作者贡献声明
Salome Wagner: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Validation, Visualization, Writing – original draft, Writing – review & editing. Leonie Sibley: Methodology, Formal analysis, Supervision, Writing – review & editing. David Weiler: Methodology, Resources, Writing – review & editing. Jan-Philipp Burde: Methodology, Resources, Writing – review & editing. Katharina Scheiter: Conceptualization, Supervision, Writing – review & editing. Andreas Lachner: Conceptualization, Formal analysis, Funding acquisition, Project administration, Supervision, Writing – review & editing.
萨洛米-瓦格纳(Salome Wagner): 概念化、数据整理、形式分析、调查、方法论、项目管理、验证、可视化、写作 - 原稿、写作 - 审核与编辑。Leonie Sibley: 方法学、正式分析、监督、写作--审阅和编辑。David Weiler: 方法、资源、写作--审阅和编辑。Jan-Philipp Burde: 方法、资源、写作--审查和编辑。Katharina Scheiter: 概念化、监督、写作 - 审核和编辑。安德烈亚斯-拉赫纳(Andreas Lachner): 构思、形式分析、资金获取、项目管理、监督、写作--审阅和编辑。
Declaration of competing interest
利益冲突声明
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
作者声明,他们没有任何可能会影响本文所报告工作的已知经济利益或个人关系。
Acknowledgement 鸣谢
We would like to thank Vincent Hoogerheide of Utrecht University (Netherlands) for sharing his study materials.
我们感谢荷兰乌特勒支大学的 Vincent Hoogerheide 分享他的研究材料。
Appendix A. Correlations of the Variables Across Conditions in Experiment 1
附录 A.实验 1 中不同条件下变量的相关性
Variable | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 |
---|---|---|---|---|---|---|---|---|---|---|---|---|
1. Prior knowledge | ||||||||||||
2. Performance in practice tasks | (.23*|.34| .24*|.41*) | |||||||||||
3. Performance in transfer tasks | (.24*|.32*| .39*|.23*) | (.66*|.60*| .68*|.52*) | ||||||||||
4. Performance in near transfer tasks | (.20*|.28*| .29*|.31*) | (.57*|.55*| .70*|.53*) | (.87|.84*| .84*|.82*) | |||||||||
5. Performance in far transfer tasks | (.21|.23*| .36*|.09) | (.55*|.43*| .40*|.35*) | (.83*|.78*| .80*|.85*) | (.43*|.32*| .34*|.41*) | ||||||||
6. Subjective difficulty after learning phase | (−.22*|−.27*| −.34*|−.36*) | (−.34*|−.44*| −.39*|−.39*) | (−.36*|−.33*| −.47*|−.27*) | (−.36*|−.35*| −.39*|−.29*) | (−.26*|−.17| −.37*|−.17) | |||||||
7. Mental effort after learning phase | (−.30*|−.05| −.12|−.30*) | (−.28*|.18| .21*|−.09) | (−.17|.25*| .20*|−.00) | (−.15|.19*| .18|−.03) | (−.13|.21| .15|.03) | (.40*|.08| .17|.21*) | ||||||
8. Arousal after learning phase | (−.03|−.16| .04|.05) | (.08|.23*| .18|.17) | (.10|.10| .03|.21*) | (.06|.09| .07|.15) | (.10|.08| −.02|.20*) | (.09|−.20*| −.14|−.20*) | (.13|.07| .30*|−.02) | |||||
9. Subjective difficulty after practice phase | (−.42*|−.38*| −.43*|−.42*) | (−.55*|−.72*| −.59*|−.77*) | (−.50*|−.52*| −.62*|−.50*) | (−.41*|−.51*| −.58*|−.49*) | (−.45*|−.32*| −.43*|−.36*) | (−.40*|.54*| .64*|.58*) | (.38*|−.10| .04|.18) | (−.13|−.18| −.13|−.31*) | ||||
10. Mental effort after practice phase | (−.27*|−.29*| −.32*|−.31*) | (−.21*|−.08| −.07|−.35*) | (−.14|.01| −.07|−.16) | (−.12|−.04| −.00|−.21*) | (−.13|.07| −.12|−.07) | (.16|.17| .29*|.48*) | (.43*|.44*| .64*|.56*) | (−.01|.19*| .24|−.07) | (.55*|.38*| .39*|.54*) | |||
11. Arousal after practice phase | (−.04|−.11| .06|.02) | (.10|.41*| .30*|.22*) | (.13|.21*| .15|.25*) | (.11|.19*| .14|.18) | (.10|.15| .10|.23*) | (.12|−.21*| −.18|−.22*) | (.14|.16| .27*|.04) | (.87*|.85*| .88*|.89*) | (−.21*|−.31*| −.25*|−.36*) | (−.01|.22*| .18|−.02) | ||
12. Monitoring accuracy practice tasks | (−.05|−.14| −.20*|−.20*) | (−.78*|−.79*| −.45*|−.78*) | (−.42*|−.46*| −.73*|−.27*) | (−.32*|−.38*| −.58*|−.31*) | (−.40*|−.37*| −.61*|−.15) | (.13|.08| .09|.14) | (.05|−.35*| −.27*|−.09) | (−.04|−.14| −.08|−.01) | (.18|.44*| .29*|.47*) | (.03|−.05| −.06|.07) | (−.03|−.34*| −.22*|−.10) | |
13. Monitoring accuracy transfer tasks | (.01|−.06| −.20*|.03) | (−.33*|−.32*| −.45*|−.23*) | (−.62*|−.71*| −.73*|−.66*) | (−.49*|−.55*| −.58*|−.53*) | (−.57*|−.61*| −.61*|−.58*) | (.06|−.10| .09*|−.05) | (−.15|−.41*| −.26*|−.21*) | (−.04|.00| .08|−.01) | (−.01|.16| .23*|.13) | (−.11|−.15| −.09|−.19) | (−.03|−.12| −.06|−.09) | (.65*|.66*| .75*|.53*) |
Note. * indicates p < .05.
Appendix B. Summary of ANCOVAs Measuring the Effects of Feedback, Strategy Instruction, and its Interaction on Students' Practice Tasks, Overall, Near, and far Transfer Performance in Experiment 1
附录 B.实验 1 中衡量反馈、策略指导及其交互作用对学生的练习任务、总成绩、近迁移成绩和远迁移成绩的影响的方差分析摘要
Variables | F(1, 432) | p | |
---|---|---|---|
Practice tasks | |||
Strategy Instruction | 9.21 | .003 | .02 |
Feedback | 0.34 | .562 | .00 |
Interaction | 2.15 | .144 | .01 |
Transfer tasks (overall) | |||
Strategy Instruction | 4.19 | .041 | .01 |
Feedback | 2.07 | .151 | .01 |
Interaction | 0.36 | .551 | .00 |
Near transfer | |||
Strategy Instruction | 0.51 | .474 | .00 |
Feedback | 2.20 | .139 | .01 |
Interaction | 1.34 | .247 | .00 |
Far transfer | |||
Strategy Instruction | 7.19 | .008 | .02 |
Feedback | 0.72 | .397 | .00 |
Interaction | 4.98 | .026 | .01 |
Note. Significant p-values are highlighted in bold.
Appendix C. Summary of Mediation Analyses Regarding the Instruction-by-Feedback Interaction Effect on Students' Far Transfer in Experiment 1
附录 C.关于实验 1 中教学与反馈对学生远迁移的交互影响的中介分析摘要
Mediator | b | SEb | 95% CI |
---|---|---|---|
Mental effort | .0003 | .0014 | [−0.0029, 0.0032] |
Subjective difficulty | .0033 | .0067 | [−0.0097, 0.0166] |
Judgement of learning | −.0005 | .0060 | [−0.0125, 0.0110] |
Monitoring accuracy | .0186 | .0099 | [−0.0008, 0.0382] |
Arousal | −.0008 | .0025 | [−0.0060, 0.0043] |
Pleasure | −.0009 | .0033 | [−0.0077, 0.0054] |
Note. Significant results are highlighted in bold letters.
Appendix D. Summary of Moderated Mediation with Valence of Feedback as Moderator in Experiment 1
附录 D.实验 1 中以反馈的价值作为调节因素的调节调解摘要
Mediator | b | SEb | 95% CI |
---|---|---|---|
Mental effort | −.0068 | .0058 | [−0.0206, 0.0019] |
Subjective difficulty | −.0003 | .0116 | [−0.0232, 0.0225] |
Judgement of learning | .0051 | .0113 | [−0.0172, 0.0270] |
Monitoring accuracy | .0335 | .0262 | [−0.0168, 0.0859] |
Arousal | .0136 | .0076 | [0.0009, 0.0303] |
Pleasure | .0071 | .0083 | [−0.0088, 0.0247] |
Note. Significant results are highlighted in bold letters.
Appendix E. Correlations of the Variables in Experiment 2
附录 E.实验 2 变量的相关性
Variable | 1 | 2 | 3 | 4 | 5 | 6 |
---|---|---|---|---|---|---|
1. Prior knowledge | ||||||
2. Performance in practice tasks | (.27*|.25*| .26*|.32*) | |||||
3. Performance in transfer tasks | (.31*|.26*| .20|.14) | (.47*|.40*| .68*|.51*) | ||||
4. Performance in near transfer tasks | (.34*|.26*| .22|.19) | (.52*|.37*| .76*|.61*) | (.89*|.81*| .84*|.81*) | |||
5. Performance in far transfer tasks | (.19|.16| .12|.05) | (.28*|.27*| .38*|.27*) | (.85*|.78*| .83*|.86*) | (.53*|.27*| .39*|.41*) | ||
6. Subjective difficulty after learning phase | (−.17|−.19| −.35*|−.33*) | (−.25*|−.30*| −.41*|−.31*) | (−.34*|−.36*| −.32*|−.37*) | (−.33*|−.30*| −.23*|−.36*) | (−.25*|−.26*| −.30*|−.27*) | |
7. Mental effort after learning phase | (−.22|−.32*| .09|−.29*) | (−.08|−.19| .18|.01) | (−.19|−.21| .07|−.04) | (−.07|−.22| .16|.00) | (−.28*|−.11| −.05|−.06) | (.20|.24*| .15|.32*) |
8. Arousal after learning phase | (.25*|−.03| .19|.11) | (.28*|.12| .16|.12) | (.16|−.06| .22|.21) | (.16|−.04| .26*|.15) | (.10|−.06| .11|.20) | (−.37*|−.12| −.24*|−.14) |
9. Subjective difficulty after practice phase | (−.32*|−.16| −.20|−.27*) | (−.37*|−.65*| −.47*|−.79*) | (−.46*|−.35*| −.41*|−.52*) | (−.41*|−.42*| −.39*|−.59*) | (−.40*|−.13| −.29*|−.30*) | (.40*|.34*| .53*|.47*) |
10. Mental effort after practice phase | (−.17|−.10| .04|−.40*) | (−.17|−.26*| −.10|−.25*) | (−.12|−.11| −.16|−.12) | (.02|−.15| −.09|−.11) | (−.25*|−.03| −.18|−.09) | (.02|.21| .25*|.33*) |
11. Arousal after practice phase | (.23*|.04| .13|.08) | (.37*|.29*| .12|.23*) | (.24*|−.00| .26*|.25*) | (.24*|.08| .28*|.19) | (.17|−.09| .14|.23*) | (−.44*|−.02| −.14|−.12) |
12. Monitoring accuracy practice tasks | (−.21|−.27*| .05|−.11) | (−.88*|−.81*| −.74*|−.70*) | (−.33*|−.28*| −.45*|−.20) | (−.38*|−.28*| −.59*|−.26*) | (−.19|−.16| −.16|−.09) | (.05|.02| −.15|−.16) |
13. Monitoring accuracy transfer tasks | (−.09|−.14| .19|.16) | (−.23*|.27*| −.17|.14) | (−.71*|−.56*| −.54*|−.49*) | (−.63*|−.42*| −.37*|−.25*) | (−.62*|−.47*| −.53*|−.54*) | (.06|−.05| −.25*|−.13) |
14. Utility of the learning phase | (−.04|−.10| .14|−.01) | (.06|−.03| .44*|.14) | (−.02|.05| .47*|.25*) | (.01|−.00| .47*|.16) | (−.05|.08| .32*|.26*) | (−.25*|−.18| −.53*|−.21) |
15. Utility of the practice phase | (.27*|.05| .21|−.06) | (.38*|.43*| .51*|.46*) | (.34*|.31*| .33*|.40*) | (.37*|.34*| .39*|.42*) | (.20|.15| .16|.27*) | (−.20|−.17| −.50*|−.05) |
Variable | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 |
---|---|---|---|---|---|---|---|---|
1. Prior knowledge | ||||||||
2. Performance in practice tasks | ||||||||
3. Performance in transfer tasks | ||||||||
4. Performance in near transfer tasks | ||||||||
5. Performance in far transfer tasks | ||||||||
6. Subjective difficulty after learning phase | ||||||||
7. Mental effort after learning phase | ||||||||
8. Arousal after learning phase | (.02|.03| .19|.15) | |||||||
9. Subjective difficulty after practice phase | (.21|.40*| .18|.10) | (−.33*|−.22| −.14|−.14) | ||||||
10. Mental effort after practice phase | (.46*|.53*| .56*|.40*) | (−.07|.02| .06|.10) | (.37*|.56*| .54*|.36*) | |||||
11. Arousal after practice phase | (−.07|−.17| .17|.19) | (.86*|.73*| .91*|.88*) | (−.37*|−.30*| −.17|−.22) | (−.03|−.18| .05|.11) | ||||
12. Monitoring accuracy practice tasks | (−.03|.19| −.20|−.08) | (−.07|.13| .08|.12) | (.11|.41*| .12|.40*) | (.16|.22| −.01|.13) | (−.12|−.15| .04|.00) | |||
13. Monitoring accuracy transfer tasks | (−.01|.02| −.05|.02) | (.12|.35*| .06|.13) | (−.15|−.34*| −.34*|−.21) | (−.03|−.12| −.11|−.02) | (.09|.32*| −.00|.15) | (.37*|.04| .34*|.23*) | ||
14. Utility of the learning phase | (.27*|.25*| .20|.29*) | (.33*|.26*| .44*|.20) | (−.04|.21| −.32*|−.23) | (.15|.18| −.02|.11) | (.27*|.21| .40*|.24*) | (.02|.09| −.08|.04) | (.00|−.10| −.07|−.00) | |
15. Utility of the practice phase | (−.19|−.22| .20|.33*) | (.15|.32*| .25*|.13) | (−.57*|−.40*| −.48*|−.45*) | (−.07|−.19| .00|.15) | (.32*|.36*| .22|.23*) | (−.17|−.33*| −.23*|−.25*) | (.13|.12| .23*|.10) | (−.01|.26*| .59*|.51*) |
Note. * indicates p < .05.
Appendix F. Summary of ANCOVAs Measuring the Effects of Feedback, Strategy Instruction, and its Interaction on Students' Practice Tasks, Overall, Near, and far Transfer Performance in Experiment 2
附录 F.实验 2 中衡量反馈、策略指导及其交互作用对学生的练习任务、总成绩、近迁移成绩和远迁移成绩的影响的方差分析摘要
Variables | F(1, 305) | p | |
---|---|---|---|
Practice Tasks | |||
Strategy Instruction | 3.78 | .051 | .01 |
Feedback | 1.93 | .166 | .00 |
Interaction | 0.37 | .545 | .00 |
Transfer tasks (overall) | |||
Strategy Instruction | 6.43 | .012 | .02 |
Feedback | 8.39 | .004 | .02 |
Interaction | 0.81 | .369 | .00 |
Near transfer | |||
Strategy Instruction | 0.58 | .447 | .00 |
Feedback | 10.18 | .002 | .02 |
Interaction | 0.19 | .662 | .00 |
Far transfer | |||
Strategy Instruction | 12.02 | .001 | .04 |
Feedback | 2.67 | .103 | .01 |
Interaction | 1.13 | .289 | .00 |
Note. Significant p-values are highlighted in bold.
Appendix G. Correlations of the Variables in Experiment 3
附录 G.实验 3 变量的相关性
Variable | 1 | 2 | 3 | 4 | 5 | 6 |
---|---|---|---|---|---|---|
1. Prior knowledge | ||||||
2. Performance in practice tasks | (.41*|.23|.32*) | |||||
3. Performance in transfer tasks | (.11|.32*|.27) | (.42*|.40*|.47*) | ||||
4. Performance in near transfer tasks | (.34*|.34*|.32*) | (.45*|.41*|.42*) | (.85*|.85*|.86*) | |||
5. Performance in far transfer tasks | (−.22|.17|.10) | (.21|.22|.35*) | (.77*|.76*|.80*) | (.33*|.31*|.38*) | ||
4. Subjective difficulty after learning phase | (−.16|−.22|−.21) | (−.17|−.36*|−.20) | (−.16|−.45*|−.30*) | (−.23|−.44*|−.30*) | (−.01|−.28*|−.19) | |
5. Mental effort after learning phase | (−.18|.10|−.03) | (−.03|.20|.10) | (.08|−.03|.11) | (−.05|−.01|.16) | (.20|−.04|.02) | (.30*|.18|−.12) |
6. Arousal after learning phase | (−.13|.19|.07) | (.01|.41*|.33*) | (.11|.29*|.31*) | (.12|.27*|.24) | (.05|.19|.29*) | (.05|−.27*|−.31*) |
7. Subjective difficulty after practice phase | (−.27*|−.32*|−.37*) | (−.40*|−.80*|−.71*) | (−.29*|−.50*|−.44*) | (−.25|−.46*|−.41*) | (−.21|−.35*|−.32*) | (.45*|.60*|.25) |
8. Mental effort after practice phase | (−.04|−.13|.05) | (−.04|−.24|−.24) | (−.08|−.29*|−.09) | (−.13|−.21|.01) | (.00|−.26*|−.18) | (.36*|.38*|.05) |
9. Arousal after practice phase | (−.15|.28*|.11) | (.19 |.47*|.30*) | (.03|.26*|.17) | (.09|.21|.12) | (−.06|.21|.17) | (−.14|−.21|−.10) |
10. Monitoring accuracy after learning phase | (−.35*|−.01|−.07) | (−.83*|−.76*|−.76*) | (−.26*|−.16|−.20) | (−.24|−.18|−.16) | (−.18|−.07|−.17) | (−.11|−.12|−.06) |
11. Monitoring accuracy after practice phase | (−.00|.04|.08) | (−.08|.26*|.04) | (−.70*|−.52*|−.63*) | (−.54*|−.41*|−.47*) | (−.62*|−.44*|−.58*) | (−.19|−.20|.12) |
12. Utility of the learning phase | (.12|.13|.11) | (.04|.45*|.47*) | (.31*|.50*|.53*) | (.32*|.45*|.50*) | (.17|.34*|.38*) | (−.45*|−.37*|−.48*) |
13. Utility of the practice phase | (.12|.13|.11) | (.30*|.56*|.51*) | (.19|.33*|.42*) | (.13|.40*|.38*) | (.18|.10|.30*) | (−.37*|−.30*|−.35*) |
Variable | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 |
---|---|---|---|---|---|---|---|---|
1. Prior knowledge | ||||||||
2. Performance in practice tasks | ||||||||
3. Performance in transfer tasks | ||||||||
4. Performance in near transfer tasks | ||||||||
5. Performance in far transfer tasks | ||||||||
4. Subjective difficulty after learning phase | ||||||||
5. Mental effort after learning phase | ||||||||
6. Arousal after learning phase | (.00|.25*|.43*) | |||||||
7. Subjective difficulty after practice phase | (−.04|−.06|.03) | (−.06|−.28*|−.22) | ||||||
8. Mental effort after practice phase | (.34*|.43*|.44*) | (.29*|.01|.16) | (.18|.53*|.22) | |||||
9. Arousal after practice phase | (.07|.09|.28) | (.81*|.80*|.82*) | (−.12|−.31*|−.22) | (.23|−.07|.30*) | ||||
10. Monitoring accuracy after learning phase | (.15|−.22|.11) | (.11|−.15|−.10) | (.06|.39*|.50*) | (.05|.03|.36*) | (.00|−.29*|−.13) | |||
11. Monitoring accuracy after practice phase | (−.10|.12|−.01) | (−.02|.12|−.13) | (−.24|−.32*|−.03) | (−.15|−.15|.08) | (.12|.10|.01) | (.23|.05|.21) | ||
12. Utility of the learning phase | (−.00|.23|.28) | (.12|.28*|.51*) | (−.27|−.45*|−.38*) | (−.18|−.07|.14) | (.17|.34*|.41*) | (.17|−.26*|−.11) | (−.07|−.09|−.23) | |
13. Utility of the practice phase | (−.05|.26|.15) | (.19|.35*|.28) | (−.59*|−.47*|−.43*) | (.01|−.01|.11) | (.18|.44*|.31*) | (−.04|−.47*|−.27) | (.22|.03|−.04) | (.45*|.68*|.58*) |
Note. * indicates p < .05.
Appendix H. Results of the Contrast Analyses in Experiment 3
附录 H.实验 3 的对比分析结果
Variables | t(162) | p | d |
---|---|---|---|
Practice tasks | |||
Control vs. experimental condition (H1) | 2.33 | .021 | 0.34 |
Flipped vs. regular condition (H2) | 0.55 | .580 | 0.07 |
Transfer tasks (overall) | |||
Control vs. experimental condition (H1) | 2.34 | .027 | 0.37 |
Flipped vs. regular condition (H2) | 0.21 | .831 | 0.03 |
Near transfer | |||
Control vs. experimental condition (H1) | 2.12 | .036 | 0.33 |
Flipped vs. regular condition (H2) | −0.36 | .716 | −0.08 |
Far transfer | |||
Control vs. experimental condition (H1) | 1.52 | .131 | 0.25 |
Flipped vs. regular condition (H2) | 0.80 | .427 | 0.15 |
Note. Significant p-values are highlighted in bold. For H1, we used the following contrast weights: control condition: −2; regular sequence: 1; flipped sequence: 1. For H2, we used the following contrast weights: control condition: 0; regular sequence: −1; flipped sequence: 1. Furthermore, we included students' prior knowledge as covariate in both contrast analyses.
References 参考资料
- Adesope et al., 2017 Adesope 等人,2017 年Rethinking the use of tests: A meta-analysis of practice testing
反思测试的使用:对实践测试的元分析 - Allen et al., 2016 艾伦等人,2016 年Computer-based writing instruction
基于计算机的写作教学C.A. MacArthur, S. Graham, J. Fitzgerald (Eds.), Handbook of writing research (2016), pp. 316-329
C.A.麦克阿瑟、S.格雷厄姆、J.菲茨杰拉德(编),《写作研究手册》(2016 年),第 316-329 页2016 - Arnsten, 2009 阿恩斯滕,2009 年Stress signalling pathways that impair prefrontal cortex structure and function
损害前额叶皮层结构和功能的压力信号通路 - Atkinson et al., 2000 阿特金森等人,2000 年Learning from examples: Instructional principles from the worked examples research
从实例中学习:从实例研究中得出的教学原则 - Azevedo, 2015 阿泽维多,2015 年Defining and measuring engagement and learning in science: Conceptual, theoretical, methodological, and analytical issues
界定和衡量科学学习的参与度:概念、理论、方法和分析问题 - Baars et al., 2013 巴尔斯等人,2013 年Completion of partially worked-out examples as a generation strategy for improving monitoring accuracy
完成部分已完成的示例,作为提高监测准确性的生成策略 - Bangert-Drowns et al., 1991
Bangert-Drowns 等人,1991 年The instructional effect of feedback in test-like events
测试类活动中反馈的教学效果 - Barbieri et al., 2023 Barbieri 等人,2023 年A meta-analysis of the worked examples effect on mathematics performance
工作实例对数学成绩影响的元分析 - Betella and Verschure, 2016
贝泰拉和弗胥,2016 年The affective slider: A digital self-assessment scale for the measurement of human emotions
情感滑块:用于测量人类情绪的数字自评量表 - Boksem et al., 2005 Boksem 等人,2005 年Effects of mental fatigue on attention: An ERP study
精神疲劳对注意力的影响:ERP研究 - Boksem et al., 2006 Boksem 等人,2006 年Mental fatigue, motivation and action monitoring
精神疲劳、动力和行动监测 - Bono et al., 2021 波诺等人,2021 年Report quality of generalized linear mixed models in psychology: A systematic review
心理学中广义线性混合模型的报告质量:系统回顾666182 - Braithwaite et al., 2013 布雷斯怀特等人,2013 年A guide for analysing electrodermal activity (EDA) & skin conductance responses (SCRs) for psychological experiments
用于心理实验的皮电活动(EDA)和皮肤传导反应(SCR)分析指南Psychophysiology, 49 (1) (2013), pp. 1017-1034
《心理生理学》,49 (1) (2013),第 1017-1034 页 - Brooks et al., 2019 布鲁克斯等人,2019 年A matrix of feedback for learning
学习反馈矩阵 - Butler and Winne, 1995 巴特勒和温尼,1995 年Feedback and self-regulated learning: A theoretical synthesis
反馈与自我调节学习:理论综述Review of Educational Research, 65 (3) (1995), pp. 245-281, 10.2307/1170684