Does learning a skill with the expectation of teaching it impair the skill's execution under psychological pressure if the skill is learned with analogy instructions?

OBJECTIVE: Having learners practice a motor skill with the expectation of teaching it (versus an expectation of being tested on it) has been revealed to enhance skill learning. However, this improvement in skill performance is lost when the skill must be performed under psychological pressure due to 'choking under pressure.' The present study will investigate whether this choking effect is caused by an accrual of declarative knowledge during skill practice and could be prevented if a technique (analogy instructions) to minimize the accrual of declarative knowledge during practice is employed.
DESIGN: We will use a 2 (Expectation: teach/test) x 2 (Instruction: analogy/explicit) x 2 (Posttest: high-pressure/low-pressure) mixed-factor design, with repeated measures on the last factor.
METHODS: A minimum of 148 participants will be quasi-randomly assigned (based on sex) to one of four groups. Participants in the teach/analogy and teach/explicit groups will practice golf putting with the expectation of teaching putting to another participant, and analogy instructions or explicit instructions, respectively. Participants in the test/analogy and test/explicit groups will practice golf putting with the expectation of being tested on their putting, and analogy instructions or explicit instructions, respectively. The next day all participants will complete low- and high-pressure putting posttests, with their putting accuracy serving as the dependent variable.

Introduction

Determining practice conditions that enhance motor learning is important to facilitate motor behavior. The value of practice conditions that enhance motor learning depends on whether the learning benefits are transferred to novel contexts (Schmidt & Lee, 2019), particularly those likely to be encountered while performing the skill and those with high importance. For example, a practice condition may improve a learner’s encoding and consolidation of a skill, however the practice condition’s efficacy is limited if the skill cannot be successfully retrieved and performed in high-stakes environments, under psychological pressure. As many skills must be performed in high-stakes environments, such as sports competition, it is crucial to determine practice conditions that enhance learning and preserve learning benefits under psychological pressure. Recently, Daou, Hutchison, et al. (2019) revealed that practicing a motor skill with the expectation of teaching it to another person loses its benefit when the learned skill is performed under psychological pressure. Therefore, the purpose of the present study is to determine whether the expecting to teach approach can be modified to preserve its learning advantage, and, in so doing, shed light on the mechanisms underlying the loss of the benefit under psychological pressure.

Some initial research of the expecting to teach approach showed that when participants study academic information with the expectation of teaching it, they exhibit augmented learning (Bargh & Schul, 1980; Benware & Deci, 1984; Nestojko, Bui, Kornell, & Bjork, 2014). However, other studies failed to reveal this effect (Renkl, 1995; Ross & Di Vesta, 1976) or demonstrated ambiguous learning effects (enhancements on short-term, but not long-term, test performance; Fiorella & Mayer, 2013; Fiorella & Mayer, 2014). Daou, Buchanan, Lindsey, Lohse, and Miller (2016) conducted the first investigation into whether expecting to teach enhances learning of motor skills, which rely more heavily on procedural knowledge than academic information does (Rosenbaum, Carlson, & Gilmore, 2001). Daou, Buchanan et al. observed having learners practice and study a motor skill with the expectation of teaching it to another person enhanced skill learning in comparison to having learners practice and study a skill with the expectation of being tested, and this effect has been replicated several times (Daou, Hutchison et al., 2019; Daou, Lohse, & Miller, 2016; Daou, Lohse, & Miller, 2018; Daou, Rhoads, Jacobs, Lohse, & Miller, 2019). Although research has failed to reveal the mechanisms underlying the learning benefit of expecting to teach, studies have consistently shown that the learning advantage occurs concomitant to large gains in declarative knowledge about the learned skill (Daou, Buchanan et al., 2016; Daou et al., 2018; Daou, Hutchison et al., 2019; Daou, Lohse et al., 2016; Daou, Rhoads et al., 2019). As motor skills acquired with large gains in declarative knowledge are highly susceptible to decrement under psychological pressure (Lam, Maxwell, & Masters, 2009a, 2009b; Hardy, Mullen, & Jones, 1996; Koedijker, Oudejans, & Beek, 2007; Liao & Masters, 2001; Masters, 1992), it was unsurprising that Daou, Hutchison, et al. (2019) revealed that the expecting to teach benefit vanished under psychological pressure, due to participants who practiced with the expectation of teaching ‘choking’ in a high-pressure posttest. Daou, Hutchison et al. concluded that participants who practiced with the expectation of teaching choked likely due to their accrual of declarative knowledge while practicing, however the authors were unable to provide evidence to support this conclusion. Nonetheless, their conclusion is consistent with reinvestment theory (Masters & Maxwell, 2008), which contends that dispositional and situational factors, such as psychological pressure, trigger individuals to use declarative knowledge acquired earlier in learning to attempt to consciously monitor and control practiced movements. This focus of attention on movement, ironically, impairs performance (Wulf, 2013). Critically, learners who accrue more declarative knowledge during skill practice are more likely to exhibit performance decrement under pressure, because they have more declarative knowledge to ‘reinvest’ in motor control.

A corollary of reinvestment theory is that motor skills learned relatively implicitly, with minimal gains in declarative knowledge, should be resilient to psychological pressure (Masters & Maxwell, 2008), and research generally supports this proposition (Hardy et al., 1996; Koedijker et al., 2007; Lam, Maxwell, & Masters, 2009b, 2009a; Liao & Masters, 2001; Masters, 1992). An effective strategy to encourage implicit motor learning is to provide learners with an analogy about how to perform the skill rather than explicit rules, strategies, and techniques regarding skill performance (Lam et al., 2009b, 2009a; Liao & Masters, 2001). With an analogy, declarative knowledge about multiple rules is reduced into a single, comprehensive rule. For example, Lam et al. (2009a) instructed participants in an analogy practice condition to “shoot as if you are trying to put cookies into a cookie jar on a high shelf” (p.344) while practicing a basketball free throw, whereas participants in an explicit practice condition group were instructed to follow a list of eight specific rules while practicing. Participants in both conditions performed low- and high-pressure posttests and were asked to recall free throw shooting rules. Participants who practiced in the analogy condition reported fewer rules, indicative of more implicit learning, and performed equally well under low- and high-pressure posttests, whereas the explicit condition group performed worse under the high-pressure than the low-pressure posttest (i.e., they choked under pressure).

Since the expecting to teach approach is a practical way to enhance motor learning, it would be beneficial to determine a way to maintain the learning advantage under psychological pressure. As Daou, Hutchison, et al. (2019) attributed the choking effect exhibited by participants who practiced with the expectation of teaching to the accrual of declarative knowledge, a promising means to prevent the choking effect is to promote implicit learning by instructing learners to use an analogy to practice a motor skill rather than a list of rules. Indeed, Daou, Hutchison et al. (2019) asked participants to study an instruction booklet containing a list of rules to follow while practicing the skill, likely prompting learners who expected to teach to attend to the rules so that they could disseminate them to another person; an analogy instruction would reduce this attention to rules. Importantly, it is unlikely that minimizing the accrual of declarative knowledge by learners who expect to teach will reduce their learning advantage, as declarative knowledge has been found to not significantly relate to motor learning in an expecting to teach paradigm (Daou, Buchanan et al., 2016). Even with analogy instructions, it is possible that learners who expect to teach could accrue greater declarative knowledge by engaging in more learning activities, such as discovery learning and hypothesis testing, than those who expect to test. Nonetheless, the practical question of whether the choking effect associated with the expecting to teach approach is prevented by using analogy instructions can still be answered.

The present study will investigate whether having learners practice a motor skill with the expectation of teaching it and using an analogy to practice it preserves the learning advantage of expecting to teach under psychological pressure. Specifically, participants will be assigned to four groups. One group will practice with the expectation of teaching a motor skill and will receive an analogy instruction (teach/analogy); one group will practice with the expectation of teaching the skill and will receive specific explicit rules related to the skill (teach/explicit); one group will practice with the expectation of being tested on the skill and will receive an analogy instruction (test/analogy); and one group will practice with the expectation of being tested on the skill and will receive explicit rules about the skill (test/explicit). One day after skill practice (6 blocks of 10 putts on a single day), all groups will perform low- and high-pressure posttests. With this 2 (Expectation: teach/test) x 2 (Instruction: analogy/explicit) x 2 (Posttest: low-pressure/high-pressure) design, we predict a 3-way interaction. In particular, we predict participants in the teach groups will exhibit superior posttest performance on the low-pressure posttest relative to their test group counterparts, but the effect of expecting to teach on performance in the high-pressure posttest will be moderated by instruction. Specifically, expecting to teach will be advantageous for participants who trained with analogy instructions, but not for participants who trained with explicit instructions. This result would indicate that practicing a motor skill with the expectation of teaching and an analogy imparts a learning advantage that can be manifested in a high-stakes environment. Crucially, this result would also strongly suggest that the reason learners who practice with the expectation of teaching choke under pressure is due to their accrual of declarative knowledge while practicing, thus addressing a shortcoming of Daou, Hutchison, et al. (2019). However, if the choking effect associated with the expecting to teach approach is not prevented by employing analogy instructions, this would not eliminate the possibility that the choking effect is caused by the accrual of declarative knowledge. Specifically, learners who practice with the expectation of teaching may accrue a relatively large amount of declarative knowledge despite receiving analogy instructions and use this knowledge during the high-pressure posttest. Importantly, we can use free recall tests of declarative knowledge use during the high-pressure posttest to shed light on this possibility.

Methods

Sample

Men and women between the ages of 18 and 30 years will participate in the study and may receive course credit for participation. This demographic is convenient to the investigators and has been used in similar past studies (e.g., Daou, Hutchison et al., 2019). Participants must have putted (anything from playing miniature golf to playing 18 holes on a standard golf course) between one and thirty times in their lifetime and not more than twenty times in the past year. Participants with this amount of experience were most sensitive to the expecting to teach and pressure manipulations in past experiments (Daou, Hutchison et al., 2019). Since these participants have at least minimal putting experience, the instructions and practice likely afford them an opportunity to improve their skill by internalizing the analogy/explicit rules rather than guiding them through a completely novel movement, and participants who expect to teach may especially take advantage of this opportunity in preparation for their teaching episode. Participants must be free from physical illness, injury, or disability that could make putting difficult. Participants will be asked to refrain from alcohol/drug consumption within 24 h of both days of the study, caffeine consumption within 3 h of both days of the study, and to get a good amount of sleep the night before each day of the study while also trying to get the same amount of sleep each night.

Sample size calculation

Since our study is novel, we are unable to estimate the effect size for an Expectation x Instruction x Posttest interaction. Thus, we are powering our study to detect an Expectation x Posttest interaction, which we estimate to be medium (η2 p = .093) in our sample, based on the effect size observed in our past research among participants who meet the inclusion and outlier criteria for the present study (Daou, Hutchison et al., 2019). We are powering the study to detect this interaction because the given instructions to the participants should moderate it to a relatively large degree, based on past research investigating the effects of instructions on posttest performance (Lam et al., 2009a; Liao & Masters, 2001). To do the power analyses, we used G*Power 3.1.9.2 (Faul, Erdfelder, Lang, & Buchner, 2007) and entered the aforementioned effect size (as in SPSS) along with the following parameters: α = .05, power = .9, number of groups = 4, number of measurements = 2, and nonsphericity correction ϵ = 1 into an ANOVA with repeated-measures testing for a within-between interaction. The required sample size was determined to be 148, but we will collect data from 164 participants to account for data loss (e.g., participant dropout, problems with data entry). The final sample submitted to statistical analysis will include at least 148 participants, and we will ensure equal n in each group (by recruiting additional participants, if necessary). In terms of data exclusion, we will only exclude participants if there is a technical error in recording their putts at the posttest or if one of their average low- or high-pressure posttest radial error values has a z-score > 3.00. In the latter case, we will report the primary statistical results with and without the inclusion of the participant.

Task

All participants will use a standard (88.9 cm) golf putter to putt a standard golf ball from a starting position indicated by a 5 cm line painted in white washable paint on an artificial grass surface to a target cross (+) comprised of two 10.8 cm lines painted in white washable paint and located 300 cm away from the starting position (Daou, Hutchison et al., 2019). Participants’ objective will be to make the ball stop as close to the center of the target as possible.

Procedure

All participants will complete the experiment individually. After consenting to the experiment, participants will complete a demographic questionnaire asking their sex, age, putting experience, any illness, injury, or disability that could make putting difficult, whether they consumed alcohol/drugs within the last 24 h, whether they consumed caffeine within the last 3 h, and how long they slept the previous night (see Appendix A). Once the experimenter confirms that the participants meet the inclusion criteria (see Sample section), participants will complete the Movement Specific Reinvestment Scale (Masters, Eves, & Maxwell, 2005). The Movement Specific Reinvestment Scale is frequently used to examine individual tendencies to reinvest in motor control (Huffman, Horslen, Carpenter, & Adkin, 2009; Kal et al., 2015; Klämpfl, Lobinger, & Raab, 2013; Malhorta, Poolton, Wilson, Ngo, & Masters, 2012; Vine, Moore, Cooke, Ring, & Wilson, 2013) and possesses good psychometric properties (Masters et al., 2005). The Movement Specific Reinvestment Scale consists of the conscious motor processing and movement self-consciousness subscales, which ask participants to indicate how strongly they agree with statements related to their tendency to attempt to control their movements and monitor their movements, respectively. Participants will respond on a 6-point scale anchored by "strongly disagree" and "strongly agree" (See Appendix B.). The Movement Specific Reinvestment Scale data may be used to explore whether individual tendencies toward reinvestment explain residual variance in the model, thus increasing the amount of variance explained by the other factors in the model, as was the case in Daou, Hutchison, et al. (2019). Next, participants will put a physiological monitoring device around their chest (BioHarness 3.0, Zephyr Technology, Annapolis, MD) to get used to wearing it, which they will be asked to do the following day as well. Physiological data such as heart rate and heart rate variability may be extracted from the device for supplemental/exploratory analyses.

Pretest. After completing the demographic questionnaire, participants will perform the pretest phase, which will consist of one block of ten putts.

Practice. After pretest, participants will be quasi-randomly assigned (based on sex) to the teach/analogy, teach/explicit, test/analogy, or test/explicit groups, and the corresponding expectation manipulation will occur. Participants in the teach groups will be told, “Tomorrow you will teach another participant how to putt,” and participants in the test group will be told, “Tomorrow you will be tested on your putting skills.” Next, the instruction manipulation will occur. Participants in the analogy groups will read the following: “Keep your body still like a grandfather clock and use your arms the same way that the pendulum of the clock operates. (A pendulum is a weight hung from a fixed point so that it can swing freely backward and forward. [See diagram on the right].)” (Vine et al., 2013). Participants in the explicit groups will read the following:

"1. Take your stance with your legs shoulder-width apart.

Set your position so that your head is directly above the ball looking down.

Keep your clubhead square to the ball.

Allow your arms and shoulders to remain loose.

In the putting action, your arms should swing freely backward and forward from your body, which should be still. Make sure that you accelerate through the ball.

After contact, follow through but keep your head still and facing down" (adapted from Vine et al., 2013). (See Appendix C for the instruction sheets participants will read.)

Participants in all groups will have 2 min to read and study the analogy or explicit instructions (Daou, Hutchison et al.). Next, participants will complete the practice phase by performing six blocks of ten putts, taking a 1 min break between each block (participants will sit in a chair during the breaks). When participants stop practicing, they will complete the Intrinsic Motivation Inventory (Ryan, 1982; see Appendix D). Intrinsic Motivation Inventory data may be used for exploratory analyses.

Posttests. Twenty-two to 26 h after completing pretest and practice, participants will return to complete the experiment. Participants will respond to the demographics questionnaire questions about drug/alcohol use, caffeine use, and previous night sleep. Participants in the teach groups will be told, “The participant who you were going to teach did not show up today, so you will actually be tested on your putting instead.” Then, participants will put on the physiological recording device. Next, they will complete low-pressure and high-pressure tests in counterbalanced order. For the low-pressure test, the experimenter will tell participants, “In this set of ten putts, your goal is to make the ball stop as close to the center of the target as possible. Please, try to do the best you can.” For the high-pressure test, the experimenter will tell participants, “In the next set of ten putts, you will be recorded and critically analyzed by a golf expert who will give you a grade.” The experimenter will affix an iPad to the edge of a table, approximately 45° to the right and 225 cm in front of participants. The iPad’s screen will face participants so that they can see themselves being recorded. After the iPad is set-up, the experimenter will tell participants, “The combination of the golf expert grade and your performance during this set will allow you to compete against the rest of the participants for the 1st prize of $50, 2nd prize of $40, 3rd prize of $30, 4th prize of $20, and 5th prize of $10. In summary, you will be putting for money.” As the experimenter explains the rewards, he will take an envelope from a cabinet, pull money from it, and display the potential monetary rewards to participants, after which he will place the money on a 91 cm high countertop, approximately 30° to the left and 100 cm in front of participants. Our pressure manipulation involves two types of pressure revealed to elicit choking in previous studies: performance-contingent outcomes and monitoring by others (e.g., DeCaro, Thomas, Albert, & Beilock, 2011).

After each posttest, participants will complete the Revised Competitive State Anxiety Inventory-2 (Cox, Martens, & Russell, 2003) to determine manipulation efficacy. The Revised Competitive State Anxiety Inventory-2 is frequently used to assess anxiety in motor skill studies (Allsop & Gray, 2014; Elliot, Polman, & Taylor, 2014; Kinrade, Jackson, & Ashford, 2015; Kuan, Morris, Kueh, & Terry, 2018; Mullen, Jones, Oliver, & Hardy, 2016) and possesses good psychometric properties (Cox et al., 2003). The cognitive and somatic anxiety subscales are of interest since the pressure manipulation is intended to modulate anxiety (nonetheless, participants will complete the self-confidence subscale as well) (Jackson, Ashford, & Norsworthy, 2006). The cognitive and somatic anxiety subscale items ask participants to report how much they are currently feeling various indicators of anxiety. All responses will be made by reporting a number between 0 and 100 on a scale with “not at all” corresponding to 0, followed by “somewhat”, then “moderately so”, and finally “very much so”, which corresponds to 100 (See Appendix E.).

After finishing posttests, participants will complete a free recall test to measure declarative knowledge use. Specifically, participants will be asked to report, in as much detail as possible, any rules, methods, or techniques they recall using to putt during the high-pressure posttest (see Appendix F). This type of free recall test is frequently used to assess declarative knowledge in motor skill studies (Daou, Buchanan et al., 2016; Daou, Lohse et al., 2016; Daou et al., 2018; Maxwell, Masters, & Eves, 2000; Maxwell, Masters, Kerr, & Weedon, 2001; Zhu, Poolton, Wilson, Maxwell, & Masters, 2011).

Data processing

Putting. Putts will be recorded with an iPad mounted to the ceiling above the target cross. We plan to measure the ball’s location relative to the target using a custom-developed program written in the National Instruments LabVIEW graphical programming language by Neumann and Thomas (2008). However, we have not been able to verify if this will be feasible in our laboratory because of work restrictions caused by the COVID-19 pandemic. If we are unable to use the LabVIEW program, then we will use Dartfish Live 9.0® motion analysis software. We have compared the output from these two programs when measuring a beanbag’s location relative to a target and observed nearly perfect correlations (e.g., rs ≥ 0.995). Putting accuracy will be indexed by recording radial error as recommended by Hancock, Butler, and Fischman (1995): , where x and y represent the magnitude of error along the respective axes (i.e., how far away from the target cross the ball stops in the horizontal and vertical directions). Precision will be indexed by recording bivariate variable error as recommended by Hancock et al.: , where k = trials in a block and  = centroid along the given axis (x or y) for that block. Radial error and bivariate variable error will be calculated over pretest (10 putts) and may be used as covariates in exploratory analyses. Crucially, we are not using either as an a priori covariate. Radial error and bivariate variable error will be calculated for the first, third, and sixth blocks of the practice phase to get a glimpse into improvement during performance without overly slowing data processing. To assess motor learning and choking under pressure, radial error and bivariate variable error will be calculated for the low- and high-pressure posttests.

Self-reported anxiety. Chronbach’s α will be calculated to determine the reliability of the Competitive State Anxiety Inventory-2 cognitive and somatic anxiety subscales for the low- and high-pressure posttests. If reliability is good (αs ≥ 0.700), then items will be averaged within the subscales. Next, a Pearson’s correlation coefficient will be calculated between the cognitive and somatic anxiety subscales for each posttest, and if r ≥ 0.500, the subscales will be averaged together for each posttest. Otherwise, the subscales will not be combined for statistical analysis.

If the subscales do not exhibit good reliability, then physiological data will serve as the primary measure of anxiety. Specifically, Bioharness data will be extracted and analyzed using Omnisense software (Zephyr Technology, Annapolis, MD). Heart rate will be averaged from the time participants were read test instructions until they completed the test for the low- and high-pressure posttest. Heart rate variability (root mean square of successive differences and high frequency [0.150–0.400 Hz]) will also be assessed for these same periods.

Free recall. Two indices of declarative knowledge use will be extracted from participants’ responses on the free recall test. First, ‘all concepts’ will refer to the number of statements about a concept (rule) (e.g., “I held my left hand over above my right”), ignoring statements irrelevant to technical performance (e.g., “I was told to putt ten times to the target”). Second, hypothesis testing will refer to statements indicating that the participant had tested hypotheses related to their putting stroke (e.g., “I adjusted the swing path of the putter after each missed ball” or “I tried to keep my head still throughout my putting stroke”). That is, hypothesis testing statements will be those that indicate the participant made a prediction about the relationship between their putting movement and putt outcome (Maxwell et al., 2001). We will ignore retrospective statements (e.g., "I held my left hand above my right" or "My feet were shoulder-width apart") that may not have been used or thought about while putting, and we also will ignore statements irrelevant to technical performance.

Movement Specific Reinvestment Scale. Chronbach’s α will be calculated to determine the reliability of the Movement Specific Reinvestment Scale. Daou, Hutchison, et al. (2019) found the Movement Specific Reinvestment Scale had good reliability when all items were considered as one scale rather than dividing the Movement Specific Reinvestment Scale into its movement self-consciousness and conscious motor processing subscales. Further, we do not expect that either subscale should account for more residual variance in our data than the other subscale. Thus, we will assess the reliability across all items and sum them into a single scale if α ≥ 0.700. If the Movement Specific Reinvestment Scale has an α < 0.700, then we will not consider conducting exploratory analyses with it.

Statistical analysis

We will conduct a 2 (Expectation) x 2 (Instruction) ANOVA with pretest radial error serving as the dependent variable. If the η2 ps of the main effects or interaction ≥ 0.0099 (Richardson, 2011), then pretest putting performance will be included as a covariate in all subsequent analyses involving putting performance.

Our primary analysis of interest will be a 2 (Expectation) x 2 (Instruction) x 2 (Posttest) mixed-factor ANOVA with repeated-measures on the last factor, and radial error serving as the dependent variable. This follows because radial error (accuracy) was more sensitive to the Expectation x Posttest interaction observed by Daou, Hutchison, et al. (2019) and reflects the objective of the putting task (accuracy with respect to target). Nonetheless, we will conduct a secondary analysis using the same model with bivariate variable error serving as the dependent variable. We predict an Expectation x Instruction x Posttest interaction, which we will follow up with separate 2 (Expectation) x 2 (Instruction) ANOVAs for the low- and high-pressure posttests. For the low-pressure posttest, we predict a main effect of expectation, such that the teach groups exhibit lower radial error (greater accuracy) than the test groups. For the high-pressure posttest, we predict an Expectation x Instruction interaction. We will follow up this interaction with separate one-tailed t-tests (expectation) for the analogy and explicit groups. For the analogy groups, we predict a significant effect of expectation, such that the teach/analogy group exhibits lower radial error than the test/analogy group. For the explicit groups, we do not predict a significant effect of expectation. Movement Specific Reinvestment Scale score and/or pretest error scores may be used as covariates/between-subjects factors in exploratory analyses of putting data.

To assess practice performance, we will conduct a 2 (Expectation) x 2 (Instruction) x 3 (Block: 1/3/6) mixed-factor ANOVA with repeated-measures on the last factor separately for radial error and bivariate variable error. We predict a main effect of block, such that participants exhibit a linear decrease in radial error and bivariate variable error as a function of block (Daou, Hutchison et al., 2019).

To assess anxiety, we will conduct a 2 (Expectation) x 2 (Instruction) x 2 (Posttest) mixed-factor ANOVA, with repeated-measures on the last factor and the total Revised Competitive State Anxiety Inventory-2 score serving as the dependent variable. If cognitive and somatic anxiety are not strongly correlated (see Self-reported anxiety section), then we will conduct a MANOVA instead, with the Revised Competitive State Anxiety Inventory-2 cognitive and somatic anxiety subscales serving as dependent variables. We predict a main effect of posttest, with higher anxiety occurring on the high-pressure posttest. If heart rate and heart rate variability need to be used to assess anxiety, they will be submitted to a 2 (Expectation) x 2 (Instruction) x 2 (Posttest) MANOVA.

To assess declarative knowledge use during high-pressure posttest, a 2 (Expectation) x 2 (Instruction) MANOVA will be conducted with all concepts and hypothesis testing free recall scores serving as the dependent variables. We predict an Expectation x Instruction interaction for all concepts (Daou, Hutchison et al., 2019). We will then conduct separate one-tailed t-tests (expectation) for the analogy and explicit groups. For the explicit groups, we predict that teach participants will recall using more concepts than test participants. We do not predict an effect for the analogy groups.

The mixed-factor ANOVAs should be robust to violations of homogeneity of variance since we will ensure equal ns. For the practice performance ANOVA, which has three levels of the repeated-measure, we will apply the Greenhouse-Geisser correction if sphericity is violated. Although ANOVAs should be robust to violations of normality, the one-tailed t-tests that follow them may not be (Field, Miles, & Field, 2012). Thus, we will test for violations of normality using the Shapiro-Wilk test and Q-Q plots, both of which we will consider when determining whether the data are non-normal. Since data from similar past research (Daou, Hutchison et al., 2019) suggests a positive skew is possible, a natural log transformation will be applied to data exhibiting a non-normal distribution. To test MANOVA assumptions, a Box test will be used to assess homogeneity of covariance matrices, and a Shapiro test will be used to determine multivariate normality. If these tests are failed, then multiple ANOVAs will be conducted instead of MANOVAs.

We will conduct sensitivity analyses excluding participants who consumed alcohol/drugs within 24 h of the first or second day of the experiment, caffeine within 3 h of the first or second day of the experiment, or report differences in sleep duration >2 h between the night before the first and second day of the experiment. If the statistical significance of results of the primary analysis does not change when excluding these participants, then they will remain in the dataset. If the statistical significance of results of the primary analysis does change when excluding these participants, then they will be removed, and we will recruit additional participants to ensure N = 148 with equal ns per group.

We have received ethics board approval for a similar protocol (Daou, Hutchison et al., 2019) in the past, and thus anticipate an expedited ethics board approval process. We plan to begin data collection in the fall 2020 semester, in which we expect to collect data from 20 participants. We expect to collect data from 40 participants in the spring 2021 semester and 20 participants in the summer 2021 semester. We expect to collect data from 40 participants per semester in the fall 2021 and spring 2022 semesters. We plan to finish data collection in the summer 2022 semester, but our data collection forecast is uncertain given the COVID-19 pandemic. We plan to process data as it is collected, so we expect to conduct statistical analyses and write as well as submit Stage 2 of the manuscript in the fall 2022 semester.

CRediT authorship contribution statement

Daniel A.R. Cabral: Conceptualization, Methodology, Formal analysis, Investigation, Data curation, Writing - original draft, Writing - review & editing, Visualization, Project administration. Marcos Daou: Conceptualization, Methodology, Writing - original draft, Writing - review & editing. Mariane F.B. Bacelar: Investigation, Data curation, Writing - review & editing. Juliana O. Parma: Investigation, Data curation, Writing - review & editing. Matthew W. Miller: Conceptualization, Methodology, Formal analysis, Writing - original draft, Writing - review & editing, Visualization, Supervision, Project administration.

Declaration of competing interest

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.