This Wiki page contains additional explanations and details of the contents provided in the project files. @[toc](Table of Contents) ### Preregistration ### As noted in the article, our investigation (in Experiment 2) of the effect of instructional enthusiasm on recall accuracy was preregistered on OSF prior to human observation of the data. The preregistration form can be found [here][2]. ### Videos ### In this study, detailed task instructions were presented as short video clips, either enthusiastic or neutral. These clips have been merged into combined video files: - Experiment 1: - [Enthusiastic][3] - [Neutral][4] - [Scripts][5] - Experiment 2: - [Enthusiastic][6] - [Neutral][7] - [Scripts][8] Each individual clip is also available in OSF Storage / Experiment [1 or 2] / Experiment Script / videos. (See below for more information about the Experiment Scripts) ### Word Lists ### English / Indonesian words pairs used in the current study were taken from a larger set used in [Kornell & Son (2009)][9]. During the study phase, participants would see an Indonesian word, and clicked the card to reveal the English translation. During test, participants were shown the Indonesian word and had to type the translation. - [Experiment 1 Word Pairs][10] - [Experiment 2 Word Pairs][11] ### Experiment Scripts ### The experiment scripts were written with [jsPsych v5.0.3](https://github.com/jspsych/jsPsych/tree/v5.0.3). Documentation for jsPsych is available at [docs.jspsych.org](http://docs.jspsych.org). The scripts can be run locally by downloading the folder containing the experiment code (click on the "Experiment Scripts" folder for either Experiment and then click *Download as zip*) and opening the index.html file. The jsPsych-based script for the study phase is contained inside the index.html file, and the test.html file contains the script for the test phase. Once downloaded, you can view the script by opening index.html with a text editor, or by opening the file in your browser of choice and selecting *View Source*. The other files in the folder are supporting code (the jsPsych library, and scripts for communicating with the server for data storage). Additionally, the *videos* folder contains the individual video clips that are called during the study. There will be a few inconsequential errors when the script attempts to save data to the server near the end of the experiment, but this should not prevent the experiment from running. ### Raw Data ### Data are recorded in long format, where every learning and test trial (as well as browser interactions, etc.) is a row. The raw data, obtained directly from the jsPsych experiment script is in the "Data and Analysis" folder, in the *Raw* subfolder. ParticipantLog.xlsx contains our notes about individual participant sessions. Day1.csv contains raw data from the study phase, and Day2.csv is the test phase. Importantly, these raw data include errors as well as participants who have been excluded. Running preProcessing.R (in the "Data and Analysis" folder) cleans these raw data, and produces the analysis files in the *Generated* folder: day1_preprocessed.csv and day2_preprocessed.csv. The columns in the day1_preprocessed.csv and day2_preprocessed.csv files are as follows: | Column name | Contents | Description | | ----------- | -------- | ----------------- | | subject_id | integer | unique subject identifier | | memory_condition | either "massed" or "spaced" | randomly-assigned sequence condition | | enthusiasm_condition | either "enthused" or "neutral" | randomly-assigned enthusiasm condition | | trial_index | integer | index of steps in the experiment routine | | trial_type | character string | indicates the type of trial | | test_question (day2 only) | binary (1 or NULL) | indicates whether the trial was a test question | | rt | integer | response time for button presses | | stimulus | html or file path | material presented on that step | | key_press | integer | keyboard code of response button | | flips (day1 only) | JSON | response time attribute of that flip trial, time spent on front of study card | | n_flips (day1 only) | 1 | [unused, but indicates that it is a flip trial] | | total_time (day1 only) | integer | response time attribute extracted from "flips" JSON | | total_time_back (day1 only) | integer | time spent on back of study card | | front (day1 only) | character string | the Indonesian element of the word pair | | back (day1 only) | character string | the English element of the word pair | | word_appearance_count (day1 only) | integer | the number of previous presentations of that word pair | | responses | JSON | responses to questions (see *demographics_and_metacognition.csv* in *Generated* subfolder for formatted survey data), raw responses to test prompts in day2. See *QuestionDataKey.rtf* for exact question wording. | | english (day2 only) | character string | correct English element of the word pair | | indonesian (day2 only) | character string | the Indonesian element of the word pair | | correct (day2 only) | binary (1, 0, or NULL) | correctness (1 is correct) of the response, compared with the correct English translation | | button_pressed (day2 only) | binary (0 or NULL) | [unused] | | view_history | JSON | [unused] | | time_elapsed | integer | [unused] | | internal_node_id | character string | [unused] | | sources | NULL | [unused] | | ts | datetime | [unused, but indicates the timestamp of the trial] | ### Analysis Scripts ### For both Experiment 1 and Experiment 2, descriptive analyses are in the R script: analysis.R. The summary statistics produced by this script should match those reported in the published article. This script also contains additional exploratory analyses that are not reported in the article. For each dependent variable (accuracy during test, number of responses during test, or questionnaire ratings), Bayesian analyses are performed in two separate files: First, the posterior distribution is produced in runJAGSmodel_[dv].R (where [dv] is either *accuracy*, *attempts*, or *ratings*), and second, the posterior distribution is investigated in analyzeJAGSmodel_[dv].R. The model for accuracy and attempts is defined (in JAGS format) in the text file p-correct-two-factors.txt (which matches [the analysis model that was preregistered][12]). There is no analyzeJAGSmodel for analyses of ratings -- these analyses are included in the runJAGSmodel_ratings.R script. These scripts should be in the same folder as DBDA2E-utilities.R and BEST.R (c/o John Kruschke). Also, in order to run these scripts, the user needs to install JAGS, as well as the runjags package for R. [1]: http://dx.doi.org/10.1371/journal.pone.0181775 [2]: https://osf.io/qmg95/register/565fb3678c5e4a66b5582f67 [3]: https://osf.io/teptz/ [4]: https://osf.io/ks83r/ [5]: https://osf.io/8xad7/ [6]: https://osf.io/ams9n/ [7]: https://osf.io/2pmxd/ [8]: https://osf.io/2nyr7/ [9]: http://dx.doi.org/10.1080/09658210902832915 [10]: https://osf.io/nbj6y/ [11]: https://osf.io/srgqt/ [12]: https://osf.io/pcnmq/