Main content

Home

Menu

Loading wiki pages...

View
Wiki Version:
**Data Set** We will collect accuracy data for each of three tasks--the change detection task with unfamiliar objects, the change detection task with an explicit strategy provided, and the 6AFC memory test. We will also collect qualitative descriptions of subjects' approach to the explicit-strategy change detection task. We plan to collect useable data from 24 participants. Participants are scheduled and tested in small groups of up to 5 participants. Depending on attendance rates at each experimental session, we may end up with slightly more than 24 participants. Whenever we must exclude a participant due to non-compliance (as listed below), we will replace that participant before viewing their data. After collecting data from a minimum of 24 participants, we will examine the data in order to exclude egregiously low performance (detailed below). If we have usable data from 20 or more participants after exclusions, we will stop data collection. If we have data from fewer than 20 participants, we will schedule additional sessions, again targeting a total of 24 participants. **Exclusion Criteria** Subject data will be excluded from analysis on the following grounds: * Subjects score at or below 20% on the 6AFC memory task * Subjects score at or below chance on more than two tasks * Subjects did not attend to the study phase of the experiment, either by looking away from the screen for an extended period of time other than during the break screen or leaving the testing booth other than during the break screen* * Subjects fell asleep during the experiment* *Any instances of this will be noted in the lab log Noncompliance exclusions will be made without examining subject data. Below-threshold accuracy exclusions will be made by an automated analysis. **Statistical Tests** We will conduct paired-samples t-tests (two-tailed), effect sizes (Cohen's d) with 95% bootstrapped confidence intervals, and calculate Bayes factors (two-tailed, medium spread) for the following comparisons: * Accuracy in the uninstructed, unstudied change detection task versus accuracy in the explicit-strategy change detection task * Accuracy in the explicit-strategy change detection task versus accuracy in the 6AFC memory test **Data Visualization** We will create the following plots: * Violin plots of individual subject accuracy on each of the three tasks * Violin plots of the difference scores for the comparisons listed above * Point plots of average accuracy on each of the three tasks, with error bars representing 95% bootstrapped confidence intervals **Exploratory Analysis** We will construct two models of subject performance. The change-detection model predicts accuracy on the explicit-strategy change detection task based on subject performance on the uninformed change-detection task, and predicts identical performance in both conditions. The combination model predicts performance based on a combination of uninformed change detection performance and 6AFC memory performance (accuracy is predicted by adding the number of items identified through change detection performance to the number of remaining items identified through 6AFC performance). We will find the distributions of the differences between actual explicit-strategy change detection performance and each of these models, and from these obtain a distribution of the differences between deviations (deviation from change detection model - deviation from combined model). We will examine this distribution for bimodality and skewness. If there are two groups in the data, each one with behavior well-defined by one of the two models, we expect a bimodal distribution, with a mode to the left of zero (indicating better predictions by the change detection model) and a mode to the right of zero (indicating better predictions by the combined model). If, however, one model better describes all or most subjects, we expect the distribution to be skewed, with a negative mean if the change detection model is the better predictor, and a positive mean if the combined model is a better predictor. We will also examine the qualitative responses taken from subjects after the explicit-strategy change detection task for patterns, common strategies, and subject perception of task difficulty.
OSF does not support the use of Internet Explorer. For optimal performance, please switch to another browser.
Accept
This website relies on cookies to help provide a better user experience. By clicking Accept or continuing to use the site, you agree. For more information, see our Privacy Policy and information on cookie use.
Accept
×

Start managing your projects on the OSF today.

Free and easy to use, the Open Science Framework supports the entire research lifecycle: planning, execution, reporting, archiving, and discovery.