Skip to content

Single case analyses

krajcsi edited this page Oct 7, 2019 · 1 revision

CogStat can handle comparisons when a single case (e.g., a patient) is compared to a group (e.g., a control group). The single case hypothesis tests are based on the solution proposed by John R. Crawford and his colleagues.

At the moment, CogStat supports these hypothesis tests:

  • Compare the performance of a single case to a control group (i.e., extremity of a task). Find more details about this modified t-test in Crawford & Howell (1998).
  • Compare the performance of a single case to a control group when the performance is measured as a slope (i.e., extremity of a task expressed as a slope). Find more details about this slope comparison procedure in Crawford & Garthwaite (2004).

To perform other single-case hypothesis tests not supported by CogStat yet, you might use the single-case study statistic packages by John R Crawford.

How to run a single-case analysis in CogStat?

Preparing the data. Load your data, where a grouping variable will tell whether a case is a single case or a member of the control group, and where there is a dependent variable. In a slope comparison analysis you also need a slope standard error variable.

For a single-case analysis, your data should look like this:

Group Dependent variable
nom int
Single case 3
Control group 3
Control group 4
Control group 2

Or if you compare slope data, your data should look like this:

Group Slope Slope SE
nom int int
Single case 3 0.2
Control group 3 0.3
Control group 4 0.4
Control group 2 0.3

Performing the analysis. Then choose Analysis > Compare groups, and set the appropriate grouping variable and dependent variable. To run a slope analysis, click on the Single case slope... button, and set the Slope SE variable, and set the number of trials per participant.

If the grouping variable includes two groups and one of the groups includes a single case (such as in the data sources above), the single-case hypothesis tests will be chosen automatically.

Note that the modified t-test will be run only if the control group is normally distributed.

Documentation for users

Introduction

Using CogStat

Advanced topics

Get help

Documentation for contributors

Statistical and methodological documentation

Documentation for developers

Clone this wiki locally
You can’t perform that action at this time.