5.4 Describing Continuous Variables Across Groups
Sometimes we want to describe a continuous variable separately for different groups. For example, we might want to know the average test score for each condition, the average scale score by class level, or the average response time by group.
In jamovi, this is often done using Split by.
When to Use Split By
Use Split by when you have:
- a continuous variable you want to describe
- a categorical variable that defines groups
The continuous variable goes in the Variables box. The categorical grouping variable goes in the Split by box.
For example:
| Continuous variable | Split by variable |
|---|---|
| Exam score | Study condition |
| Depression total score | Clinical group |
| Boredom mean score | Class level |
| Mind wandering score | Gender |
Why Split-By Descriptives Matter
Split-by descriptives are often the first step before group comparisons. Before running an independent t-test or ANOVA, you should understand what the group descriptives look like.
However, descriptive differences are not the same as inferential evidence. If one group has a higher mean than another group, that tells us what happened in the sample. It does not, by itself, tell us whether the difference is statistically meaningful or likely to generalize beyond the sample.
When you split a continuous variable by a categorical variable, you may see that one group has a higher mean than another group. That is useful information, but it does not tell you whether the difference is statistically meaningful or likely to generalize beyond your sample.
For now, describe what you see. Later, inferential tests such as t-tests and ANOVA will help us decide whether group differences are large enough, relative to variability and sample size, to reject a null hypothesis.
Running Split-By Descriptives in jamovi
To describe a continuous variable across groups:
- Go to Analyses.
- Select Exploration.
- Select Descriptives.
- Move the continuous variable into the Variables box.
- Move the categorical grouping variable into the Split by box.
- Select the statistics you want for each group.
Useful statistics often include:
NMissingMeanMedianStd. deviationMinimumMaximum
Interpreting Split-By Descriptives
When interpreting split-by descriptives, ask:
- How many observations are in each group?
- Are any groups very small?
- Which group has the highest mean or median?
- How variable is each group?
- Are the minimum and maximum values plausible?
- Do the descriptive patterns suggest a question for inferential testing later?
Do not write that one group is “significantly higher” than another group based only on descriptive statistics. Statistical significance requires an inferential test.
You want to describe average boredom scores separately for each class level. Which variable goes in the Variables box and which goes in the Split by box?
Answer
The boredom score goes in the Variables box because it is the continuous variable being described. Class level goes in the Split by box because it defines the groups.