"Slicing and Dicing" the Data

Subgroup analysis can be thought of as “slicing and dicing” the data according to key characteristics among your target population that may affect the outcomes youth experience as well as varying levels of participation in the program (i.e., dosage and duration).
Subgroup analysis is important because overall group averages and results often do not tell the whole story. For example, you may find that although the overall mean (or average) for all youth suggests increased knowledge of recycling, but analysis of results for girls when compared to boys reveals that girls experienced a large increase while boys experienced a slight decrease. These are actionable results! Looking at results forsubgroups helps you focus on program improvement efforts that are most likely to yield better results.
Potential groups to include in subgroup analysis

  • Demographic characteristics (e.g., gender, age, race/ethnicity, education level, socioeconomic status, etc.)
  • Other target population criteria (risk factors, readiness factors, and other baseline conditions)
  • Program dosage (e.g., number of months or years in program, number of classes received, etc.)
  • Groups of participants who did exhibit change compared to those who did not

Subgroup Analysis: Sample Data

In the example below, we look at responses by program session for the question, “Since participating in the program, I am more interested in helping improve my local environment.” You can create a similar table for any other subgroup of interest (such as target population characteristics).

As you can see in the highlighted column, the weighted average is lower for the school-year session than it is for the summer session (2.2 vs. 3.0). In fact, nearly one-third of participants who participated in the school-year session marked “not at all.”

This finding, which you would uncover only through subgroup analysis, can prompt you to dig deeper with staff and youth to explore potential reasons why participants in the school-year and summer sessions might be experiencing outcomes differently.
Sample Subgroup Analysis (by Program Session):
“Since participating in the program, I am more interested in helping improve my local environment.”

Learn More about subgroup analysis

Below is a list of resources cross-tabulating your data to look at results by sub-groups in a relatively straightforward fashion.

BetterEvaluation also has several resources about data analysis on their website. See their “Analyse Data” page here: http://betterevaluation.org/plan/describe/look_for_patterns.
Potential resources for subgroup analysis

  • Excel Pivot Tables: You can create a pivot table in Excel to “slice” one question’s results by another question’s results. For example, you could look at level of satisfaction with the program by race/ethnicity.
  • If you don’t know how to use pivot tables or have bandwidth to take an online tutorial, you can also use “filters” in Excel columns to create separate datasets for each sub-group.
  • SurveyMonkey: When running SurveyMonkey reports in the paid version of SurveyMonkey, you can run cross tab reports to show the relationship between two or more survey questions, or you can “filter” your results by different criteria (such as answers to a specific survey question like grade level).
  • StatWing: StatWing is an online platform that lets you analyze data quickly, simply by clicking on the variables you want to compare. StatWing can also run tests for statistical significance.

Photo credit: Seven Teepees