How to analyse a survey with fewer than 30 respondents
Source:vignettes/articles/guides/analyse-a-survey-with-fewer-than-30-respondents.Rmd
analyse-a-survey-with-fewer-than-30-respondents.RmdPlan and review an analysis for a survey with a small sample, such as a pilot.
The example in this guide is A pilot with 18 attendees. A pilot study with 18 participants, answering the main study’s questions with methods suited to a small sample.
In SurveyStudio
- In RStudio, open the Addins menu, choose
surveyframe: Try a demo and pick
small_sample. - On Design, choose Plan. The example plan uses methods often considered for small samples: a rank test, an exact test and logistic regression with Firth’s correction.
- Choose methods that fit your own design, measurements and assumptions, and record the reason in each research question.
- On Analyse, run the plan and report the intervals beside each estimate.

The Design screen with the A pilot with 18
attendees study open.
In R
d <- sframe_demo("small_sample")
results <- run_analysis_plan(d$responses, d$instrument)The research questions and what the report gives
| Research question | Method | What the report gives |
|---|---|---|
| Do the 2 formats differ, without assuming normality? | Mann-Whitney U test | The APA 7 sentence and a table, with a chart where the method has one |
| Is format associated with returning, on an exact test? | Fisher’s exact test | The APA 7 sentence and a table, with a chart where the method has one |
| Does attending more sessions predict returning, with a small-sample correction? | Firth logistic regression | The APA 7 sentence and a table, with a chart where the method has one |
Run the example in SurveyStudio or R to see the results, each marked as demonstration data.