Introduction
In the Codeframe Panel in codeit, the frequency statistics for each theme can be calculated on a Response or Respondent basis. This page explains the difference between the two methods.
Multi Response Data
The distinction between Response and Respondent level statistics only matters when coding multiple response data.
This is data where a respondent can provide multiple verbatims for a single question.
For example, a looped question or a multiple response question (e.g. a brand mention question with multiple fields for answers).
If you are coding simple single response data (where each respondent provides only one verbatim) then there is no difference between the Response and Respondent level statistics.
Example Data
As an example of multi response data, consider the following example:
| Respondent | Response | Verbatim | Code Applied |
|---|---|---|---|
| 1 | 1 | Coca Cola | 1 - Coke |
| 1 | 2 | Diet Coke | 1 - Coke |
| 1 | 3 | Fanta | 2 - Fanta |
| 1 | 4 | Dr Pepper | <Uncoded> |
Response Based Calculations
When calculating on a response basis, counts are calculated at a verbatim level.
For a given theme, the count is the number of verbatims that have that theme applied.
In the example data above, the count for "1 - Coke" would be 2, because two verbatims have that theme applied.
The frequency percentage is the count divided by the total number of coded verbatims (i.e. verbatims with any themes applied). In the example data above, the percentage for "1 - Coke" would be 67% because 2 verbatims have that theme applied, out of a total of 3 coded verbatims.
Respondent Based Calculations
When calculating on a Respondent basis, counts are calculated at a respondent level.
For a given theme, the count is the number of respondents that have that theme applied.
In the example data above, the count for "1 - Coke" would be 1, because one respondent has been coded with that theme, regardless of the fact that the theme has been applied twice.
The frequency percentage is the count divided by the total number of coded respondents (i.e. respondents with any themes applied). In the example data above, the percentage for "1 - Coke" would be 100% because 1 respondent has that theme applied, out of a total of 1 coded respondents.