Multiple response analysis

Multiple response analysis

Overview edit

Multiple response analysis is a frequency analysis for data which include more than one response per participant, such as to a multiple response survey question.

Examples of survey items which create multiple responses:

  1. "Tick all responses that apply." (multiple dichotomies)
  2. "List the reasons you do physical exercise." (multiple responses)

Such data is well suited to multiple response analysis.

Rather than treat the 1st, 2nd, 3rd etcetera responses as separate variables, multiple response analysis allows the set of responses to be combined and collectively analysed.

Multiple response analysis can be performed using the SPSS Multiple Response Sets function. This is a two step process:

  1. Define multiple response set(s)
  2. Conduct multiple response analysis(es)

This will produce a multiple response analysis showing the frequencies and percentages by cases and responses.

Also consider creating a graph (e.g., bar graph) to illustrate the frequencies or percentages for cases and/or responses.

General steps edit

The three general steps are:

  1. Define a set of two more responses (you cannot do step 2 without doing this step first)
    1. Analyze - Multiple Response - Define Sets
      1. Add the multiple response variables to the "Variables in Set" box. Then click
        1. Dichotomy (if there were only two categories of response) or
        2. Categories (if there were several categories of response) and indicate the category range
      2. Add a name for this new set - Label is optional
      3. Click Add to create the set and then close
  2. Obtain multiple response frequencies (or cross-tabs) of the set you created - this will provide frequencies and percentages of each response option by total number of responses and by cases
    1. Analyze - Multiple Response - Choose either of:
      1. Frequencies: Add the multiple response set into the tables box and click OK
      2. Crosstabs:
        1. Add the multiple response set into either the columns or rows
        2. Add an independent variables of interest (e.g., gender) into the columns or rows (the opposite of the one for the set)
        3. Options - The results will show frequencies, but you can also get percentages click options to also get cell percentages based on columns or rows for cases and responses
  3. Create a graph: It can be useful to graph frequencies or percentages (bar graph). You will need to decide whether to graph responses or cases (or do both). Options for graphing this data:
    1. Use a word processor or spreadsheet - e.g., see Making a Microsoft Word Graph or Graphing with Excel
    2. If graphing responses, you could use SPSS to make a new data file and copy all the responses into a single column (variables), also copy the value labels, then Graphs - Bar Chart

Qualitative data coding edit

If there are multiple open-ended responses, then these should be coded into categorical data, based on themes, prior to conducting multiple response analysis.

FAQ edit

Question: What is the difference between % of responses and % of cases?


  • % of responses indicates what % of the total responses were in each category e.g., 300 out of 1000 (30%) responses may have been about A, 500 (50%) about B and 200 (20%) about C. Note that these %s will sum to 100%.
  • % of cases indicates what % of cases mentioned each category e.g., 250 out of 600 cases (45%) may have mentioned A, 400 may have mentioned B (67%) and 150 (25%) may have mentioned C. Note that these %s will sum to more than 100% if at least one person made more than one response.

External links edit