Survey research and design in psychology/Tutorials/Qualitative analysis/Multiple response analysis

From Wikiversity
Jump to navigation Jump to search

Multiple response analysis[edit]

Gnome-settings-background.svg View the accompanying screencast: [1]


In this content analysis approach, qualitative data is recoded as numeric numeric data which indicate categories of themes which are then summarised statistically and graphically:

  1. Immerse yourself in the data and create a coding sheet:
    1. Read over the responses
    2. Make a list of the main themes and allocate a number to each e.g.,
      • 1 = Parking,
      • 2 = Teaching quality,
      • 3 = Computing facilities etc.
  2. Consider whether to group some of the themes into a broader category - if in doubt, keep as separate themes because there will still be a chance to recode them if necessary later on (e.g., after looking at the results)
  3. Themes/Coding:
    1. Insert new variables in your data file - one for each of the multiple response options (e.g. participants could list up to two areas of satisfaction, so add two variables e.g., satis-response1 and satis-response2).
    2. Enter category codes for each case's comments. If they made multiple comments, then code one response per column, up to the max number of possible responses.
    3. If there was less than the maximum number of responses, then leave those cells as missing data (no code.
    4. You may also need a code for "Other": In doing your coding, you may find responses that don't seem to fit into the main themes you identified. These could be coded as an extra category called something like "Other"
  4. Inter-rater reliability:
    1. It could be a good idea to do the coding in conjunction with someone else
    2. Ideally, create the coding sheet together, then code the data independently, then go over the codes for each response together and discuss any codings you disagreed on. This is the basic way of inter-rater reliability.
  5. Multiple response analysis
    1. Its a good idea to add value labels (it will make the output easier to interpet).
    2. Perform multiple response analysis (download some of the examples from this page about how to do this).
      1. Note that you can also analyse the multiple response frequencies by an independent variable (e.g., gender) using Multiple Response - Crosstabs.
    3. Examine the frequency and percentage of responses and the percentage of cases - present these summary statistics as a table in your results and possibly also as a bar graph
  6. In writing up the results, don't lose the richness of the qualitative data; try to provide a rich description of the themes to accompany the statistical summaries.