How do numbers and statements fit together? (comparing cases, mixed methods)

In brief

The case comparison table puts each case in one row: the numbers and groups on the left, the statements next to them, for example an interview quote. The numbers give a first suggestion for groups. Reading the texts, re-sorting cases and interpreting the groups is part of the method. Statable only describes here; it tests nothing and calculates no p.

What it is for

There are numbers (such as a grade or a questionnaire score) and texts (interview, open answer) for the same people. The question is how the two fit together: do the statements confirm the numerical profile, or do they explain why people with the same profile differ (Schoonenboom and Johnson, 2021)? The steps:

1. One row per case, numbers and texts side by side.
2. First groups from the numerical values (Statable splits number columns at the median and group columns by their levels).
3. Read the quotes: do they fit the group? Do they explain differences within the same profile? Mark the important passages.
4. Interpret and write a tentative hypothesis in the note of the group.
5. Move cases, rename or create groups and read again.

Genuinely qualitative data (interview text) differ from data converted into numbers (coded, counted): the table brings both together but does not replace an analysis of the texts (Schoonenboom, 2023). Transcribing and coding long texts stays with MAXQDA, ATLAS.ti, QCAmap or QualCoder.

Requirements

• One row per case. You need 1 to 4 columns with numbers or groups and 1 to 3 text columns.
• A case ID column is optional but advisable: with it, groups, notes and marks are kept when the rows change. Without an ID the row number of the data counts.
• Comparing cases is meant for a manageable number of cases: with a note from 500 cases, not possible above 3000. Texts up to 32,767 characters are shown in full.
• Paragraphs in long answers become spaces when the data are read in.
• In MAXQDA, document variables and segments can be exported as an Excel table and opened in Statable.

Reading the result

• The table shows each case with its group, the numbers and the full texts. Clicking a heading sorts (display only). In the "Group" column you choose where a case belongs; "Undo" takes back the last change.
• The suggestion splits a number column at the median: "up to" includes the value itself. If several cases lie exactly at the median, the groups differ in size; Statable states both sizes. Under "More options" you can set your own threshold. Cases with a missing entry form groups of their own, "no answer for …".
• For each group the table shows n, mean, standard deviation (with n − 1; "-" for one case), median and smallest to largest value, for group columns the frequencies with percent of the cases with an entry. They update with every move.
• Quotes are marked: the whole answer with a button, part of a sentence by selecting the text. The note of each group records the interpretation.

Reporting the result

The report contains a joint display (Guetterman, Fetters and Creswell, 2015): one row per group with the number of cases, summary figures, the marked quotes and the note. Word, PDF and Excel contain it; Excel also contains the full table of all cases.

The 16 cases were divided into groups at the median of practicum grade and self-efficacy; the groups were read against the interview statements, adjusted and interpreted (Schoonenboom and Johnson, 2021).

Limits

• The table describes these cases. It tests nothing and cannot be generalized to other people; there is no p, no confidence interval and no effect size.
• The groups are the researchers' interpretation, not a finding. Splitting at the median is a starting point: another threshold gives other groups.
• Statable does not form groups with artificial intelligence and does not code texts; the reading is left to the researchers.

Source

• Schoonenboom, J. and Johnson, R. B. (2021): case comparison table, in: The Routledge Reviewer's Guide to Mixed Methods Analysis.
• Schoonenboom, J. (2023): Forum Qualitative Sozialforschung, 24(1).
• Guetterman, T. C., Fetters, M. D. and Creswell, J. W. (2015): Integrating quantitative and qualitative results in health science mixed methods research through joint displays. Annals of Family Medicine, 13(6), 554-561.