How do I interpret the Respondent-Centric Analysis (RCA) in Card Sorting?
In this help, we'll discuss how to interpret the Respondent-Centric Analysis (RCA) in open and hybrid card sort analysis.
Read further to learn about:
- What is RCA and how it's used
- Adjusting RCA parameters
- The meaning of high and low support
- PDF Export of RCA
What is RCA and how it's used
The RCA (a.k.a. Respondent-Centric Analysis) seeks to answer the question "Which respondents best represent popular ways of thinking?". As the name implies, RCA revolves completely around respondents and their answers:
- Answers from all respondents are tallied as votes towards answers from other respondents that they agreed with.
- Completed card sorts with the hightest number of votes are selected as the result of RCA. As such, the results are actual answers from actual people.
How this works is we take all pairings of cards that were placed together into the same category. If two respondents have a sufficient percentage (50% and higher by default) of shared pairings, we count their answers as 'voting' in favor of each other. From completed card sorts ordered by the number of votes, we select the ones with the highest number of votes, which haven't already supported an answer with higher number of votes with their vote. Doing this helps us pick popular answers that are sufficiently distinct from each other.
The popularity of an RCA result is expressed by the support score - the number of votes that the result received. For each RCA result, you see which categories the respondent created and which cards they placed inside them. Each category also provides a list of alternative suggested names, based on names of similar categories from supporting answers.
You can further explore categories and the level of agreement between voting answers by clicking Details. You will see a similarity matrix, with rows and columns for all cards found in the category. The percentages in the matrix say how many of the respondents who supported this RCA result have also placed these cards in the same category.
Adjusting RCA parameters
RCA in UXtweak Card Sorting has three parameters that modify how the RCA results are calculated. This allows for flexible analysis an exploration of the similarity between emergent concepts.
- Minimum similarity - What percentage of pairs shared between answers is required for the answers to vote for each other. The defautl value is 50% (or higher, if the increased value doesn't cause any of the results to lose supporters)
- Number of categories - Limit results to answers that contain a number of categories within specified range
- Number of results - How many RCA results are returned
Try to increase minimum similarity to create RCA results where voters achieved a higher level of agreement. If you find few or none results that are supported by others on current settings, try to decrease the minimum similarity to find answers that many others supported at least partially. Investigate how the results and support scores change depending on minimum similarity. Use what you learn to find results that are well descriptive of their groups of supporters.
Naturally, RCA can only return as many results as there are answers that don't support each other. If all answers agree on the minimum percentage of pairings, you will get only one RCA result, no matter the number of results you ask for.
The meaning of high and low support
A low similarity score means that the RCA result represents only a small number of other respondent's answers. This can happen if most respondents came up with different ways to sort the cards.
To solve this, you can try to decrease the minimum similarity threshold. Perhaps you're asking that the answers be too similar, which can be a problem with more complex card sorting. You can also try to recruit more respondents to give concepts more opportunity to emerge, and/or use the Similarity matrix and Dendrograms to look for patterns in a different way.
PDF Export of RCA
You can create a PDF report of all the information listed above, exactly like you can see it in the web interface. To create an export, click PDF Export in the upper right corner of the tab contents.
If you wish to create a report using more than one section of your study's results, go to the Export tab