Causal Cognitive Mapping as innovative interview method?

Causal Cognitive Mapping [1] is, on a first look, a way to draw maps (a bunch of text statements connected with arrows; see also figure 1 left). But that is like saying the portrait is the art. What about the artist and the person portrayed? Often, cognitive maps are developed within an interview. Therefore, you may well consider Causal Cognitive Mapping as an interview method - but not a general one. Each map is structured so that, as a whole, bottom assertions and options lead over actions or strategies to top goals or values (figure 1 right and left).

Figure 1 How a map looks like (zoomed out)


 

Give it a shot or not?

It might not make much sense to map your 10 answers to your Likert questions with Causal Cognitive Mapping. You may only really be interested in a particular assertion or action of your interviewee. For instance: "What do you do if it rains?" "Open umbrella." You can now build classes of what people do when it rains, or so. If that is all you want to know, done.
More broad and open opinion interviews and the like may be more compatible. With Causal Cognitive Mapping you can explore individual assertions and how they lead to goals or values.

Figure 2 Example causal chain

If you were interested in this full causal chain in figure 2, then Causal Cognitive Mapping is worth a shot. Assume again you have figured out that your interviewee opens her umbrella when it rains. Why does she do that? What if it does not rain? What if she forgets the umbrella? You see, with a Causal Cognitive Mapping interview, you systematically would build an increasingly rich map (=interview). Therefore, whether Causal Cognitive Mapping may be useful as an interview method depends on what you are interested in when doing your interviews.


An actual interview

Let us have a look at an actual interview. In figure 3, which is based on a job interview, you see labels that represent sets (categories).They were named after interviewer questions. "Why did you apply?" became "Why" for instance. Each set contains all the statements related to the associated question. So the "Why" label, for example, is the set that contains all statements of the interviewee's answer to the why-question.

Figure 3 Interview questions/answers as set matrix


Figure 3 shows what set (left side) links to what other sets (top side). Differently put, it shows what answer is linked to other answers. You see for instance that the interviewee backed her case up by linking her work experience with why she applied; four times in total. But you also see empty space.





Actual use cases

Okay, but where can this be useful? Two scenarios come to mind. Above I already implied the case when you want to build richness. This relates to qualitative interviews [3] that are less strictly, or even not at all, tied to a predetermined interview structure. Differently put, in those interviews you may make up questions on the go. That is usually done when you want to get every bit of data out of each interviewee (e.g. case study research [4]). You can use emerging empty spaces as an indicator for where you need to ask more, explore hot spots, etc.

In (rather) quantitative interviews you usually do not want richness. Imagine you pose a question to 50 interviewees. If you phrase it loosely and keep it general, even encourage interviewees to keep talking, you may well get 50 distinct answers without anything in common. That would not be helpful if you wanted to generalise which quantitative interviews often seek to do.

But that does not necessarily mean Causal Cognitive Mapping is useless for quantitative interviews. You usually want answers to be isolated from one another. That is because there is an effect according to which interviewees seek to appear consistent with what they have said before [2]. They subconsciously say something because of what they have said previously which they would not have said otherwise. This effect may light up in the set matrix as coloured space. More generally speaking an entirely empty set matrix would be the ideal outcome because then no interference between answers would have occurred.


Some caveats

As always there are some drawbacks to be considered.

Remember that in figure 2, sets on the left side link into sets on top. This can become a little confusing because this relationship is not a chronological one. It does not mean "left" was asked before "top". What counts is whether statement A is causally [8] connected to statement B - irrespective of which one of them was uttered before the other. Anyway, if you were interested in a chronological order, you could list the sets in the same rank order as the question order. If the left side set has a lower rank than the top side set, "left" was asked before "top" - and vice versa. Figure 4 clarifies.

Figure 4 Given that sets are rank ordered by question order, if question 5 links to question 3, 3 was asked before 5. Therefore, a chronological arrow would be inversed.

Furthermore, finding labels for interview questions is not always easy. Some qualitative interviews are more like actual conversations rather than distinct question-answer blocks (e.g. [5]). This is one reason why interviewers often seek certain themes [6] or topics (that cross answers) in their interviews. Discovering such themes can take quite some time though. Keep that in mind.
Finally and particularly for qualitative interviews, a trade-off might be that if you map while interviewing there is a risk that you interrupt the conversation flow - and eventually risk annoying your interviewee. So, just map the transcripts then? But then you will have to schedule follow-up interviews if interesting features show up on the map.


Summary

So far I have not read interview guides that suggest graphical mapping of interviews. With Causal Cognitive Mapping you have a straightforward method readily available to be applied to your next interview.
I guess the best case can be made for qualitative interviews where you are after richness. Quantitative interviews? Maybe. Obviously for very short interviews you will not get much out of mapping. However, I suggested mapping for showing that answers do not influence each other. For this case it depends on how valuable graphical maps of your interviews are to you. Is creating them worth the extra effort - in particular if you conduct a large amount of interviews?


But wait...

So you decided to give it a shot and feel a bit lost where to start? Well, obviously computer support is highly recommended because managing sets can be tedious, if not impossible, depending on map size. The first stop is probably the software "Decision Explorer". It gets you through the mapping and the set management up until printing the sets as a number list on screen.
Another very promising programme comes from the University of Eastern Finland called "CMAP3". I have not had the chance to try it yet but it is probably the most "scientific" version of Causal Cognitive Mapping software. You can get CMAP3 for free (http://www.uef.fi/web/cmap3) and read about use cases in [7]. I am not sure about its capability of managing sets. Please let me know if you tried it.
Finally, if you want to get something like in figure 3, I need to selfishly point you to http://jo-richter.de/ccmapp/index.php/getting-started. If you do not mind becoming a guinea pig, the app is for free right now...

(C) CC BY (https://creativecommons.org/licenses/by/4.0/), Jo. Richter, http://causal-cognitive-mapping.blogspot.com/2017/01/causal-cognitive-mapping-as-innovative.html


[1] http://www.springer.com/us/book/9781848828087
[2] Schuman, H. and Presser, S. (1981), Questions and Answers in Attitude Surveys - Experiments on Question Form, Wording, and Content, Academic Press, New York.
[3] DiCicco‐Bloom, B. and Crabtree, B.F. (2006), "The qualitative research interview", Medical Education, Vol. 40 No. 4, pp. 314–321.
[4] Taylor, R. and Thomas-Gregory, A. (2015), "Case study research", Nursing Standard, Vol. 29 No. 41, pp. 36–40.
[5] Spradley, J.P. (1979), The Ethnographic Interview, Holt, Rinehart and Winston, USA.
[6] Turner, D. W. (2010). Qualitative Interview Design: A Practical Guide for Novice Investigators. The Qualitative Report, 15(3), 754-760. Retrieved from http://nsuworks.nova.edu/tqr/vol15/iss3/19 (http://creativecommons.org/licenses/by-nc-sa/4.0/).
[7] Mauri Laukkanen and Päivi Eriksson. (2013), "New designs and software for cognitive causal mapping", Qualitative Research in Organizations and Management: An International Journal, Vol. 8 No. 2, pp. 122–147
[8] I will discuss in a later post whether statements in Causal Cognitive Mapping must be connected causally.