Meet with a NCA author - Maciej Karwowski

NCA has papers published in various fields. Dr. Maciej Karwowski wrote two papers in Psychology using the NCA method. One of them, which shows to what extent Intelligence is a necessary condition for creativity, is among the most cited papers of the Intelligence journal. The NCA team interviewed him and got valuable insights about his experience with the method.

An interview in which Maciej explained that some of his peers called NCA a possible “game-changer” in psychology.

Find the interview below:

[NCA member]
[Maciej Karwowski]

Where did you find NCA/ how did you know existence of NCA?
This is actually Jan who contacted me around 2015 while I was studying so-called threshold hypothesis. In the psychology of creativity, the threshold hypothesis assumes a curvilinear relationship between intelligence and creativity. To state it differently, it predicts that intelligence matters for creativity, but its role differs at different levels of the continuum of intelligence.  While low intelligence usually makes creativity hardly possible, high intelligence does not guarantee it. Dr. Dul was the first who realized that this is precisely a Necessary Condition idea. Hence, we explored several independent datasets to examine if an NCA-like pattern occurs and compared the new analytical approach with more classical data analysis methods. The results were very consistent and showed a clear NCA pattern.

What were your personal considerations to use NCA?
My first impression was that what NCA proposes is quite straightforward and very elegant. It can be very easily explained and understood. That was my first reaction. In a sense, NCA changes the status quo as it is not based on averages/means as typical correlation/regression analyses are. It was simple and refreshingy new. 

What were the main arguments in favor of using NCAhen discussing NCA in your team of co-authors/peers ?
All thought that NCA was an excellent supplementary method. Both methods (NCA and regression-based) were used in my team, and the results were compared. It was, therefore, easy to use as it was not about skipping the classical methods. 

What were the main arguments against using NCA in your team?
- At the very beginning, we were especially interested if the results obtained are trustworthy: reliable and robust. In psychology, there is always a measurement error, which is especially problematic at very high and very low levels of the distribution of traits being measured. It might have consequences for the NCA estimates. Of course, that is not a problem of NCA per se: NCA assumes that the measurement is reliable, but there is no perfect measurement. 
- We also thought that NCA was rather descriptive and not explanatory at the very beginning. For instance, there were some effect sizes estimated, but no p-value reported. Now the permutation-based significance test is available, so the situation changed.

What were the challenges during the data analysis with NCA (e.g. software, information about NCA)?
Not really, it was very easy from the beginning. The package is very straightforward. I never had to use the web calculator. It was really simple and straightforward.

What were the challenges during interpretation of the results?
When we used NCA for the first time, there was yet no significance test. So, we had to decide, somehow, whether or not there is a necessary condition in our data. What was available are some proposed thresholds of effect size (small, medium, and large) that we used. We recognized this is somehow arbitrary. 

What were the challenges regarding NCA during writing the paper?
The process was a gratifying endeavor, especially for our first paper published in a journal Intelligence. It was precisely NCA research problem from the very beginning. It is ironic that no one looked up for this approach before, as in 1967, Joy Paul Guilford, one of the greatest psychologists who studied both intelligence and creativity, published a scatter plot that visualized the relationship between intelligence and creativity. This scatter plot was exactly like NCA illustration in contemporary papers. So, it was very satisfying as it was as if for fifty years, we had used imperfect statistical tools, and NCA could be much more appropriate. 
That is the beauty of simplicity. Sometimes we tend to complicate things too much, to use highly complex models, and that is fine if the question requires it. 
NCA provides some breadth; it is simple but, at the same time, robust. It was a pleasure to write both papers. 

What were the challenges regarding NCA during the review process (editor, reviewers)?
The process was very rigorous and we did our best to explain well what NCA is. In this paper, there was a whole section about it. The reviewers understood it and accepted it. 
We didn't only use NCA. I believe if we have used only NCA, it would probably have been harder to publish it, as NCA was still quite unknown. What we did is that we explained NCA and compared it to more established methods, like segmented regression. Both approaches showed similar results, which is not always the case. 
People intuitively get what NCA is. I see this as an insight when people get this "Wow" or "Aha" moment. And this is important because statistical methods are usually requiring a lot from the audience. But here, even without in-depth mathematical training, you can at least understand the basics. Of course, there are some things you might not get as quickly, but the basic idea you can understand quite easily, and I think it is a great asset.

What were the reactions of the readers of your paper on NCA?
Most of them were very positive. I received comments of people in my field saying that NCA will be a "game-changer" in our thinking of the processes. 

What do you consider the main advantages of NCA?
NCA is something you can quite nicely explain. It is not only robust, but it is also quite communicative.

Do you intend to use NCA in your future work?  Please explain.
Yes, of course. There are some ongoing projects with NCA being our primary statistical analysis, yet this is too early to talk about details. 

References: Karwowski, M., Dul, J., Gralewski, J., Jauk, E., Jankowska, D.M., Gajda, A., Chruszczewski, M.H., Benedek, M. (2016). Is creativity without intelligence possible? A Necessary Condition Analysis, Intelligence 57, 105-117.