Dear Yimin,

What your recite from the review you got is by no means specific to QCA. Some researchers simply seem to think that, only because one has not measured all potentially relevant variables ("...environmental outcome depends on other kind [sic] of external factors that are not included in the analysis..."), the purpose of a method breaks down. I highly recommend the recent article "The C-Word: Scientific Euphemisms Do Not Improve Causal Inference From Observational Data" by Harvard epidemiologist Miguel Hernán "The C-Word: Scientific Euphemisms Do Not Improve Causal Inference from Observational Data" (Article The C-Word: Scientific Euphemisms Do Not Improve Causal Infe...

Kevin C. Johnson has also already asked a similar question on ResearchGate in this thread: https://www.researchgate.net/post/Want_to_help_with_QCA_Frequently_Asked_Questions_FAQ_List.

Best wishes,

Alrik

Dear Prof.Thiem,

Thank you for answering my questions immediately all the time.

I am working on causal inference in the area of environmental governance. My last paper employed synthetic control method to examine the CAUSAL effect of democratic transition on carbon intensity. In that paper, my causal interpretation was also met with skepticism from the anonymous reviewers.

I will read the references you have suggested. Additionally, your package QCApro is great.

Yimin

Dear Yimin,

You're most welcome. I wish you the best of luck with your QCA paper.

Best wishes,

Alrik

Dear friends,

I am trying to use fsQCA to study conditions for Foreign Direct Investment in Latin America, including why some countries can apply some policies without pussing out investors, but other countries can not...

I think that the causal question is related at least with two issues: the type of causality you are looking for, and the number of observations you have. When you use QCA, you are allowing your data to tell you something about sufficient conditions, even if that conditions can not explain the phenomenon in most cases (they are not necessary). Mahoney 2008 and Ragin's papers show how correlational methods look for sufficient and necessary conditions. When you do not have a lot of observations, maybe you can posit some causal inference with OCA and your good judgement. This is what Ragin arguments. So I think that saying "I do not think that the authors are able to identify any causality effect with this kind analysis(i.e. fsQCA)." is a preconception or prejudgement, and that's why it's a mistake. I think that fsOCA does not identify causality, it is the way how you use it that can make it useful or wrong.

Mahoney, J. (2008), “Toward a Unified Theory of Causality”, Comparative Political Studies, 41, DOI: 10.1177/0010414007313115, http://cps.sagepub.com/cgi/content/abstract/41/4-5/412

Ragin. C. (2009), “Qualitative Comparative Analysis Using Fuzzy Sets (fsQCA)”, Chapter 5 in Rihoux, B. and Ragin, C. (eds.), Configurational Comparative Analysis, Sage Publications

Dear Alrik and Gustavo,

Thank you for your answers.

A related question is about the interpretation of the QCA results.

For example, there are four conditions(A,B,C,D) and one outcome (E). The QCA result is A*B--->E, with high score of consistency and coverage( e.g. 0.9 and 0.7). Then, what is the appropriate interpretation of this result from causality perspective?

Here is my interpretation: According to the QCA result, A in combination with B is sufficient for the presence of E.

However, many researchers, including myself, are suspicious about this causal interpretation. I think A*B may be the INUS cause of E because the QCA result may change into A*B*F--->E if I include F as a condition into this model.

Could you provide some suggestions to clarify my thoughts and optimize my interpretation?

Yimin

Dear Yimin,

To say that something is sufficient for something else has, per se, nothing to do with causality. Thus, when you say that "A in combination with B is sufficient for the presence of E", this is just what the data, on the basis of the laws of Boolean algebra, tell you functionally. It neither implies that A and B are conjunctive causes of E, nor that nothing else apart from A and B is needed to effect E. Unfortunately, QCA researchers, and social scientists in general, tend to heavily overinterpret the concepts of "sufficiency" and "necessity" (see http://cps.sagepub.com/content/49/6/801.abstract).

However, under the INUS theory (and its extensions), certain sufficiency relations represent causal structures. These causal structures have the following, general functional form: P*Qi + R S (see Mackie 1965:246-47). Strictly speaking, every QCA result should thus add Q to any conjunction, and R, indicating that there may be other unmeasured or unknown causes of S. For example, your result would have to look like this: A*B*Q1 + R -> E.

A last point, which is not to be underestimated in this connection, but which is often swept under the carpet: a proper causal theory and an established algebraic system only exist for crisp (bivalent and multivalent) sets. I know of no work to date that has successfully extended Mackie's INUS theory to "fuzzy causes". Either something is a cause, or it is not. It can be a weak cause, or a strong cause, but what is a cause that is a cause to degree x and is not a cause to degree 1 - x? I have never seen a substantive argument that speaks of conditions that are causes and non-causes at the same time. This is why there are still so many unresolved problems in fsQCA, unlike in csQCA and mvQCA.

Best wishes,

Alrik

References

• Mackie, J. L. 1965. “Causes and Conditions.” American Philosophical Quarterly 2(4): 245-64.

Hi

A good contributor to QCA on this forum is Eva Thomann. I recommend you to check her contributions on RG.

Best

Guillaume

QCA as any other qualitative or statistical technique never identifies causality. New advances in causal inference techniques (e.g. regression discontinuity, and the like) can provide some "inferences," but the problem is impossible to solve. The best way to deal with such comments is to acknowledge that QCA, as any other technique, does not has causal identification as a goal. In this article we use the technique and phrase it that way. And it was published :) Article US hegemony and regime change in Latin America