I have one categorical Independent variable --Background Experience(CNC and Conventional) and one DV which is also categorical variable --Intentions to adopt system ( CNC, Conventional , Either ). To test to different hypotheses such as "employees with CNC background exp tend to adopt CNC systems more than conventional systems", I have used joint probability to test these hypotheses. I have been asked by journal to go about this through QCA.
can any body help me in identifying which technique of QCA to use ( fs QCA, mv QCA or cs QCA) and which parameters to take into account to test directional hypotheses?
Thanks in advance!!
Rafique
Dear Rafique
my understanding of qca is to uncover how different combinations of "driver" variables may lead to a specific outcome (level).
I can't see how this method suits the study of a relationship with one independent variable. If, however, you study how the extent of CNC experience AND the extent of Conventional experience jointly impact adoption intention, then qca may be helpful.
I am not sure about the nature of scaling of your variables, but if you can dichotomise your dependent and ind't variables, then crisp set qca is the choice, if you have a more nuanced scaling, then fs qca.
Hope this helps
Oliver
Thank you Koll for your valued reply.
In my case there are three outcomes and a dichotomous IV . i do not know how to go about this?
Regards
Rafique
Rafique
in that case I can't see qca as a useful path. There are no combinations of causal factors to be investigated - if indeed for every person all you have is one driver with two levels and one outcome with three levels it looks like a cross tab is the right path.
Oliver
Thanks Oliver,
I think i need to incorporate other IVs in my data for analysis and use of qca such as education level with three categories and gender (my IVS will become CNC, NCNC, GRADUATE, MASTER, SPECIALISTS, MALE , FEMALE) but my question remains same whether we can deal 3 outcomes in QCA If yes then how? should I go for 3 models with same IVs and separate outcome ( adoption of CNC, Adoption of NCNC and EITHER)
kind regards and once again thanks for response.
Rafique
yes - I would run three models separately then:
what are the combinations of IVs that most likely lead to each of these outcomes?
kind regards
Oliver
thank you Oliver, for contribution from problem to a meaningful solution.
Kind Regards
Rafique
Dear Rafique,
For testing configurational hypotheses of the kind "A and B are sufficient for C", you do not need QCA (see the attached article of mine on how to test configurational hypotheses). But if your goal is causal inference with observational data, QCA is the right tool. However, as soon as one of your exogenous factors has more than two levels, each of which is important to your analysis, you should use mvQCA, not fsQCA. The number of categories of the endogenous factor is irrelevant, since all QCA variants are designed to analyze a single outcome at a time, which can come from either a bivalent endogenous factor or a multivalent one. Last but not least, note that you should use the QCA package for R, since it is currently the only software for QCA that is capable of deriving ALL alternative models that fit your data equally well.
Article Mill's Methods, Induction and Case Sensitivity in Qualitativ...
Thank you Thiem for valued input. Your point is well taken. What is still not clear is that whether i can go for QCA with 2 DVs or single outcome is mandatory for this analysis.
Regards
Rafique
- Badges
- Science topic
- Similar topics
- Biological Science
- Anatomy
- Tissue