2 Correlation and Causation.pptx
Document Details
Uploaded by ImpressedRed
Tags
Full Transcript
Correlation and Causation Prof. dr. Bart Cambré Antwerp Management School Causality issues If A is associated (=correlates) with B, then not only is it possible that A causes B, but it is also possible that B is the cause of A ‘Wrong’ causality often leads to bad managerial recommendations ›...
Correlation and Causation Prof. dr. Bart Cambré Antwerp Management School Causality issues If A is associated (=correlates) with B, then not only is it possible that A causes B, but it is also possible that B is the cause of A ‘Wrong’ causality often leads to bad managerial recommendations › High debt causes low economic growth · Managerial recommendation: save to get out of recession! · Solution? › Refugees show more criminal behavior · Recommendation: keep them out! Characteristics of ‘wrong’ causality 1. There is no explanation Paul: 8 correct predictions (2012) or 1/256 But so many predictions and predictors, there will always be one that is correct Thus: RANDOMNESS, there is no explanation for the causal relationship 2. It is the other way around Rain causes the use of an umbrella (=correct) What about the correlation between burnout and part-time work? What comes first? Causality = cause + effect (time) Solution: longitudinal research or experiment Fisher (1958, 21): “Is it possible that lung cancer (…) is one of the causes of smoking cigarettes? I don’t think it can be excluded” 3. The missing variable Women with mastectomy show a higher death rate But: › Age › General health condition › So other conditions explain more Selection bias: two groups differ on more than the variable(s) studied Therefore: multivariate instead of bivariate analysis Do storks really deliver babies? 4. Other reasons Wrong model testing e.g. dealing with a mediator/moderator and not test for it (Lipperman-Kreda et al., 2009) Important unmeasurable variables Causal complexity Multi-causality instead of mono-causality (all variance attributed to single variable) Correlation is not Causation Professor Bart Cambré bart.cambré@ams.ac.be Antwerp Management School Opening minds to impact the world