Last week I wrote about “conscientiousness in the brain” and the (artificial) example
“sportiness → engine size → max.horsepower → max.speed”
which illustrates that arrows into a path diagram don’t have anything to do with causality.
Well, just some hours after that a student presented her research project on multimedia learning in physics, and she had the following real-life example of a poorly designed path diagram:
“gender → previous knowledge → motivation → effective use of learning strategies” etc.
This example is wrong in even more ways than the previous, artificial one: not only is gender a different kind of effect (e.g. you can’t manipulate it and it won’t change over time — usually), even worse is the lack of arrows from gender to the other constructs if one even decides to include it into the model as a causal effect. Consider the following statement: “Jane’s being a girl caused her to have less previous knowledge”. Of course, it might be argued that being a girl is on average causes exposure to stereotypes (“Girls aren’t good at physics, they should rather learn to knit and sew”) etc. etc., but then gender is only a distal effect and surely not directly related to previous knowledge. But if we accept gender as a distal effect, then there should also be an arrow to the other constructs: Surely the stereotype exposition would just as well lead to a general reduction of motivation, effective use of learning strategies etc.
P. S. Is that grammatically valid, “path diagramming”? Well, what I’m writing about isn’t really “valid”, so who cares.