Symptom Validity

I’ll be posting an update soon with a number of papers and such from the last year, along with links to those PDFs and a brief discussion of what I found in each and how I think about it. I wanted to make a few notes about SVTs and symptom validity testing here, and how I’m seeing the literature broadly. It seems fairly clear to me that there are not reliably and incrementally useful differences in the supposed domains of over-reporting (e.g., cognitive, psychological, somatic) and that things, instead, seem to function in a more unified manner – with over-reporting as a general function.

My way of thinking about this currently is that as abstraction processes loosen, as a function of controlled executive function, mood regulation, or other cognitive process – intentionally [consciously] or not, the greater the interpretive band about item meaning becomes. As it loosens, I expect that people try to apply novel ideas to bounded language concepts – leading to poor description (such as when trying to feign psychosis, e.g., dissimulation) or other inaccuracy of type (misunderstanding normal or symptom trajectories). I think this is why we are seeing the same patterns repeat across MMPI-2/RF/3 and PAI+ over-reporting scales. Its why the same moderators appear, and why effect size ranges tend to standardize. I think scores outside of this range are likely an unrealistic example of abstraction processes – its also why they most often occur in simulation designs with students, the weakest feigning/malingering/etc design. It seems like what is happening is that there is a general bounded ceiling to which most people adhere (with, of course, deviations from bounding) – similar to how without speed limits everyone has a speed they feel ‘most comfortable with’ while driving. There are speed demons too, of course.

I was looking at some of the ratios of change between conditions/moderators across meta-analyses (Herring et al., Ingram & Ternes, Sharf et al., etc etc.) and it seems like the ratio between the domains decreases as the hedge’s g/cohen’s d of the overall effect size increase. This reinforces my belief that its a bounded system driven by this more cohesive and unified over-reporting process.

I’ll be posting some of the articles that focus on this in detail that published this last year shortly, and will have several updates about papers upcoming that tend to point to the same findings, including national VA samples with the MMPI-2-RF