The common factor of conspiracy mentality: a bifactor analysis of four short-form measures of general conspiracy thinking

TomAlright, I saw this and had to bring it to you: The common factor of conspiracy mentality: a bifactor analysis of four short-form measures of general conspiracy thinking. What's going through your head?

CarolYou're asking me my gut reaction when the real problem is that everyone in that field is treating four different tests like they're the same coin flip. If those scales aren't measuring the exact same thing, then every conclusion drawn from them about who falls for nonsense is basically built on sand. I want to know why we'd trust a "common factor" without checking if the tools actually agree with each other first.

TomWait, hold on, I think you’re getting ahead of yourself here. They’re explicitly running a test because they suspect those scales might not be perfectly interchangeable, so saying it’s all built on sand misses the whole point of their work. But I see where you’re coming from—it’s scary how easily a bad measurement can poison a relationship dynamic, even if they’re trying to fix it.

CarolThat comparison is a stretch. A scale isn’t a partner; it’s a tool, and the authors are literally auditing whether that tool works before anyone uses it again. The issue isn’t moral failure, it’s statistical validity—if the bifactor model confirms a strong common component, then using these four measures interchangeably is defensible, but until that check passes, no one should cite them as equivalent.

TomYou’re making this sound like a courtroom drama over a spreadsheet, Carol. Look, if the bifactor analysis says there’s a solid core linking those four measures, then the field has a usable baseline for spotting where people drift into paranoid patterns. Without that shared ground, we’re just guessing at boundaries between healthy skepticism and actual delusion. It’s a practical fix, not a philosophical crisis.

CarolCalling it a philosophical crisis misses the mark entirely. This is a technical audit of four specific instruments, not a debate about human nature. Until that bifactor model proves the common variance is robust enough, the field needs to stop citing those scales as if they’ve already passed inspection.

TomThough focusing solely on the raw numbers ignores the messiness of how people actually latch onto narratives. From a coaching perspective, knowing which aspect of conspiracy thinking is driving the belief matters more than just confirming a generic score. If the bifactor model isolates that unique variance, we finally get a handle on what specifically triggers the spiral rather than just counting heads.

CarolThat’s a neat distinction, but it assumes the unique variance is even stable across those four instruments. Before you tell me what “triggers” the spiral based on leftover scores, I need to see if those residuals are consistent or just random noise hiding behind a complex model. If the common factor swallows most of the signal, your “specific trigger” is probably just an artifact of poor calibration.

TomYou keep waiting for perfect data before doing anything, which is exactly why conversations stall out. In my world, a rough estimate is good enough to spot red flags early while we gather better info later. We don't freeze up just because the initial readout isn't lab-grade precise.

CarolFreezing isn't the goal here; acting on garbage data is. If those four scales aren't reliably capturing the same trait, any "early warning" system is just amplifying noise and mislabeling normal skepticism as pathology. You can't coach someone out of a delusion you haven't correctly identified in the first place.

TomBut that argument cuts both ways. If the residual variance turns out to be real and distinct, then collapsing everything into one big "conspiracy mentality" bucket actually erases the very differences needed to understand why some people stay skeptical while others spiral. Your demand for a clean common factor risks proving too much by smoothing over the nuance the authors are trying to isolate.

CarolSmoothing over nuance is a risk you take when you ignore the loadings on the unique factors. If the residuals are statistically insignificant, they aren't deep psychological traits—they're measurement error—and building a coaching strategy on top of random noise just guarantees you'll miss the people who actually need help.

TomYou’re treating a statistical quirk like a moral failing, which is backwards. The authors ran a three-step test precisely to sort out whether those leftovers are real signals or just sloppy math, so stop preemptively dismissing them before the verdict comes down. If the common factor holds, great—we use it. If it doesn’t, we scrap those shortcuts and find better tools, instead of assuming everyone who questions authority is broken.

CarolAssuming everyone who questions authority is broken isn't coming from the stats; it's coming from ignoring who gets flagged by these scales. A clinical psychologist has to know if the tool separates actual distress from healthy civic doubt, otherwise we end up pathologizing people who simply ask hard questions. The three-step test tells us if the math works, but it doesn't tell us who the damage falls on until we look at the false positives.

TomYou’ve got a point there—if we skip checking who gets caught up in the net, we’re just guessing where the harm lands. But the research still needs to finish sorting out whether those extra bits of data matter before anyone starts diagnosing people with them. Let’s wait for the math to settle before deciding if the label is useful or just wrong.

CarolWaiting for the math to settle doesn’t protect anyone from getting mislabeled in the meantime. The bifactor analysis proves the structure exists, but it says nothing about whether the score accurately reflects reality versus just measuring how easily someone gets swept along by bad information. Until we verify that link, every high score remains a guess rather than a diagnosis.

TomFine, admit it then—you’re stalling on the obvious fact that without validating the link between the raw score and actual behavior, the whole scale is just a fancy echo chamber. Until someone proves that high number predicts real-world paranoia instead of just gullibility, I’m done pretending those four shortcuts are anything more than expensive guesses. That’s the gap you keep dancing around, Carol, and it’s the only thing standing between this metric and total uselessness.