Social & Critical Psychology · Simplified
Where the Prevalence Comes From
A description becomes a thing when it acquires a line, an instrument and somebody with a use for it.
Construction is not fabrication
The five people below are describing something. They are tired, they dread opening their messages, they stay up late for an hour of quiet. Nothing here suggests they are inventing it, and nothing here is a criticism of anyone who finds a category useful for naming their own experience. The question is a different one: what happens to a description once it is given a name, a threshold, an instrument and an institutional user, and what stops being asked as a result. The category, its features, its instrument and its prevalence are all invented for teaching. No real diagnosis, construct, published criteria or published estimate is described or reproduced, and nothing is administered to you.
The raw material
Five accounts
Five fictional people were asked what the last six months had been like. None of them used a name for it, because at this point there is not one.
| Person | What they said | Features coded |
|---|
Step 1 of 3
First, a judgement
Two years later a paper reports that eighteen per cent of working adults meet the criteria for the condition these accounts became. What does that figure most depend on?
Step 2 of 3
Four hundred people, and a line
Four hundred simulated people were asked about the same six features. Here is how many of the six each of them has. Nothing about any of them changes from here on. The only thing that moves is how many features you decide are enough.
Key terms
- Threshold
- How many features somebody must have before they are counted as a case. It is chosen, and it is rarely chosen by the people it will be applied to.
- Prevalence
- The share of a population above the line. When the underlying thing is continuous, the line is doing most of the work.
- Object status
- How much a category behaves like a thing in the world rather than a description somebody wrote. A name, a number and an instrument each add to it.
One bar per number of features. The shaded bars are the people who count as cases at the line you have set. The bars never change height; only which of them are shaded.
Step 3 of 3
Three items, and who they find
Six features is too many to ask about in a screening questionnaire, so pick three. Somebody is counted as a case by your instrument if they would endorse at least two of your three items. The five people from the start are still there, and they have not changed either.
| Person | Items they would endorse | Found by this questionnaire? |
|---|
What this shows
Two numbers that were made, not found
Key idea: Nobody in the four hundred moved, and the prevalence went from most of the population to a small minority, because a prevalence is a line as much as it is a fact about people. Then the questionnaire decided something else: not how many, but which. The three features that sound most like the condition are the ones two of the five original people never mentioned, and both of those two described difficulty that comes from their hours and their obligations rather than from anything inside them. No set of three items finds all five. Once the name, the number and the instrument are in place, the category starts behaving like a thing that was there all along, and the question that quietly stops being asked is the one those two people were answering: what would have to change for this to stop. None of this means the difficulty is not real. It means the object is partly made of decisions, and it is worth being able to say which ones.
This is not an argument that categories are bogus. Classification is how any science starts, a name for something people could not previously name is a real gain, and a threshold has to be somewhere if anybody is going to be offered anything. Three cautions. The four hundred people are simulated from a model with a single underlying dimension, which is a strong assumption and is what makes the line look so obviously arbitrary; a category with genuine natural joints would behave differently, and whether any given category has them is an empirical question this page cannot settle. The instrument rule here, two endorsements out of three, is a crude stand-in for real scoring. And nothing on this page refers to any real diagnosis, construct, criteria set or published estimate.
The longer version adds the naming stage and what a name imports before any evidence, a comparison set that shows which features fail to distinguish anything, separate duration and interference rules, and a fifth stage in which institutional decisions harden the category or hold it accountable. It is at Constructing a Category in the main collection.