If you want to know whether this project is any good, start here rather than with anything I have built.
Lenhausen, Bleidorn and Hopwood (2023) gave 1,194 people a Big Five inventory seven times, each under a different framing: unprompted, compared to people in general, compared to close others, compared to people your age, your gender, your past self, your ideal self.
The between-person framings — including people in general versus close others — moved scores by d < .05. That is essentially nothing. The within-person ones, past self and ideal self, moved scores by up to d = .98.
So the specific manipulation that most resembles mine has already been tested at scale and come back empty.
The escape, and why it is thin
Their instruction changed the comparison standard: rate yourself relative to close others. Mine changes the behavioural target: how you actually behave toward close others.
Those are different psychological operations. One asks you to move a yardstick. The other asks about different events in your life. I think the second is a real manipulation where the first is nearly a no-op, and there is theoretical support for that.
But notice what the defence rests on. It is carried entirely by item wording. Which turns a theoretical argument into an engineering constraint, and makes it something that can be got wrong quietly:
- Every stem has to be unreadable as a comparison. "I go out of my way to help the people closest to me" is a target. "Compared to people close to me, I go out of my way" is a standard, and would be fatal.
- The original items are never modified, and never differenced against the new ones.
- Pilot respondents should be asked to paraphrase stems back. Any paraphrase that returns as a comparison is a wording failure, not a respondent failure.
What I have committed to
If the gradients here also come out at d < .05, the construct is dead and I say so on this site, in the same place and at the same size as I would have said the opposite.
That is not modesty. It is the only thing that makes the earlier claim worth anything. A prediction registered before the data, with a stated condition for being wrong, is evidence. The same words written afterwards are decoration.