Multivariate effects of demand characteristics on the analogue assessment of heterosocial competence.

The team ran a lab test with neurotypical college students.

They wanted to see if telling students what results the study "should" give would change how the students acted.

Each student did a short role-play chat with an opposite-sex partner.

Some students got hints that shy behavior was normal. Others got hints that outgoing behavior was expected.

The researchers then scored 16 different social skills.

Only 3 of the 16 scores moved at all.

Those shifts were tiny and showed up only on self-rating sheets, not on observer counts.

In plain words, the hints barely nudged the students.

Cullinan et al. (2001) and Lancioni et al. (2008) also used short lab tests, but with kids who had developmental disabilities. Their tests quickly showed clear behavior changes when consequences changed.

That looks like a clash: why did their analogue tests pick up big effects while Annable et al. (1979) did not? The gap is in what was manipulated. The later studies changed real reinforcers like attention or escape. The 1979 study only changed verbal hints about what was "normal." Hints are weaker than consequences.

Foti et al. (2015) used a similar college-lab set-up and found strong learning effects after brief naming training. Together these papers show that analogue sessions work fine; you just need to use powerful variables, not weak demand cues.

You can relax about demand cues in brief social-skills probes. Telling a client what is "expected" probably will not swamp your data. Spend your time controlling real reinforcers instead of hunting for hidden hints. If you need a clean test, run a short functional analysis like Cullinan et al. (2001) rather than trying to out-smart every possible expectancy.