Measuring Technology Acceptance
Building the instruments — micro scenarios, conjoint designs, semantic differentials — that make "would people accept this?" an answerable question.
Technology acceptance research has an instrument problem. A great deal of it reduces to agreement ratings on a handful of items, which produces numbers that are easy to collect, easy to publish, and hard to act on. If the answer is “4.2 out of 5”, nobody knows what to change.
So a good part of my work is method building rather than method application. Micro scenarios present many short, concrete situations instead of a few abstract statements, which recovers the structure that averaging destroys and lets public opinion and individual differences be read from the same data. Conjoint analysis makes people trade attributes against each other rather than endorse all of them, which is how preferences behave outside a questionnaire. More recently I have been pushing on the response format itself, replacing agreement scales with a semantic differential that measures the construct without the acquiescence that agreement items invite.
The newest piece of this is an open-source platform that uses generative AI to build and run conjoint studies, which lowers the cost of asking these questions properly. That matters more than it sounds: most acceptance work is underpowered and ad hoc not because researchers do not care but because doing it well has been expensive.
User diversity sits at the centre throughout. Age, gender, technical self-confidence, and domain expertise reliably move acceptance, and a result reported as a single population mean has usually hidden its most useful finding.
