The reason a self-selected panel overstates strong opinions

Eleven per cent respond. Of those, a third are furious, a third are enthusiastic, and the report describes a divided workforce. The seventy-nine per cent who did not reply were mostly people for whom the topic did not rise to the level of an opinion.

What everyone sees

A low response rate is understood as reduced precision — a smaller sample with wider intervals, correctable by weighting to known demographics. Weighting addresses composition, not intensity. Whether someone bothers to answer is correlated with how strongly they feel, and no demographic variable in the file carries that information.

What is actually happening

Rogelberg and Stanton set out the structure of organisational survey nonresponse, distinguishing passive non-response from active refusal and showing that respondents and non-respondents differ on attitudinal variables in ways that demographic weighting cannot repair, because the propensity to answer is itself an attitude. Keeter and colleagues compared a standard survey with a much more intensive effort achieving far higher response and found that many estimates were similar, while a subset — particularly measures of civic engagement and issue intensity — differed systematically. The nuance matters for honesty: low response does not corrupt everything, but it reliably overstates whichever variables correlate with the willingness to answer, and intensity of opinion is the clearest of those.

Why it stays hidden

The distortion hides because the report reads as a distribution. Percentages sum to a hundred, subgroups are compared, and the indifferent majority appears nowhere — not as a category, not as a caveat. Findings are then acted on as if the workforce were divided, which produces communications aimed at a conflict most people were not in.