First Impressions Towards Autistic People: A Systematic Review and Meta-Analysis.

Nobody in these studies met anyone. Impressions were formed from photos, recordings or transcripts, so the review cannot say what happens once an autistic and a non-autistic person actually interact. The autistic people used as stimuli were small in number, and several studies reused the same recordings or photos. Agreement between studies may therefore be agreement between new groups of raters looking at the same faces, not independent confirmation. The funnel plot suggested publication bias, and correcting for it cut the pooled difference from 0.54 to 0.34. The studies varied widely in their results (I2 = 88.09), so a single underlying effect is unlikely. The three item categories were chosen by the review authors from the literature rather than worked out from the data, and the authors note some items could sit in more than one category. Most raters were non-autistic, and only a few studies included autistic raters, so the finding that autistic raters share the bias rests on a small base.

The authors declare no conflicts of interest. The supplied text contains no funding statement.

Do not turn this into a target for therapy. These studies only asked observers to rate photos, voices and recordings. No treatment was tested, and nothing here shows that changing how an autistic person speaks or presents would change how they are treated. The review's own reading is that these judgments come from social norms and stigma, because autistic raters showed the same bias as non-autistic ones.