From Bayes through marginal utility to effect sizes: a guide to understanding the clinical and statistical significance of the results of autism research findings.

Ekas et al. (2011) wrote a how-to guide. They show how to turn autism study numbers into real-world meaning.

The paper walks readers from p-values to effect sizes to clinical payoff. No new data—just clear rules.

The authors found that most autism papers stop at “significant.” They say that is not enough.

You need effect size, cost, and client value to judge if an intervention is worth using.

Tromans et al. (2018) looked at 529 autism trials and saw most are tiny. Ekas et al. (2011) already warned that tiny trials can flunk effect-size tests—same worry, different angle.

Provenzani et al. (2020) counted 327 different outcome tools in 406 trials. Ekas et al. (2011) say this chaos hides true effect sizes; the two papers together push for common metrics.

Stewart et al. (2018) meta-analysis shows parent-training gives small but real gains. Ekas et al. (2011) give the ruler that turns those small numbers into “yes, clients will notice” decisions.

Next time you read an autism study, skip the p-value paragraph. Jump straight to the effect-size table. If authors did not report it, email them. Demand Cohen’s d or percentage of non-overlapping data. Tell supervisors, “Small effect with big cost equals don’t adopt.” Make effect size your gatekeeper for new programs.