> The key analysis was a 2 × 2 mixed ANOVA with Time (Pre vs. Post) as the repeated (within subjects) measure and Group (Control vs. Treated) as the across-subjects measure. Given the matching of other variables including age, gender and years of education, we did not include these variables as covariates in the model. A two-sided P less than 0.05 was considered statistically significant. We used SAS9.4 for all statistical analyses.

Can anyone explain why is it reasonable to assume normality here?

I am not sure about this specific case, but I want to point out that there is a big misunderstanding with this normality assumption in t tests: it's not the distribution of the data that needs to be ~normal, but rather the distribution of the sample mean, which usually is ~normal, thanks to the central limit theorem.

https://journals.physiology.org/doi/full/10.1152/advan.00064...

But the government of the US can issue gag orders

Is there a way to add similar functionality to Garmin? These watches can vibrate, which might not be as effective, but still a reminder