When I had introductory parametric statistics, it was explained that one in twenty was deemed an acceptable number of good calls vs bad back inna day. Nineteen to one and all that. Good enough for lowest bar. But if it is important or the decision is a critical one, ya go p=<.01 as a minimum. One inna 100 and all that.
In isolation, RESPECT is a loser. If it was the only study, a case can be made that it was as likely an outlier in the positive direction as not.
But it isn't and as such it is an underpowered study trending in the expected direction as shown by other better designed studies and highly unlikely to be an outlier.
Market Data 
Markets 
AI
