With big data we’re finding statistical significant correlations but no measurable elasticities. E.g. vehicle acceleration change of 0.0045m/s^2 for one unit change in a road environment change.
-
-
-
This is what typically happens with increasing sample size, a robust effect that is undetectable for a human, which in turn is meaningless in the greater context. Yours is a perfect example. Wish I could adjust my acceleration in those steps when biking :-).
- 1 more reply
New conversation -
-
-
You agree that there's nothing wrong with testing for statistical significance, right? What you call "statistical-significance thinking" is substituting significance testing for more meaningful evaluations of the data. These researchers didn't even have a representative sample!
-
I think he means it's unwise to think dichotomously. Consider: p=.03 p=.06 p=.45 Stat sig thinking would have us categorize the last two as non-signif and the first as signif, even though the first two are nearly identical. Better to think on a continuum without thresholds.
- 4 more replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.