I’ve been locked away for the last few days, analyzing data from one of my theoretically boss projects. The results are not quite as boss as hypothesized…cute but not sexy.
I was really looking for sexy.
My grad students and I are going to apply one more old, somewhat straightforward, technique to grind analyze these data, although there is some contention about the method’s validity in my particular field of study:
Spirit, oh Spirit, what is the true p-value?
Dammit.
June 11, 2009 at 3:06 AM
Sometimes cute is better than sexy.
June 11, 2009 at 7:03 PM
p=0.06. well, it is statistically significant at the 10% significance level.
June 11, 2009 at 9:51 PM
The sanctity of p<0.05 is a load of crap, IMO. Decision rules are arbitrary, so I think that 10% is just fine in many cases. Low N in field studies, anyone?
June 12, 2009 at 2:27 PM
Like EGF, my default p-value is 0.1, because I tend to deal with extremely messy datasets (yay, “natural variability.”)
June 14, 2009 at 11:53 PM
p<0.05 is far far too high.
You can immediatley remove your disapointment with your results by retrospectively adjusting alpa to <0.001.
That way you're no where near close and can relax.
June 23, 2009 at 4:08 PM
Love it. What did the Spirit say?
June 26, 2009 at 10:46 PM
Spirit said: work harder.
June 29, 2009 at 2:40 AM
LOL. Maybe the Magic 8 Ball will have a more gentle message…