The same myths seem to occur very frequently on ResearchGate, and, indeed, elsewhere. The myth of the need for data to be normally distributed leads the way, but there are others (one pet hate of mine is the mythology of acceptable values for Cronbach's alpha).
What are the myths that you struggle with in students, colleagues, the media?
That a sample size of 30 is adequate (as basic stats can be applied)
- p(D|H0) == p(H0|D)
- smaller significant p-values mean higher effect size
- significant p-value means that H1 is true
- not significant p-value means that H0 is true
- You can always use non-parametric methods to test the same hypothesis if your data does not allow for parametric methods
It's sad to see how this interessting question is not discussed and not more contributed to in comparison to the usual Why-Is-My-Regression-not-working-Thread on RG.
As Mafalda says…
http://4.bp.blogspot.com/-gVlF5wxFuGA/U-1m_Wvh2DI/AAAAAAAABiE/cyNbLRgDIZo/s1600/140813_quino_lo-urgente-y-lo-importante_mafalda-oficial.jpg
Chalamalla – Thank you for an excellent reading list! You are clearly interested in the area too…
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics