Stats week 10

Stats week 10

1. variance

   • R2 is the variance in y explained by the model relative to the total Vance in y - we can express r2 as a % by multiplying it by 100 - in regression we assume that x explains the variance in y
   • zero order = total variance in y explained by x
     • how much variance in record sales could radio plays explain if they were the only predictor?
   • partial2 - proportion of the remaining variance in y explained by x when other predators are removed
     • what proportion of variation in record sales that cannot be predicted by tv interviews, can be explained by radio plays?
   • part2 - unique variance in y explained by x
     • using SPSS, you need to square the 0.something numbers and x by 100 to get proportion
     • how much variance in record sales could radio plays explain on their own?

2. power

   • refers to the ability to find a statistically significant effect when one exists (ability to reject the null when it is false) - measured on scale of 0 - 1 where 0 = no power
   • power of 0.1 means you have a 10% chance of finding an effect if one exists
   • factors influencing it - effect size = small effect size, harder to find effect - criterion significant - more conservative, harder to find effect, sample size, stats test (harder to find with non-parametric, subjects design - harder with independent designs, one or two tailed hypithsis, easier with one tailed
     • calculated by knowing number of participants, effect size and criterion significant level