Title | 10. Inferential Statistics. Analysis of Variance ( Anova) – One Way Anova |
---|---|
Course | Statistics |
Institution | Indiana University - Purdue University Indianapolis |
Pages | 8 |
File Size | 246.6 KB |
File Type | |
Total Downloads | 33 |
Total Views | 170 |
This lecture discussed inferential statistics, including the analysis of variance (ANOVA), and one way ANOVA....
Inferential Statistics: Analysis of Variance (ANOVA) – One Way ANOVA Multiple Tests Why is that so bad? o When we reject the null, we have no idea whether or not the null is actually false! That's not what the NHST tells you! o So for any given rejection, I don't know if it was a correct decision or a Type I error! This is why we want to minimize Type I error rate. In the field of psychology, we say we're okay with falsely rejecting the null 5% of the time, but we don't want to make that wrong decision more than that! Multiple tests "inflate" Type I error rate o i.e., reject the null when the null is true o Alpha = 0.05…for a SINGLE TEST or per comparison C Alpha less than or equal to p(at least 1 Type I error) less or equal to 1 - (1 alpha)^C less than or equal to Ca Error rate for a single test Mid-case/likely scenario Worst case scenario o If you run 6 tests, what is the worst case scenario for at least one Type I error rate? C * Alpha 6 (0.05) 0.30 0.05 or p < 0.05, so write that (whichever goes with the decision) o You can leave effect size and descriptive stats blank F(2,6) = 10.66 F(crit) = 5.14 o "We the null hypothesis that null hypothesis here, test-statistic (degrees of freedom) = _____, p = _____, effect size name = _____, (interpret what the conclusion means using descriptive stats, your verbal hypothesis, and "significant" or "not significant"> o We reject the null hypothesis that there is no effect relatability on ratings of purchasing LaCroix, F(2,6) = 10.66, p < 0.05, effect size = _____. There is a significant effect of relatability on ratings. The ratings are significantly different between relatability conditions. Relatability significantly affect the ratings of how likely participants are to purchase LaCroix Where 's the difference, thought? Do I go with both savvy social media AND traditional? Is savvy social media greater than traditional? o ANOVA does not tell you that! According to a post-hoc analysis called Schaffa's test, the only significant difference is between the influencer and savvy social media.
...