Anova and Chi Square

Anova

compare means of 1 qualitative with 2+ groups and 1 quantitative

unadjusted type of variable

ANOVA null hypothesis

the overall means of the groups are the same

ANOVA null hypothesis reading

if highly significant < 0.05

one of the means is different

Bonferroni Post Hoc

adjusts p-values for acculumated 5% alpha error

variances are equal

not significant Levene

Tamhane’s

adjusts p-values

variances are not equal

significant Levene

Levene’s test

aka homogeneity of variables

tests for which post hoc test

Kruskal-Wallis

aka k-independent samples

compare medians of 3+ samples or in abnormal qualitative variable

adjusted type of variable

Kruskal-Wallis null

all of the medians are the same

Kruskal-Wallis null reading

if highly significant < 0.05

difference in medians between groups

Finding which median is different

graphs —> chart builder —> bar with error bars

scale Y and quantitative X

change to medians and hit ok

Confidence Intervals reading

no overlap means significantly different

Chi-square test

evaluates if two quantitative variables are related to each other

shown as numbers and %s in a cross tabulation

Chi-square test null

two quantitative variables are not related/associated to each other

means that they are statistically independent

Chi-square test null reading

if the p < 0.05 there is a significant association between them

Expected values

total row x total column / total total

Chi-square assumptions

no expected frequency should be less than 1

no more than 2-% of the cells should have an expected frequency less than 5

observations must be considered to be independent not a post and pre

fisher’s exact test

only used if expected frequencies are too small

Generating Chi-Square

analyze → descriptive → cross-tabs

add the 2 variables

exact → monte carlo 95%

statistics → chi-square and mcnemar

cells → observed, expected, column

Mcnemar test

if the two variables are the same pre and post

2 × 2

Mcnemar-Bowker test

if there are more than two variables that are the same

3 × 3 table