Stats 1 Flashcards
name the three steps in hypothesis testing
1) define a null hypothesis
2) compare observed and expected
3) work out probability observed expected difference happened by chance
p
significant - reject null hypothesis. unlikely to have happened by chance
If an odds value is 1 then do you accept/reject the null hypothesis?
accept
How do you determine Chi2 - three steps
a - work out expected ((row total observed x column total observed )/ overall)
b – sum of ((expected- observed)2 / expected)
c – look up P value against table
What degree of freedom is
a) Chi2
b) t test
1) 1
2) 2
What is important about numbers for the expected count in a Chi2
must be over 5 for it to work
What hypothesis test would you do for normally distributed continuous data
t tests
Which test for categorical data?
Chi2
Which test doesn’t work with unpaired data?
Chi2
What do you calculate in an independent t test and how
T value. Use differences between mean. diff/Standard error = T value. Then look up p value sing table
What are you looking for with 95% confidence intervals of t test mean values
Do they cross 0? if 0 then no differences between the means so no statistical significance
What would you do if your continuous data is not normally distributed?
try log or use none parametric test like mann-whitley
When do you use odds ratio
Categorical outcomes
when do you use mean difference
continous outcomes