Statistical test hypothesis testing (week 2) Flashcards
why do we use statistical tests
to figure out how likely our results are to be a product of chance
what does the t-test take into account
expected mean, and the standard error of the mean (estimated on the sample)
how does the t distribution change with number of participants
gets more narrow (closer to normal)
three types of t tests
one sample design (one group with values from different people), independent measures (comparing two different groups) repeated measures (same group over multiple conditions)
one sample design info (when to be used, what it is, advantages, disadvantages)
used for one group, with values got from different people. Advantage is it can be used when we know information about the group. Disadvantages, we don’t always know whole population values, we may want to compare groups, or a group over time.
independent measures design info (when to be used, what it is, advantages, disadvantages)
comparing two groups, people are in one group OR the either. Advantages, measures are independent, dont have to participants learning effects. Disadvantages, people in groups vary alot, so we need large sample sizes or to counterbalance, cannot study over time.
repeated measures design info (when to be used, what it is, advantages, disadvantages)
same group over multiple conditions to test wether the change in condition changes the response to the same test. Advantages, dont have to think about differences in personality etc., study changes over time, usually have fewer people. Disadvantages, measurements not independent (calculate variance differently), people know treatment after first time, need to counterbalance conditions.
difference between one and two tailed, when we would use them
use one tail when the hypothesis is directed (like if you expect not just any difference in results, but specifically a higher value). Use 2 tailed when hypothesis is not directional (just expecting some difference, for it not to be in the centre. )
when would you use non-directional hypothesis, and what effect on the graph
when the hypothesis is expecting ANY difference. like for H1: u>10 OR u<10. or like for “predicted there would be a SIGNIFICANT DIFFERENCE in mep’s”. For graph, means 2.5% on either side is significant.
what are the two types of effect size measures
cohens d, r^2
cohens d size guide
0.2 is small, 0.5 is medium, 0.8 is big
r^2 size guide
0.01 small, 0.09 medium, 0.25 large
what are the assumptions for when running a t test (the rules per say)
no influence from other participants or biases of assigning people to groups. sample populations must be normal. if multiple groups, both sample variance must be same.
order of writing results
indicate what tests were run, what the results were, and what the results indicate in terms of hypotheses
