is this a foundation year. i swear to god.
t tests:
examine differences between groups of people
compare distributions of groups - central tendency and variability
shows us whether there is a statistically significant difference between two groups
data needs to meet assumptions:
normal distribution
interval/ratio data
homogeneity of variance = data sets are spread out the same amount
tested via levene’s test - significant test = no homogeneity of variance
no extreme data scores
parametric
non parametric equivalent = wilcoxon test for repeated measures, mann whitney for independent
see slides for actual equation but the components are:
difference between means of sample (m1-m2)
variability of data in samples (sd1 and sd2)
sample sizes (n1 and n2)
2 types of t test:
independent measures = used for independent groups design - data from 2 different groups
repeated measures/paired samples = ppts are assessed twice and give two sets of scores
mann whitney:
assumptions:
non parametric data
comparing 2 groups
independent groups
if you really want to know how jamovi does it for you see slides
wilcoxon signed rank:
assumptions:
non parametric data
comparing 2 groups
repeated measures
again there’s a slide of useless information about how software does it for you
pros and cons of non parametric:
pros:
do not rely on restrictive parametric assumptions
unaffected by outliers
cons:
less powerful than parametric
more observations needed for this than parametric to find significance
not as good for smaller samples
standard error:
shows us how accurately our sample represents the population
standard deviation over square root of number of people in the sample
estimated standard deviation for the central tendency of an infinite population
small se indicates representative sample
if you need a refresher on critical value tables give your head a wobble but they’re in there