Please enable JavaScript.
Coggle requires JavaScript to display documents.
STAT193 (HYPOTHESIS TESTING (t-test for 1 sample (is mean of the sample=…
STAT193
HYPOTHESIS TESTING
Sign test
-
-
-
-
calculate p-value with binomial distribution. p-value is what is the chance of observation or more extreme occuring?
-
-
t-test for 1 sample
-
kind of like the sign test but with mean, so effected by skew
-
-
-
-
calculate test statistic with t formula 
to calculate p-value, uses the t-distrubtion like it's normal.
t-test for differences
-
same process as with the t-test for 1 sample, however instead of using raw values, we're using difference values
-
-
test statistic calculated with t equations again, with mu=0
-
z-test
-
same as t-test, however uses the standard normal distribution. used when we don't know if the distribution is normal or not
-
-
-
-
chi^2 test
-
-
-
-
-
p-value will always be one tailed and is just the probability of finding the observation or more extreme in a chi^2 dist
-
ANOVA test
-
-
must have equal variance, normal distribution and random
-
-
-
will also get confidence intervals to analyse as part of the test to figure out which groups have the association
t-test for two means
-
same as ANOVA, but can be two tailed and only 3 groups in the categorical
-
-
-
-
errors
-
-
-
-
-
as significance level gets smaller, reduce liklihood of type 1, increases type 2
power of test is the ability to detect a difference that exists in the population. can be calculated by 1-type 2 error
-
DISTRIBUTIONS
-
normal
-
conditions
-
symmetrical, unimodal, bell
-
-
-
-
-
-
F distribution
-
also relies of degrees of freedom, but note there are two different types of degrees of freedom
CENTRAL LIMIT THEOREM
-
-
-
-
-
-
overall, allows us to approximate values based off our assumption of normality
DISPLAYING DATA
Variables
numerical
continuous
-
dotplot, boxplot, histogram
-
categorical
-
nominal
categories don't have an order, just a name
-
-
-
-
-
-
-
-