Prelim #2 Flashcards
Types of Test
Test of a single proportion
Goodness of fit test
Test of association
null/reference distribution of a test of single proportion
binomial distribution
null/reference distribution of goodness of fit test
chi-square distribution
null/reference distribution of a test of association
chi-square distribution
assumptions for chi-square test
- adequate cell sizes
- all cells must have expected counts >=1
- at least 80% of the cells need expected counts>=5
assumptions for binomial distribution
- the number of trials, n, is fixed
- individual trials are independent
- there are only 2 outcomes for every trial
- the probability of success (p) is the same for every trial
assumptions of poisson distribution
-observations are independent and the probability of an observation is constant throughout
clumped
variance>mean
odds ratio
a/c/b/d (odds of success in one group divided by the odds of success in another group)
relative risk
pr(worse outcome in group 1)/pr(worse outcome in group 2)
finding p value in R
1-pchisq(teststat, df)
finding critical value in R
qchisq(percentile, df)
Type 1 error
rejecting a true null hypothesis
Type 2 error
(beta) failing to reject a false null hypothesis
hypothesis testing assumes that…
sampling is random
if test statistic is >critical value
reject null
if p-value is <0.05
reject null
significance level
the probability used as a criterion for rejecting the null hypothesis
null distribution
the distribution of the test statistic, when assuming that the null hypothesis is true
if odds ratio confidence interval contains 1?
no association
if odds ratio does not contain 1?
statistically significant association
standard error of the mean formula
sd/sqrt(n)
CI for sample mean
sample mean +/- 2 SE
power
the ability of a test to reject a false null hypothesis
1-beta
the smaller beta (type 2 error)…
the more power a test has
a larger sample gives more/less power?
more
dispersed
variance is less than mean
p-value
The probability of getting the result, or something as unusual or more unusual, if the null hypothesis were true
