Module 4 Flashcards
when to use z (normal) distribution
if pop SD is known and the sample is above 30
t distribution
- used to see if a sample belongs to a pop and can be used to compare two dif groups
- used when pop SD is unknown or sample size is less than 30
- mean = 0, area = 1
- defined by df (more samples 30+ similar to z distribution, less heavier tails)
- greater tail area so CI has greater variability (high df closer to z)
dependent t test assumptions + df
assumptions:
- scale
- paired (pre, post)
- independent samples
- normality
df = observation number - 1
independent t test assumptions + df
assumptions:
- scale
- unpaired
- independent samples
- normality
- equal variance
df = observation number - 2
normality check
use a histogram or shapiro wilks test (greater than 0.05 = normal)
equal variance check
levenes test (greater than 0.05 = equal)
how to express t value
t(df) = x
parametric vs non parametric
parametric makes assumptions about pop (normality), non parametric do not (uses median, ordinal or scale, no equal/normal)
t test non parametric equivalents
dependent - wilcoxon signed ranked test
independent – mann whitney u test
comparing proportions
categorical data, uses contingency table, compares counts
- left column compared groups
- top characteristic of interest
- must include totals
chi square distribution
- used for contingency tables
- squared so only positive
- right skewed
- although two sided upper tail only considered
chi square df + measures of central tendency
df = (rows-1)(colums-1)
- small df more skewed, large less
mean = df
variance = 2df
mode = df - 2
chi square distribution assumptions
assumptions:
- df must be at least 1
- no more than 20% of cells have less than 5 observations
- independent groups
- mutually exclusive
chi square nonparametric
fishers exact test - more than 20% of cells with less than 5 observations
mcnemars test - paired groups, not independent