Session 7a - comparing independent means Flashcards
t-distribution is…
Used to compare a continuous variable in 2 groups (the groups are treated as a binary categorical variable)
A set of normal distributions - for mean group differences
Defined by degrees of freedom
df for t-distribution
Total sample size minus number of groups
n-2
(if not independent groups, df=n-1)
critical t-score is
Compared to observed t-value
The cut off values for different proportions of the population - can use for 95%, 99% significance etc
Standard Error of the Mean is
an estimator of the population standard deviation
measure of uncertainty around the mean
Standard Error of the Mean (se) is affected by
Size and Variability of a sample
Degrees of freedom for t-distribution calculation
Total sample size minus number of groups
Confidence Intervals (CIs)
CI for a statistic - range that is likely to contain the statistic for that population
Standard = 95% CI
Calculation of CI
mean ± critical statistic (e.g. critical t) * standard error of the
mean
Confidence limits =
Values that state the boundaries of the confidence interval
The mean difference can be assumed to be statistically significant
If the interval excludes 0
Confidence intervals can be displayed graphically for group differences by…
Whiskers - NOT box-plot
Overlapping CIs suggests non significant test results
Independent sample t-test assumptions
Observations are independent
Observations come from normal distributions
Distributions have equal variance (Levene’s)
Levene’s test for equality of variance
H0 is that the groups are equal variance
If p<0.05, variances are different and therefore need adjusting for - read from ‘equal variances not assumed’ row