Please enable JavaScript.
Coggle requires JavaScript to display documents.
Reliability - Coggle Diagram
Reliability
estimating reliability
internal consistency
KR20
all possible forms (mean) of split-half
works only for dichotomous items -requires nr of right or wrong responses on an item
alpha coefficient
measure of first factor saturation
equal to reliability in conditions of essential tau-equivalence
lower bound of reliability of a test
more general version of KR20-sum of item variance instead of sum of right-wrong
mean of all split-half reliabilities
split-half
underestimated because half length - correction: Spearman-Brown formula
test-retest
parallel forms
estimating reliability of behavioural observations (multiple observers)
kappa
phi
factor analysis
if a test measures different(sub) traits
divided the test into independent yet internally consistent parts
correction for attenuation
ratio of true variance to observed variance