Reliability: extent to which same result is obtained upon repeated measuring. The stability of a measure (important for making predictions, e. g. height) Validity: does it measure what you want it to measure. Assess degree to which to raters/judges give same measure. Decreases because of: poor operational definitions, untrained observer, procedural errors, bias. Determined using a correlation between the judges" scores: should be at least 0. 90 or higher. Test of temporal stability: same test administered twice. Parallel forms reliability: parallel form of same test administered on two occasions. Determined, using a correlation between two scores: should be at least +. 75. Keep in mind the amount of time elapsing between tests. Consider half the test as one administration and other half as another administration. Score on one split correlated with score on other split: should be at least +. 65. Cronbach"s alpha: average of all possible split half correlations.