Squaring the correlation coefficient gives the percent of variance accounted for. Statistical significance may show how likely it is that an association is not due to chance. To be scientifically useful, the observations and measures used to do constructs should be consistent/ reliable: the same test is given to the same group of people on two occasions, a test/retest correlation shows temporal reliability. Internal consistency: the correlation between parts of a single form. Interscorer agreement/ reliability: the degree to which different scorers/ judges agree. Content validity: shows if the terms of a test sample the domain they are supposed to. Criterion validity: shows the relationship between a behavior and the score on other relevant measures. Construct validity: does a test measure what it is supposed to. Independent variable: the condition that is manipulated by the experimenter. Dependent variable: the resulting behavior that is measured. Experimental group: the cohort that receives the manipulated condition.