In cqnstruct validity, the investigator is concerned with the questions the concept under investigation being adequately measured? Is there a valid basis for inferring the scores? This begins when the investigator formulates hypotheses, based on theory, about characteristics of those who have high scores on a test (compared to those with low scores). The measure can be dependent variable in some hypotheses and an independent variable in others. Evidence of construct validity is not established within a single study.
Variables like anxiety, creativity, job satisfaction, attitude towards numerous phenomena, intelligence, empathy, dependency, power, self-concept, etc. are all constructs. These variables have been constructed from theory and empirical data (i.e. quantitative data from research). For measures of variables that are constructs, evidence must be presented that scores represent the degree to which an individual does indeed possess or exhibit the construct. One common approach to construct validation is "known-groups technique" where the group which is expected to differ on the particular attribute because of some known characteristics are administered the test. For example, in validating the tool to measure fear in pregnancy of primigravida, the researcher would also administer the tool to multigravidae. Because one would logically infer that women who have not given birth (primi) would express more fear than multigravidae.