Blackwell Reference Online

Extract

Reliability refers, at a general level, to consistency of measurement. Consistency can be conceptualized somewhat differently for different forms of reliability estimation, but in all cases reliability is focused on whether a measurement yields consistent results. Such consistency is critical to research practice, where variables must be operationalized and measured in some fashion. For example, the measurement of socioeconomic status can be operationalized as average family income, whether a child receives a reduced lunch rate at school, education level of parent, or by other variables. Regardless of the way in which the variable is defined, however, it must be measured with consistency within the research study such that the scores obtained reflect dependable characterizations of the units of observation (e.g., people, families) on the variable of interest. In the example above, if a head of household does not know his or her average family income, he or she might simply guess at an estimate. Conversely, another head of household in the same study may give an accurate average family income. In such a case, the variable is not being consistently measured across the units of observation. There are three dominant measurement theories that can be used to conceptualize reliability of scores: classical test theory , generalizability theory , and item response theory . In research ... log in or subscribe to read full text