In this section, we discuss factors to consider when selecting measures for inclusion in a behavioral intervention research trial. It is not always possible to meet all criteria; however, having an understanding of the issues is important and will help guide the choice of measures. The relevance of some criteria also varies according to the stage in the pipeline.
Psychometric Properties and Scale of Measurement
The psychometric properties of a measure are important factors to consider in measurement selection. Psychometric properties influence the quality of the information gathered from a measure and include factors such as level of measurement, reliability, validity, and sensitivity.
Level of Measurement
The measurement properties or level of measurement of a measure is also an important consideration as it affects the selection of statistical techniques that can be used to analyze the data and the interpretation of the data. Nominal levels of measurement simply assign responses or individuals by category such as male versus female or married versus single. Nominal scales do not imply any ordering among the responses. Ordinal scales also involve mutually exclusive and exhaustive categories in terms of responses such as degree of agreement with a statement, but the categories do not have equal intervals. Interval level scales involve order and equal intervals between units of measurement such as the units on a Fahrenheit scale of temperature. However, there is not a true zero point on interval scales. Finally, ratio scales are interval scales with a true zero point that indicates presence or absence of a quality. As noted, the concern about scale of measurement in behavioral intervention research is the implications for the appropriate statistical analyses.