Measurement of Constructs

  • 5 or 7 point scales
    • Bipolar or bidirectional scales (also known as semantic differential scales): unlikely / likely, disagree / agree, good / bad.
    • Unipolar or unidirectional scales: very little control / complete control, not at all worried / extremely worried.
  • Scoring from -3 to +3 such that:
    • Belief that behavior will result in a positive outcome contributes positively to measure.
    • Belief that behavior will result in negative outcome contributes negatively to measure.
  • Elicitation studies: studies conducted to identify and select the content for the construct measures.
    • Example: Before behavioral beliefs can be measured, the researcher must know what they are for the health behavior and population studied.
    • It is always important to pilot test your measures / instrument
  • Direct measurement refers to measurement on a major construct (example: attitude, subjective norms, perceived behavioral control). These are usually more strongly associated with outcomes than indirect measures. Demonstrate these associations before proceeding to indirect measures.
  • Indirect measurement refers to measurement on the minor constructs that comprise the major construct and then calculating these scores across the outcomes of the behavior.
    • Example:
      • major construct = attitude, indirect measurement measures
      • minor constructs = behavioral beliefs, evaluation of behavioral outcomes; calculation discussed later
  • It is important to show strong association between these measures and the major construct to ensure the appropriate beliefs were measured. If so demonstrated, these are of most interest.
  • Assess instrument reliability by test-retest reliability and/or internal consistency reliability.