From: Quantitative measures of health policy implementation determinants and outcomes: a systematic review
Scale | Domain | Definition |
---|---|---|
Pragmatic criteria | Brevity | Number of items; excellent < 10 items |
Language simplicity | Readability of items, ranging from accessible only to experts (poor) to readable at or below an 8th grade level (excellent) | |
Cost to use instrument | Monetary amount researchers pay to use the instrument; excellent = freely available in the public domain | |
Training ease | Extent of assessor burden due to required trainings versus manualized self-training; excellent = no training required by instrument developer | |
Analysis ease | Extent of assessor burden due to complexity of scoring interpretation; excellent = cutoff scores with value labels and automated calculations | |
Psychometric properties | Norms | A measure of generalizability based on sample size and means and standard deviations of item values |
Internal consistency | Reliability | |
Convergent construct validity | Observed association in data of two theoretically related constructs, assessed through effect sizes and correlations | |
Discriminant construct validity | Observed differentiation (lack of association) of two theoretically distinct constructs, assessed through effect sizes and correlations | |
Known-groups validity | Extent to which groups known to have different characteristics can be differentiated by the measure | |
Predictive criterion validity | Extent to which a measure can predict or be associated with an outcome measured at a future time | |
Concurrent criterion validity | Correlation of a measure’s observed scores with scores from a previously established measure of the construct | |
Responsiveness | Extent to which a measure can detect changes over time, i.e., clinically important not just statistically significant changes over time | |
Structural validity | Structure of test covariance, i.e., extent to which groups of items increase or decrease together versus a different pattern, assessed by goodness of fit of factor analyses or principal component analyses |