Review and analysis of evaluation measures used in nutrition education intervention research

The purpose of this review is to provide a summary of the kinds of evaluation measures used in 265 nutrition education intervention studies conducted between 1980 and 1999 and an analysis of psychometric issues arising from such a review. The data are summarized in terms of tables for interventions with each of six key population groups: preschool children, school-aged children, adults, pregnant women and breast-feeding promotion, older adults, and inservice preparation of professionals and paraprofessionals. Measures evaluating knowledge and skills or behavioral capabilities were most widely used in studies with preschool, school-aged, and inservice populations (50%-85%) and less widely used in studies with the other groups, particularly breast-feeding promotion (5%). Measures of potential psychosocial mediators or correlates of behavior such as outcome expectancies, self-efficacy, or behavioral intention were used in 90% of behaviorally focused studies with school-aged children and in about 20% of studies with adults. Dietary intake measures were used in almost all studies, primarily food recalls, records, and quantitative food frequency questionnaires. Short frequency instruments involving only foods targeted in the intervention such as fruits and vegetables are increasingly being used. Measures of specific observable behaviors are also increasingly being used. Physiologic parameters were used in about 33% of behaviorally focused interventions with school-aged children and adults, 20% with older adults, and 65% with pregnant women and/or their infants. Criterion validity of newly developed intake instruments and content validity of instruments measuring mediating variables were reported in the majority (range 50%-90%) of studies. Reliability and stability of measures of mediating variables were reported in 50% to 75% of studies, with reliability coefficients mostly about .6 to .7. Two major conclusions from this review are that evaluation measures should be appropriate to the purpose, duration, and power of the intervention and that measures should have adequate validity and reliability in relation to both the outcomes and characteristics of the target audience. Major implications are that considerable preliminary work needs to be done before any intervention study to develop and test evaluation instruments so that they are appropriate and have adequate psychometric properties, and cognitive testing of published instruments with each new target audience is essential. We will then be better able to make judgments about the effectiveness of nutrition education.