Selecting items for a food behavior checklist for a limited-resource audience.

OBJECTIVE: To report 6 psychometric properties of food behavior checklist (FBC) items and then to use these properties to systematically reduce the number of items on this evaluation tool.

DESIGN: Random assignment to the intervention and control groups.

SETTING: Low-income communities.

PARTICIPANTS: Women (N = 132) from limited-resource families.

MAIN OUTCOME MEASURES: Reliability, internal consistency, baseline differences by ethnicity, sensitivity to change, and criterion and convergent validity of subscales.

RESULTS: The fruit and vegetable subscale showed a significant correlation with serum carotenoid values (r =.44, P <.001), indicating acceptable criterion validity. Milk, fat/cholesterol, diet quality, food security, and fruit/vegetable subscales showed significant correlations with dietary variables. Nineteen items have acceptable reliability. Twenty items showed no baseline differences by ethnic group. Eleven of the 15 items expected to show change following the intervention demonstrated sensitivity to change.

CONCLUSIONS AND IMPLICATIONS: This brief food behavior checklist (16 items) is easy to administer to a client group, has an elementary reading level (fourth grade), and has a low respondent burden in addition to meeting requirements for validity, reliability, and sensitivity to change. This study establishes a process that can be used by other researchers to develop and further refine instruments for use in community health promotion interventions.