Household Nutrition Security Scale (HNSS): Further Establishing Validity and Reliability in a Statewide Sample in...

One of the first survey measures for assessing nutrition security in the United States is referred to as the Household Nutrition Security Scale (HNSS). Although the analyses conducted in the original HNSS development study supported the validity and reliability of the scale, some survey development steps were beyond the scope of the original study, such as confirmatory factor analysis (CFA) and differential item functioning (DIF).

This study aims to conduct CFA and DIF analyses on the HNSS within a new statewide sample in Massachusetts.

From November 2022 to January 2023, the Greater Boston Food Bank and Mass General Brigham conducted a cross-sectional, statewide, representative survey of Massachusetts adults. In addition to other topics, respondents answered questions about sociodemographics and completed the HNSS. To assess CFA model fit, the Satorra-Bentler scaled chi-squared test, root mean square error of approximation (RMSEA), standardized root mean square residual (SRMR), and the Bentler Comparative Fit Index (CFI) were calculated. Cronbach's alpha was calculated to assess internal consistency. The Cochran-Mantel-Haenszel procedure was used to investigate DIF across demographic groups.

Participants in the analytic sample (n = 2903) were mostly at or under age 54 (60.7%), women (58.0%), non-Hispanic White (71.3%), and with household incomes <$75,000 (71.1%). CFA model fit metrics were well within thresholds for adequate model fit [chi-square 5.213 (Degrees of freedom (DF) = 2, P = 0.074), RMSEA = 0.024 (95% confidence interval: 0.000, 0.049), SRMR = 0.006, and CFI = 0.999]. Cronbach's alpha was 0.923. There was no indication of DIF for any of the 4 items across groups for age, race/ethnicity, gender, educational attainment, income, having children in the household, or household size.

These findings further support construct validity and reliability of HNSS, and the further use of the scale for research and program evaluation purposes within sociodemographically diverse samples of households.