Pengembangan dan Evaluasi Psikometrik Awal Instrumen Konsumsi Makanan Ultra-Proses Berbasis Klasifikasi NOVA pada Orang Dewasa Indonesia

Measuring ultra-processed food (UPF) consumption requires a brief instrument that is relevant to the NOVA classification and suitable for local use. The objective of this study is to develop and evaluate the content validity, exploratory factor structure, and internal consistency of a UPF consumption instrument. This methodological study involved NOVA-based item development, content validation by five experts, and field testing among 258 respondents. The field-test instrument comprised 13 items with seven frequency categories referring to the previous month. Analyses included descriptive statistics, I-CVI and S-CVI/Ave, the Kaiser–Meyer–Olkin (KMO) measure, Bartlett’s test, parallel analysis, exploratory factor analysis using minimum residual (MINRES) extraction, Cronbach’s alpha, McDonald’s omega, and item–rest correlations. Participants had a mean age of 30.36±10.11 years, 64.3% were women, and 57.4% resided in North Sulawesi. For the 13 retained items, the mean expert rating was 3.65/4 (92%), S-CVI/Ave was 0.969, and all I-CVIs were ≥0.80. The data were suitable for EFA (KMO=0.892; χ²(78)=1026.460; p<0.001). Parallel analysis supported one factor. The one-factor MINRES solution explained 33.90% of the variance, with loadings ranging from 0.460 to 0.708 and moderate model fit (RMSEA=0.078; SRMR=0.057; CFI=0.892; TLI=0.870). Internal consistency was good (α=0.859; ω=0.860), with item–rest correlations ranging from 0.428 to 0.661. The 13-item instrument demonstrated excellent content validity and good internal consistency, with a dominant general factor but suboptimal construct fit. Further validation using test–retest reliability, polychoric correlations, independent-sample CFA, and comparison with established dietary assessment methods is warranted before widespread use.