Development and validation of a short form Children's power of Food Scale
Matthew D Stone,Brittany E Matheson,Adam M Leventhal,Kerri N Boutelle
DOI: https://doi.org/10.1016/j.appet.2019.104549
IF: 5.016
2020-04-01
Appetite
Abstract:Objective: To develop and validate a shortened form of the Children's Power of Food Scale (C-PFS), which measures anticipated reward from consuming highly palatable foods (i.e., hedonic hunger). Presently, two gaps exist with the C-PFS: the need for a shorter tighter measure, and evidence to support similar item function across populations. Method: Ninth grade students (N = 3277; 14.1 ± 0.4 years; 53.5% Female; 47% Hispanic) from 10 Los Angeles high schools completed the C-PFS and other surveys in class. Factor analysis, graded response modeling, and differential item functioning explored the structure of the 15 C-PFS items and identified a reduced set that parsimoniously taps hedonic hunger across the latent continuum and exhibits item-level invariance across sex, race/ethnicity, and weight status. Construct validity was examined via associations of self-reported dietary intake, impulsivity, and body mass index (BMI) to C-PFS scores. Results: Factor analytic models supported a single, primary dimension of hedonic hunger that accounted for 61% of the variance across all 15-items (α = 0.94). Adequate severity, discriminatory ability, and non-overlapping item-difficulty were observed for 11-items, of which 9-items were found to have item-level invariance across demographic and weight status groupings. Poor performing items were removed to create a 9-item scale (C-PFS-9; α = 0.93). Construct validity was demonstrated as higher C-PFS-9 scores were significantly related to greater sweet (β = 0.32, [95%CI = 0.23, 0.41], p < .001) and fatty food intake (β = 0.34, [95%CI = 0.26, 0.43], p < .001) and impulsivity resulting from positive (β = 0.11, [95%CI = 0.02, 0.21], p < .05) and negative mood (β = 0.36, [95%CI = 0.28, 0.45], p < .001). Females, relative to males, reported higher C-PFS-9 scores (β = 0.10, [95%CI = 0.02, 0.17], p < .05) and associations with BMI were mixed. Conclusion: The C-PFS-9 possesses excellent psychometric properties and retains the original construct coverage of hedonic hunger without a marked decrease in information obtained.