A psychometric evaluation and application of a measure of food literacy among young adults

Abstract

Background: Food literacy is an emerging study area encompassing the knowledge, skills, and attitudes required to navigate dynamic food systems. Food literacy has been suggested as a potential leverage point for improving diet quality and health outcomes. The emerging adult population, facing unique dietary and life transitions, is highlighted as an important group in public health research. Following calls for standard and well-evaluated measures of food literacy, a collaborative working group of public health nutrition practitioners in Ontario, Canada, led the development of a comprehensive food literacy measure, the FLit50, for use with young adults. To support the broad use of the measure in public health practice, assessment of the construct validity of the FLit50 and the development of a shortened version were desired by the public health nutrition practitioners.

Objectives: The objectives of this dissertation were to: (1) evaluate the construct validity of the FLit50 measure among post-secondary students (Chapter 5); (2) analyze the characteristics of the FLit50 items to facilitate the development and evaluation of a shortened measure (Chapter 6); and (3) explore the demographic, income adequacy, studentship, and health correlates of food literacy among post-secondary students (Chapter 7). This work was undertaken in collaboration with the public health nutrition practitioners.

Methods and results: The first study (Chapter 5) drew upon data from 457 post-secondary students in Ontario, Canada, to evaluate the measure's construct validity by assessing whether it could differentiate between groups hypothesized to have differences in food literacy. The FLit50 showed adequate construct validity, evidenced by higher median food literacy scores among students in food and nutrition programs (KW χ2) = 108, p < 0.001), women (KW χ2= 49.2, p < 0.001), those with adequate health literacy (KW χ2 = 20.6, p < 0.001), those with higher general health (KW χ2 = 49.5, p < 0.001), those with higher mental health (KW χ2 = 17.4, p < 0.001), and those experiencing food security in the past 12 months (KW χ2 = 21.9, p < 0.001), as hypothesized. No differences were observed by age (KW χ2 = 5.24, p = 0.63) or perceived income adequacy (KW χ2 = 4.21, p = 0.38). Differences in group means were observed as hypothesized for scores on items reflecting the underlying domains of food and nutrition knowledge and self-efficacy and confidence, but not food skills or ecological factors.

The second study (Chapter 6) involved the application of the 2-parameter Item Response Theory, using data from postsecondary students (n=457) along with data collected from young adults across Canada during the initial development of the measure (n=351). Item difficulty (mean= -1.72 SD from the sample’s average food literacy ability level, range: -3.64 to 3.05 SD) and item discrimination (mean= 1.78, range: 0.33 to 8.43) characteristics were estimated for items on the FLit50. The parameters informed discussions with the public health nutrition practitioner partners to select sixteen items for inclusion in the shortened measure, the FLit16. The correlation coefficient between scores on the FLit50 and the FLit16 was estimated, and whether the FLit16 could differentiate among groups hypothesized to have different levels of food literacy, as per study 1, was examined. Scores from the two versions of the measure were strongly and positively correlated (Rho = 0.87, p < 0.01) and the short measure was able to differentiate among groups, consistent with the full measure.

The third study (Chapter 7) drew upon data from the sample of postsecondary students to explore associations between demographic, income, studentship, and health characteristics, identified a priori based on the emerging food literacy literature, and food literacy scores using multiple linear regression analysis. Data from 413 students were included in these analyses after accounting for missing data on the characteristics of interest. Food literacy was assessed using the FLit50 measure, with a mean score of 42 of 49 points. Adjusting for other characteristics, food literacy was higher among women compared to men (β = 2.509, p < 0.001) and those who reported positive or neutral general health (β = 1.743, p < 0.001). Food literacy was lower among individuals identifying as East/Southeast Asian (β = -2.036, p < 0.001), South Asian (β = -2.409, p < 0.001), and other racial/ethnic identities (β = -1.568, p = 0.005) compared to those identifying as White. Food literacy was also lower among those who lived on-campus (β = -1.457, p = 0.073) and those experiencing food insecurity (β = -1.274, p = 0.004). Food literacy scores did not differ by age, income adequacy, domestic or international studentship, whether students attended college or university, household composition, or self-reported mental health status. The regression analysis yielded an R2 of 0.403, indicating that 40% of the variance in food literacy scores was explained by this model, with an overall significance of F(15, 396 DF) = 17.79, p < 0.001.

Conclusions: This dissertation continues the work of the collaboration of public health nutritionists by furthering two of its main aims: establishing the construct validity of the FLit50 and developing a shortened measure that demonstrates construct validity. The availability of two well-evaluated measures presents opportunities to better understand and monitor food literacy in emerging adults and assess associations with factors such as diet quality.