Psychosocial constructs related to healthy eating in female college freshmen: Reliability analysis and influence of body mass index

Kedem, Leia E.

Permalink

https://hdl.handle.net/2142/18618 Copy

Description

Title

Psychosocial constructs related to healthy eating in female college freshmen: Reliability analysis and influence of body mass index

Author(s)

Kedem, Leia E.

Issue Date

2011-01-21T22:51:57Z

Director of Research (if dissertation) or Advisor (if thesis)

Chapman-Novakofski, Karen M.

Department of Study

Nutritional Sciences

Discipline

Nutritional Sciences

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Keyword(s)

• social cognitive theory
• internal consistency
• Cronbach's alpha
• psychosocial
• self efficacy
• outcome expectations
• health
• questionnaire
• survey
• attitudes
• beliefs

Abstract

There were three objectives of the present study, which evaluated nutrition belief questionnaires used with a sample of female college freshmen (n=268) participating in a weight gain intervention. Surveys measured social cognitive constructs including self-efficacy (SE), outcome expectations (OE), food environment (FE), and goals (G); items from Project EAT (EAT) and the Weight Efficacy Lifestyle Questionnaire (WELQ) were also used. The first objective of the study was to investigate surveys for reliability using Cronbach’s α and principal components analysis (PCA). Cronbach’s α assessed internal consistency while PCA compared underlying data patterns with previously published item groupings. Second, validity was assessed through comparison of these two methods. Third, psychometric properties of surveys were compared among normal- and overweight subjects. Body mass index (BMI) was calculated using baseline height and weight data; subjects were then categorized as normal- or overweight using classifications issued by the National Institutes of Health. If scales had adequate Cronbach’s α values (α > ~0.70), composite scores were calculated as mean scores of all items within a construct. Composite scores for reliable constructs were compared by BMI category using Mann-Whitney U. Factor structures for the different surveys were also compared with PCA using body mass index (BMI) category as an independent variable. SE, OE, EAT, and WELQ had good internal consistency (α > ~0.70) based on previously published item groupings. Previous groupings were not available for G and FE, so only those indicated by PCA were tested for internal consistency; α values were high for G subscales but were not adequate for FE. Agreement between PCA and Cronbach’s α indicated that the item grouping for OE was a valid reflection of the construct; most SE and EAT constructs were also similar when comparing the two methods. WELQ groupings for PCA vs. Cronbach’s α original groupings were considerably different than described, indicating that these groupings may not be valid in this sample and other groupings may be more reliable. Scales may better address population concerns by integrating such issues as social pressure, emotional eating, food availability, and alcohol use. FE should be refined and further tested due to low α values. Only a few composite scores were significantly different according to BMI. Overweight subjects were more concerned about their weight (p<.01) and had lower SE to consume vegetables when physical effort is required (p<.05). More obvious differences were seen through PCA, with factor structures differing by BMI for each questionnaire except for OE. Normal- and overweight subjects may have varying cognitive patterns regarding nutrition and health, possibly due to differing influences and habits regarding health behaviors. Questionnaires may not have been constructed so that items had the same meaning for all subjects; psychosocial variables may not have been adequately measured in order to make valid comparisons. An additional finding was that in the sample as a whole, scores were very positive for OE but less so for SE and WELQ. Scores for G indicated that subjects had goals to incorporate fruits & vegetables and calcium-rich foods less often as compared to lower-fat foods. EAT scores were mostly positive except for weight concern and taste barriers. Due to less positive scores for SE, G, certain WELQ subscales (availability, social pressure, emotional eating), and taste barriers from EAT, interventions in this population may be more successful by incorporating strategies to overcome these barriers.

Graduation Semester

2010-12

Permalink

http://hdl.handle.net/2142/18618

Copyright and License Information

Copyright 2010 Leia E. Kedem

Owning Collections