- Email: [email protected]
The diet and health knowledge survey: Development of a short interview format
Eating Behaviors 7 (2006) 235 – 242
The diet and health knowledge survey: Development of a short interview format Emily E. York-Crowe *, Marney A. White, Sahasporn Paeratakul, Donald A. Williamson Louisiana State University, Yale Psychiatric Research, and Pennington Biomedical Research Center, United States Received 29 August 2005; accepted 29 August 2005
Abstract The primary aim of this study was to create a condensed short-form version of the structured interview named the Diet and Health Knowledge Survey (DHKS 1994–1996) via factor analysis. Data from 5765 participants of the 1994–1996 Diet and Health Knowledge Survey were used in the factor analysis. Appropriate sampling weights were used in the multivariate analyses (n = 5233) that compared subgroups (i.e., ethnic group, sex, etc.) on the subscales of the DHKS. A 50-item, 12-factor condensed version of the 149-item original DHKS was derived from the analyses. Multivariate analyses, using age and education as covariates, indicated significant differences in dietary beliefs, practices, and behaviors for individuals of different ethnic groups, sex, dieting status, and BMI. Based on our analyses, the condensed short-form version of the Diet and Health Knowledge Survey appears to be a convenient and efficient tool, used in sections or in its entirety, for examining various dietary practices and beliefs of adults. The results from analysis of the short-form DHKS indicate that although a very large proportion of adults in the U.S. are overweight or obese, only a small proportion report current dieting. Examination of the DHKS indicates that although adults varied in their dietary beliefs and practices, dieters tended to report healthier dietary practices and attitudes than non-dieters. BMI level was positively related to self-reported amount of fat, cholesterol, and calories consumed. D 2005 Elsevier Ltd. All rights reserved. Keywords: Diet; Dietary knowledge; Dietary practice; Dietary attitudes; Obesity; Factor analysis; Diet and Health Knowledge Survey
There are only a few measures that can be used to examine dietary knowledge and practices. One such measure is the Diet and Health Knowledge Survey, an interview conducted by the U.S. Department of Agriculture in 1994 through 1996. This interview includes (a) self-perceptions of the adequacy of food intake; (b) dietary and health knowledge; (c) perceived importance of certain nutrients; (d) eating behaviors related to fat intake and food safety; (e) self-perceptions of weight; (f) knowledge about cholesterol and fat sources; (g) use of food labels; (h) accuracy and confidence in the use of food labels; and (i) frequency of using food labels. The content of the DHKS survey was shaped around the need for data concerning the population’s knowledge and attitudes about the Dietary Guidelines for Americans (USDA/DHHS, 1990), fat intake as related to dietary behavior, and questions surrounding the use and understanding of food labels. Kuchler and Lin (2002) examined the Diet and Health Knowledge Survey, and found that the choices and attitudes of the surveyed individuals were related to BMI and adiposity, although the factors affecting BMI * Corresponding author. Tel.: +1 919 699 2929, +1 919 572 9496. E-mail address: [email protected] (E.E. York-Crowe). 1471-0153/$ - see front matter D 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.eatbeh.2005.08.010
236
E.E. York-Crowe et al. / Eating Behaviors 7 (2006) 235–242
differed systematically by gender. Unfortunately, this interview is not extremely time efficient for research purposes (35 min), furthermore, each question was treated as an independent datum. There is a need for a shortened version of the DHKS that develops subscales so that the interview can be used with all or a subset of the questions of the original version of the DHKS. Such a tool could be used to gather important information for dietary and obesity research and educational purposes. No study to date has examined the dietary knowledge, beliefs, and practices of such a large (and recent) representative sample of U.S. adults as the DHKS. Based upon these considerations, we believe that the DHKS is an ideal interview to convert into a validated short-from version. To demonstrate the utility of the DHKS for investigating specific trends in dietary attitudes and behaviors, we conducted factor analyses on the most current version (1994–1996) with the goal of creating a more condensed, more efficient, short-form version of the DHKS. Based on the subscales identified by the factor analysis, we then examined the dietary knowledge and practices of a representative sample of U.S. adults as a function of a variety of demographic variables. For the purposes of this study, factors (subscale scores) were examined as a function of ethnic group, sex, BMI, dieting status, and age. 1. Method The 1994–1996 Diet and Health Knowledge Survey (DHKS 1994–1996) was a telephone follow-up interview to the Continuing Survey of Food Intakes by Individuals (CSFII). The interview was conducted in a nationally representative sample of adults in the United States using a stratified, multistage, area probability sampling design. The details of survey design and methods are available elsewhere (USDA: ARS, 2000). The DHKS utilized information on dietary and health knowledge and attitudes from individuals 20 years of age and older who participated in the CSFII. The original 1994–1996 DHKS was comprised of 149 items assessing dietary practices, attitudes, and knowledge.
Table 1 Sociodemographics of population used in factor analysis Variable Sex Male Female Age 20–40 41–59 60 and over Ethnic group Caucasian African-American Hispanic Household income(pct pov) 0–130% 131–350% N350% Education V High school NHigh school BMI group b20 20–25 25–30 30–40 N40 Dieting status No Yes
Sample size
% of sample
2897 2868
50.3 49.7
1852 1952 1961
32.1 33.9 33.3
4245 621 425
80.2 11.7 8.0
1481 2217 2067
25.7 38.5 35.9
3750 1952
55.8 44.2
380 2047 2060 1031 247
6.6 35.5 35.7 17.9 4.3
4764 1000
82.6 17.3
E.E. York-Crowe et al. / Eating Behaviors 7 (2006) 235–242
237
1.1. Participants The sample used in the factor analyses consisted of 5765 individuals. The sample was divided in half using a random seed, and the exploratory factor analysis was conducted on the first half of the sample. To cross-validate the confirmatory factor analysis, the remaining half of the sample was again divided in half with a random seed. The measurement model was then analyzed twice to further test the stability of the factor structure (i.e., two confirmatory factor analyses were conducted). For the multivariate analyses, missing data were imputed based on estimation using the mean for each variable, within ethnic and sex groups. Those individuals who did not fit into the ethnic and sex grouping were excluded. Alaskan natives, Asians, Pacific Islanders, and American Indians (total n = 364) were not present in adequate number to provide a reliable estimate and were excluded, resulting in a final sample of 5233 individuals. 1.2. Variables of interest Variables examined in this study included 1) ethnic group, 2) sex, 3) BMI (from self-reported height and weight), 4) dieting status, 5) age, and 6) education level (see Table 1). The sample was divided into 3 categories of racial/ethnic groups: Caucasian, African-American, and Hispanic. Body Mass index (BMI) was derived from selfreported height without shoes (in feet and inches) and weight data (in pounds). BMI was then divided into 5 groups based on the World Health Organization’s classification system (WHO, 2000): V 19.9; 20–24.9; 25–29.9; 30–39.9; z 40 kg/m2. According to these criteria, overweight was defined as having a BMI between 25 and 29.9 kg/m2 and obese was defined as having a BMI above 30 kg/m2. Dieting status was determined by endorsement of any of the following questions: 1) bAre you on a weight loss or low calorie diet?Q; 2) bAre you on a low fat or cholesterol diet?Q; 3) bAre you on a low salt or sodium diet?Q; 4) bAre you on a high fiber diet?Q; or 5) bAre you on a diabetic diet?Q. Education level, ranging from 0 to 17, was defined as the number of years of schooling completed. Responses ranged from bnever attended school or kindergarten onlyQ (0) to bfive or more years of collegeQ(17). 2. Results 2.1. Factor analysis Table 1 shows the characteristics of the sample used in the factor analysis. To conduct both exploratory and confirmatory factor analyses, the sample was divided in half (n = 2880; n = 2885) using a random selection of cases. Items on the survey were analyzed via principle component analysis using an orthogonal (VARIMAX) rotation. The Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy and Bartlett’s test of sphericity were calculated to assess the appropriateness of the data for factor analysis. The KMO index was .89 and the Bartlett’s test of sphericity was highly significant ( p b .001), indicating that the data were appropriate for analysis. Questions of the DHKS were retained if they had a factor loading of .35 or higher and if they loaded on only one factor. This procedure yielded a 12-factor structure composed of 53 items that accounted for 53.74% of the total variance. Only factors that had eigen values greater than 1.0 were retained. Also, a scree plot of eigenvalues confirmed the presence of a 12-factor solution for the factor analysis. To cross-validate the Confirmatory factor analysis, the remaining half of the sample (n = 2885) was again divided in half based on a random seed. The measurement model was then analyzed twice (n = 1446; n = 1439) to further test the stability of the factor structure (i.e., two confirmatory factor analyses were computed). In both confirmatory analyses, the 12-factor model was stable. The Root Mean Square Error of Approximation (RMSEA) was adequate for both (.046 and .043 respectively). Further, the Non-Normed Fit Indices (NNFIs) were acceptable (.84 and .83), as were the Comparative Fit Indices (CFIs) (.86 and .85). RMSEA values of .08 or less and NNFIs and CFIs approaching .90 are considered desirable. Analysis of the confirmatory factor loadings indicated that 3 questions did not load highly on the predicted factor (i.e., loaded less than .35 in one or both of the confirmatory analyses). Therefore, those items were dropped, yielding a 50-question final interview. Table 2 lists the individual questions and factor loadings for the three factor analyses for the 12 factors that were retained.
238
Table 2 Factor loadings for the exploratory and confirmatory factor analyses Factor 1
Perceived value of healthy eating Eating low-fat diet Eating adequate fiber Diet low in cholesterol Diet high in fruit and vegetables Diet low in saturated fat Maintaining healthy weight Moderating calories Moderating salt Eating variety of foods Plenty of bread, cereal, rice, pasta
0.71; 0.70; 0.70; 0.69; 0.69; 0.66; 0.62; 0.54; 0.54; 0.44;
Use of food labels Never used/never seen Use nutrition panel Use list of ingredients Use health/nutrients benefits label Use serving size Use bshort phrasesQ (e.g., light) Consumption of low-fat/low-calorie foods Eat/use low-fat cheese Eat/use low calorie salad dressing Eat/use 1% milk Eat/use ice milk/frozen yogurt Eat/use lower fat luncheon meat Knowledge of food guide pyramid servings Fruit group Vegetable group Bread, cereal, rice, pasta group Milk, yogurt, cheese group Meat, poultry, fish, dry beans, eggs Perception of adequacy of energy intake Fat Saturated fat
0.82; 0.67; 0.78; 0.69; 0.76; 0.56; 0.58; 0.48; 0.51; 0.36;
2
3
4
5
0.81 0.67 0.76 0.66 0.74 0.64 0.59 0.54 0.47 0.38
0.78; 0.72; 0.73 0.77; 0.84; 0.86 0.76; 0.79; 0.80 0.75; 0.73; 0.72 0.72; 0.70; 0.72 0.70; 0.75; 0.75
0.78; 0.70; 0.68; 0.60; 0.58;
0.75; 0.71; 0.59; 0.62; 0.63;
6
E.E. York-Crowe et al. / Eating Behaviors 7 (2006) 235–242
Factor labels and item numbers
0.72 0.73 0.63 0.65 0.61
0.75; 0.74; 0.72; 0.70; 0.60;
0.76; 0.76; 0.60; 0.54; 0.48;
0.71 0.76 0.64 0.53 0.42
0.82; 0.78; 0.78 0.80; 0.80; 0.79
Cholesterol Calories
0.67; 0.65; 0.59 0.61; 0.47; 0.47
Added fats Butter/margarine to vegetables Butter/marg/sour cream to potatoes Cheese/creamy sauce to vegetables Butter/margarine to breads/muffins Factor interpretations
0.78; 0.61; 0.58 0.66; 0.69; 0.65 0.58; 0.35; 0.36 0.56; 0.52; 0.55 Factor 7 0.65; 0.50; 0.63; 0.54;
Values affecting food purchasing decisions How safe food is to eat Nutrition Perception of nutrient intake adequacy Iron Fiber Calcium Consumption of fiber through fruit and vegetable peels Eat vegetable peels Eat fruit peels Avoidance of extra fat Trim fat off meat Remove skin from chicken Eat fried chicken Consumption of junk food Cakes, cookies, donuts Potato/corn chips
8
9
10
11
12
0.51 0.49 0.63 0.50
E.E. York-Crowe et al. / Eating Behaviors 7 (2006) 235–242
Awareness of consequences of poor nutrition Eating too much fat 0.69; Being overweight 0.66; Eating too much cholesterol 0.66; Eating too much salt/sodium 0.65;
0.57; 0.51; 0.45 0.50; 0.78; 0.83
0.71; 0.53; 0.46 0.60; 0.44; 0.38 0.58; 0.45; 0.47
0.83; 0.57; 0.72 0.83; 0.86; 0.73
0.72; 0.46; 0.37 0.72; 0.64; 0.59 0.38; 0.52; 0.53
0.71; 0.42; 0.46 0.70; 0.76; 0.77
The first column within each cell reflects the exploratory factor loadings; the factor loadings from both confirmatory analyses appear in the second and third columns of each cell.
239
240
E.E. York-Crowe et al. / Eating Behaviors 7 (2006) 235–242
The first factor, Perceived Value of Healthy Eating, consisted of 10 items measuring the extent to which participants endorsed the importance of moderating food intake and eating specific types of foods or nutrients (e.g., low saturated fat, high fiber). The second factor consisted of 6 items and assessed the extent to which participants reported using various features of food labels in making dietary selections; this factor was labeled Use of Food Labels. The third factor, Consumption of Low-fat/Low-Calorie foods, consisted of 5 items measuring participants’ use of low-fat alternatives in making food selections. Factor 4, Knowledge of Food Guide Pyramid Servings, asked participants to estimate the number of recommended daily servings of food groups classified according to the Food Guide Pyramid guidelines (Davis, Britten & Myers, 2001). Factor 4 consists of 5 items. Factor 5 asks participants to rate their belief in the extent to which their diet conforms to healthy amounts of calories, cholesterol, fat, and saturated fat intake. Factor 5 was labeled Perception of Energy Intake Adequacy. The sixth factor, Added Fats, assesses 4 eating behaviors that increase the amount of fat intake, such as adding butter to foods. Factor 7, Awareness of Consequences of Poor Nutrition, consists of 4 items assessing participants’ knowledge of negative health consequences associated with poor eating habits. The eighth factor, Values affecting Food Purchasing Decisions, consists of 2 items tapping food-purchasing decisions, and asks participants to rate the importance of nutrition and safety when making food selections. Factor 9, Perception of Nutrient Intake Adequacy, consisted of 3 items asking participants to rate the extent to which they believe they are consuming appropriate amounts of iron, fiber, and calcium. Factor 10 consisted of 2 items measuring the Consumption of Fiber through Fruit and Vegetable Peels. Factor 11 consisted of 3 items that assessed dietary practices that reduce the amount of fat intake, such as trimming skin from poultry and fat from meat; this factor was labeled Avoidance of Extra Fat. The final factor retained 2 items that asked participants to estimate the frequency of consuming sweets and chips and was labeled Consumption of Junk Food. Table 3 Summary of main effects Effect
Multivariate effect1
Sex
F(12, 5162) = 4.28
Subscale
Estimated marginal means Female 5.02b
F(1, 5173) = 25.49, p b .001
12.01a
13.06b
F(1, 5173) = 5.62, p b .020
Non-dieter 34.55b 3.52b 11.80b 4.46b
Dieter 36.06a 3.09a 10.08a 4.27a
F(1, 5173) = 10.84, p b .002 F(1, 5173) = 12.45, p b .001 F(1, 5173) = 13.80, p b .001 F(1,5173) = 5.08, p b .030
12.10b
12.99a
F(1, 5173) = 4.09, p b .05
Hispanic
7.36a 9.49a
AfricanAmerican 7.63b,c 9.01b,c
7.68c 7.68c
F(2, 5173) = 6.89, p b .001 F(2, 5173) = 10.66, p b .001
3.94a
4.77b
3.99a
F(2, 5173) = 16.31, p b .001
13.42b
12.21c
F(2, 5173) = 13.35, p b .001
6.23 b 11.90b,c
5.73c 12.22c
F(2, 5173) = 5.08, p b .007 F(2, 5173) = 7.04, p b .002
Perception of adequacy of nutrient intake Knowledge of Food Guide Pyramid servings Diet
F(12, 5162) = 2.98 Perceived value of healthy eating Consumption of junk food Use of food labels Aware of negative health consequences of poor nutrition Knowledge of Food Guide Pyramid servings
Ethnic group F(24, 10,324) = 5.80
Caucasian Food purchasing decision rules Perception of energy intake adequacy Consume fiber by fruit and/or vegetable peels Consumption of low fat/low calorie foods Avoidance of extra fat Knowledge of Food Guide Pyramid servings
BMI group
F(48, 19,887) = 2.55 Perception of energy intake adequacy
Univariate effect
Male 5.59a
11.93
a
5.89 a 13.41a
b 20 kg/m 2 20–25 kg/m 2 25–30 kg/m 2 30–40 kg/m 2 N40 kg/m2 8.21a 8.48b 9.21c 9.38d 10.10e
1 All multivariate effects were significant, p b .001. Different subscripts (a, b, c) denote significant group differences. Results from all univariate follow-up analyses are reported and discussed elsewhere (York-Crowe, 2002).
E.E. York-Crowe et al. / Eating Behaviors 7 (2006) 235–242
241
2.2. Examination of dietary practices Multivariate analysis of variance (MANCOVA) using the subscales as dependent variables revealed age and education were significant covariates ( F(12, 5162) = 90.10, p b .001; F(12, 5162) = 54.98, p b .001). Based on results of the MANCOVA, ANCOVAs were performed using age and education as covariates for each of the individual subscales. Ethnic group, sex, BMI, and dieting status were entered as independent variables, all of which were significant (all p-values b .001). In terms of gender, men reported consuming higher, more adequate, levels of iron, fiber, and calcium than women; women, believed that a person should eat more servings from the food groups in the Food Guide Pyramid than did men. In terms of dieting status, individuals who reported current dieting (dieters) were more likely to believe in the importance of healthy food, report consuming less junk food, using food labels more often when purchasing food, and endorse a belief that a person should eat more servings from the food groups in the Food Guide Pyramid. Dieters also reported a higher awareness of the consequences of the following behaviors: eating too much fat; eating too much salt or sodium; eating too much cholesterol; being overweight. Differences in ethnicity were also found. When buying food, Caucasian participants were the least concerned with nutrition and safety of food. Caucasians also report consuming higher levels of fat, cholesterol, and sugar and believed that a person should consume more servings from the food groups in the Food Guide Pyramid than both African-American and Hispanic participants. Caucasian participants also reported consuming fiber through fruit and/or vegetable peels more frequently and also consuming more low fat/low calorie food products than both African-American and Hispanic participants. African-American participants reported less healthy food preparation practices, such as trimming the fat off meat, than both Caucasians and Hispanics. Finally, contrasts across BMI groups indicated that the reported level of fat, cholesterol, and calories consumed increased with increasing BMI. The main effects are summarized in Table 3 (all p-values b.001). 3. Discussion The current study sought to shorten and condense the DHKS. This goal was accomplished using exploratory and confirmatory factor analyses, which indicated that the DHKS could be abbreviated to a 50-item, 12-factor scale assessing specific domains of dietary attitudes and behaviors.1 To demonstrate the utility of the resulting short-form DHKS, the factors were then analyzed as a function of demographic, behavioral, and body mass variables. Results from this study indicate that dietary attitudes, beliefs, and practices differ as a function of sex, diet, ethnic group, and BMI group. Compared to men and non-dieters, women and dieters reported needing greater numbers of servings from the food guide pyramid to be in accordance with current recommendations, and compared to non-dieters, dieters reported consuming junk food at a lower frequency, a stronger belief in the importance of healthy eating, and using food labels more often when purchasing food. These findings are consistent with previous research (Jeffery & French, 1996; Neumark-Sztainer et al., 2000). Taken together, these findings indicate that knowledge of healthy eating practices is associated with intentional attempts to modify dietary intake. Overall, results of the multivariate analyses were consistent with previous research (Kumanyika & Krebs-Smith, 2001). We believe there are several potential uses for the short-form DHKS. For example, the 12 factor groupings could be used individually to examine knowledge of particular constructs (e.g., buse of food labels). Secondly, subsections of the DHKS could be useful for specific target populations. For instance, the bPerceived Value of Healthy EatingQ items could be administered to men and women from various age groups and geographical locations to identify potential barriers to adopting healthy-eating behaviors. Third, the entire short-form could be used to examine current trends across various demographic groups in the U.S. related to dietary knowledge, beliefs, and practices in a more time-efficient manner than the original version. Our study has some limitations. First, we relied on self-reported height and weight data to calculate BMI. The accuracy of self-reported data has been investigated thoroughly and results indicate that the use of self-reported height and weight data is adequate when measured height and weight data are not available (Bowman & Delucia, 1992; Marietta, Welshimer & Anderson, 1999; Rowland, 1990). Second, the definition of dieting status used in this study 1
Requests for a copy of the measure should be sent to Donald A. Williamson, Ph.D. Pennington Biomedical Research Center 6400 Perkins Road, Baton Rouge LA, 70808.
242
E.E. York-Crowe et al. / Eating Behaviors 7 (2006) 235–242
was limited to individuals who endorsed current dieting to lose weight, reduce cholesterol and/or salt, and to increase dietary fiber and did not include individuals who were trying to lose weight by other means. Third, dietary recall data were not compared to self-reported intake on the DHKS; however, there is a well-known trend of over- and underreporting in dietary data, although the difference between self-reported data on the interview and self-reported intake data was not compared. Another important point to recognize is that when looking at the factor structure and item loadings, the question format may relate to the factor solution. However, not all items from the same block of questions loaded together, or at all, so it is the opinion of the authors that although the question format is important to consider, it is unlikely that it was the sole determinant of the factor structure on the short form DHKS. Finally, the format of the questions on the DHKS does not currently allow for an overall total score, because some of the questions are open-ended. However, devising a simple scoring system is a viable option, as demonstrated by Obayashi et al., who examined the reliability and validity of the scales of an older version of the DHKS (Obayashi, Bianchi & Song, 2003). Despite these shortcomings, the utilization of part or all of the short-from DHKS derived from this study could be useful to examine current dietary knowledge and practices for population-based research (e.g., examination of trends, sociodemographics, and practices of adults in the U.S.). The development and use of assessment instruments, such as the short-form DHKS, could be especially useful for identifying trends and potential barriers to the adoption of healthy eating habits to lose weight or to modify specific features of an individual’s diet. References Bowman, R. L, & Delucia, J. L. (1992). Accuracy of self-reported weight: A meta-analysis. Behavior Therapy, 23, 637 – 655. Davis, C. A., Britten, P., & Myers, E. F. (2001). Past, present, and future of the food guide pyramid. Journal of the American Dietetic Association, 101, 881 – 885. Jeffery, R. W., & French, S. A. (1996). Socioeconomic status and weight control practices among 20- to 45-year-old women. American Journal of Public Health, 86, 1005 – 1010. Kuchler, F., & Lin, B. H. (2002). The influence of individual choices and attitudes on adiposity. International Journal of Obesity and Related Metabolic Disorders, 26, 1017 – 1022. Kumanyika, S. K., & Krebs-Smith, S. M. (2001). Preventive nutrition issues in ethnic and socioeconomic groups in the United States. Nutrition, 18, 325 – 356. Marietta, A. B., Welshimer, K. J., & Anderson, S. L. (1999). Knowledge, attitudes, and behaviors of college students regarding the 1990 Nutrition Labeling Education Act food labels. Journal of the American Dietetic Association, 99, 445 – 449. Neumark-Sztainer, D., Rock, C. L., Thornquist, M. D., Cheskin, L. J., Neuhouser, M. L., & Barnett, M. J. (2000). Weight-control behaviors among adults and adolescents: Associations with dietary intake. Preventive Medicine, 30, 381 – 391. Obayashi, S., Bianchi, L. J., & Song, W. O. (2003). Reliability and validity of nutrition knowledge, social–psychological factors, and food label use scales from the 1995 Diet and Health Knowledge Survey. Journal of Nutrition Education and Behavior, 35, 83 – 92. Rowland, M. (1990). Self reported weight and height. American Journal of Clinical Nutrition, 52, 1125 – 1133. U.S. Department of Agriculture, Agricultural Research Service. (2000). Popular diets: A scientific review. Washington, DC7 Office of Research, Education, and Economics. U.S. Departments of Agriculture and Health and Human Services (USDA/DHHS). (1990). Nutrition and your health, dietary guidelines for Americans (3rd ed.). Home and Garden Bulletin No. 232. World Health Organization. (2000). Obesity: Preventing and managing the global epidemic. Report of a WHO consultation, vol. 894 (pp. 1 – 253). Geneva7 World Health Organization Technical Report Series. York-Crowe, E. E. (2002). Who’s not dieting in America and who should be? Results from the 1994–1996 Diet and Health Knowledge Survey (DHKS 1994–1996). Unpublished master’s thesis, Louisiana State University.
The diet and health knowledge survey: Development of a short interview format Emily E. York-Crowe *, Marney A. White, Sahasporn Paeratakul, Donald A. Williamson Louisiana State University, Yale Psychiatric Research, and Pennington Biomedical Research Center, United States Received 29 August 2005; accepted 29 August 2005
Abstract The primary aim of this study was to create a condensed short-form version of the structured interview named the Diet and Health Knowledge Survey (DHKS 1994–1996) via factor analysis. Data from 5765 participants of the 1994–1996 Diet and Health Knowledge Survey were used in the factor analysis. Appropriate sampling weights were used in the multivariate analyses (n = 5233) that compared subgroups (i.e., ethnic group, sex, etc.) on the subscales of the DHKS. A 50-item, 12-factor condensed version of the 149-item original DHKS was derived from the analyses. Multivariate analyses, using age and education as covariates, indicated significant differences in dietary beliefs, practices, and behaviors for individuals of different ethnic groups, sex, dieting status, and BMI. Based on our analyses, the condensed short-form version of the Diet and Health Knowledge Survey appears to be a convenient and efficient tool, used in sections or in its entirety, for examining various dietary practices and beliefs of adults. The results from analysis of the short-form DHKS indicate that although a very large proportion of adults in the U.S. are overweight or obese, only a small proportion report current dieting. Examination of the DHKS indicates that although adults varied in their dietary beliefs and practices, dieters tended to report healthier dietary practices and attitudes than non-dieters. BMI level was positively related to self-reported amount of fat, cholesterol, and calories consumed. D 2005 Elsevier Ltd. All rights reserved. Keywords: Diet; Dietary knowledge; Dietary practice; Dietary attitudes; Obesity; Factor analysis; Diet and Health Knowledge Survey
There are only a few measures that can be used to examine dietary knowledge and practices. One such measure is the Diet and Health Knowledge Survey, an interview conducted by the U.S. Department of Agriculture in 1994 through 1996. This interview includes (a) self-perceptions of the adequacy of food intake; (b) dietary and health knowledge; (c) perceived importance of certain nutrients; (d) eating behaviors related to fat intake and food safety; (e) self-perceptions of weight; (f) knowledge about cholesterol and fat sources; (g) use of food labels; (h) accuracy and confidence in the use of food labels; and (i) frequency of using food labels. The content of the DHKS survey was shaped around the need for data concerning the population’s knowledge and attitudes about the Dietary Guidelines for Americans (USDA/DHHS, 1990), fat intake as related to dietary behavior, and questions surrounding the use and understanding of food labels. Kuchler and Lin (2002) examined the Diet and Health Knowledge Survey, and found that the choices and attitudes of the surveyed individuals were related to BMI and adiposity, although the factors affecting BMI * Corresponding author. Tel.: +1 919 699 2929, +1 919 572 9496. E-mail address: [email protected] (E.E. York-Crowe). 1471-0153/$ - see front matter D 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.eatbeh.2005.08.010
236
E.E. York-Crowe et al. / Eating Behaviors 7 (2006) 235–242
differed systematically by gender. Unfortunately, this interview is not extremely time efficient for research purposes (35 min), furthermore, each question was treated as an independent datum. There is a need for a shortened version of the DHKS that develops subscales so that the interview can be used with all or a subset of the questions of the original version of the DHKS. Such a tool could be used to gather important information for dietary and obesity research and educational purposes. No study to date has examined the dietary knowledge, beliefs, and practices of such a large (and recent) representative sample of U.S. adults as the DHKS. Based upon these considerations, we believe that the DHKS is an ideal interview to convert into a validated short-from version. To demonstrate the utility of the DHKS for investigating specific trends in dietary attitudes and behaviors, we conducted factor analyses on the most current version (1994–1996) with the goal of creating a more condensed, more efficient, short-form version of the DHKS. Based on the subscales identified by the factor analysis, we then examined the dietary knowledge and practices of a representative sample of U.S. adults as a function of a variety of demographic variables. For the purposes of this study, factors (subscale scores) were examined as a function of ethnic group, sex, BMI, dieting status, and age. 1. Method The 1994–1996 Diet and Health Knowledge Survey (DHKS 1994–1996) was a telephone follow-up interview to the Continuing Survey of Food Intakes by Individuals (CSFII). The interview was conducted in a nationally representative sample of adults in the United States using a stratified, multistage, area probability sampling design. The details of survey design and methods are available elsewhere (USDA: ARS, 2000). The DHKS utilized information on dietary and health knowledge and attitudes from individuals 20 years of age and older who participated in the CSFII. The original 1994–1996 DHKS was comprised of 149 items assessing dietary practices, attitudes, and knowledge.
Table 1 Sociodemographics of population used in factor analysis Variable Sex Male Female Age 20–40 41–59 60 and over Ethnic group Caucasian African-American Hispanic Household income(pct pov) 0–130% 131–350% N350% Education V High school NHigh school BMI group b20 20–25 25–30 30–40 N40 Dieting status No Yes
Sample size
% of sample
2897 2868
50.3 49.7
1852 1952 1961
32.1 33.9 33.3
4245 621 425
80.2 11.7 8.0
1481 2217 2067
25.7 38.5 35.9
3750 1952
55.8 44.2
380 2047 2060 1031 247
6.6 35.5 35.7 17.9 4.3
4764 1000
82.6 17.3
E.E. York-Crowe et al. / Eating Behaviors 7 (2006) 235–242
237
1.1. Participants The sample used in the factor analyses consisted of 5765 individuals. The sample was divided in half using a random seed, and the exploratory factor analysis was conducted on the first half of the sample. To cross-validate the confirmatory factor analysis, the remaining half of the sample was again divided in half with a random seed. The measurement model was then analyzed twice to further test the stability of the factor structure (i.e., two confirmatory factor analyses were conducted). For the multivariate analyses, missing data were imputed based on estimation using the mean for each variable, within ethnic and sex groups. Those individuals who did not fit into the ethnic and sex grouping were excluded. Alaskan natives, Asians, Pacific Islanders, and American Indians (total n = 364) were not present in adequate number to provide a reliable estimate and were excluded, resulting in a final sample of 5233 individuals. 1.2. Variables of interest Variables examined in this study included 1) ethnic group, 2) sex, 3) BMI (from self-reported height and weight), 4) dieting status, 5) age, and 6) education level (see Table 1). The sample was divided into 3 categories of racial/ethnic groups: Caucasian, African-American, and Hispanic. Body Mass index (BMI) was derived from selfreported height without shoes (in feet and inches) and weight data (in pounds). BMI was then divided into 5 groups based on the World Health Organization’s classification system (WHO, 2000): V 19.9; 20–24.9; 25–29.9; 30–39.9; z 40 kg/m2. According to these criteria, overweight was defined as having a BMI between 25 and 29.9 kg/m2 and obese was defined as having a BMI above 30 kg/m2. Dieting status was determined by endorsement of any of the following questions: 1) bAre you on a weight loss or low calorie diet?Q; 2) bAre you on a low fat or cholesterol diet?Q; 3) bAre you on a low salt or sodium diet?Q; 4) bAre you on a high fiber diet?Q; or 5) bAre you on a diabetic diet?Q. Education level, ranging from 0 to 17, was defined as the number of years of schooling completed. Responses ranged from bnever attended school or kindergarten onlyQ (0) to bfive or more years of collegeQ(17). 2. Results 2.1. Factor analysis Table 1 shows the characteristics of the sample used in the factor analysis. To conduct both exploratory and confirmatory factor analyses, the sample was divided in half (n = 2880; n = 2885) using a random selection of cases. Items on the survey were analyzed via principle component analysis using an orthogonal (VARIMAX) rotation. The Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy and Bartlett’s test of sphericity were calculated to assess the appropriateness of the data for factor analysis. The KMO index was .89 and the Bartlett’s test of sphericity was highly significant ( p b .001), indicating that the data were appropriate for analysis. Questions of the DHKS were retained if they had a factor loading of .35 or higher and if they loaded on only one factor. This procedure yielded a 12-factor structure composed of 53 items that accounted for 53.74% of the total variance. Only factors that had eigen values greater than 1.0 were retained. Also, a scree plot of eigenvalues confirmed the presence of a 12-factor solution for the factor analysis. To cross-validate the Confirmatory factor analysis, the remaining half of the sample (n = 2885) was again divided in half based on a random seed. The measurement model was then analyzed twice (n = 1446; n = 1439) to further test the stability of the factor structure (i.e., two confirmatory factor analyses were computed). In both confirmatory analyses, the 12-factor model was stable. The Root Mean Square Error of Approximation (RMSEA) was adequate for both (.046 and .043 respectively). Further, the Non-Normed Fit Indices (NNFIs) were acceptable (.84 and .83), as were the Comparative Fit Indices (CFIs) (.86 and .85). RMSEA values of .08 or less and NNFIs and CFIs approaching .90 are considered desirable. Analysis of the confirmatory factor loadings indicated that 3 questions did not load highly on the predicted factor (i.e., loaded less than .35 in one or both of the confirmatory analyses). Therefore, those items were dropped, yielding a 50-question final interview. Table 2 lists the individual questions and factor loadings for the three factor analyses for the 12 factors that were retained.
238
Table 2 Factor loadings for the exploratory and confirmatory factor analyses Factor 1
Perceived value of healthy eating Eating low-fat diet Eating adequate fiber Diet low in cholesterol Diet high in fruit and vegetables Diet low in saturated fat Maintaining healthy weight Moderating calories Moderating salt Eating variety of foods Plenty of bread, cereal, rice, pasta
0.71; 0.70; 0.70; 0.69; 0.69; 0.66; 0.62; 0.54; 0.54; 0.44;
Use of food labels Never used/never seen Use nutrition panel Use list of ingredients Use health/nutrients benefits label Use serving size Use bshort phrasesQ (e.g., light) Consumption of low-fat/low-calorie foods Eat/use low-fat cheese Eat/use low calorie salad dressing Eat/use 1% milk Eat/use ice milk/frozen yogurt Eat/use lower fat luncheon meat Knowledge of food guide pyramid servings Fruit group Vegetable group Bread, cereal, rice, pasta group Milk, yogurt, cheese group Meat, poultry, fish, dry beans, eggs Perception of adequacy of energy intake Fat Saturated fat
0.82; 0.67; 0.78; 0.69; 0.76; 0.56; 0.58; 0.48; 0.51; 0.36;
2
3
4
5
0.81 0.67 0.76 0.66 0.74 0.64 0.59 0.54 0.47 0.38
0.78; 0.72; 0.73 0.77; 0.84; 0.86 0.76; 0.79; 0.80 0.75; 0.73; 0.72 0.72; 0.70; 0.72 0.70; 0.75; 0.75
0.78; 0.70; 0.68; 0.60; 0.58;
0.75; 0.71; 0.59; 0.62; 0.63;
6
E.E. York-Crowe et al. / Eating Behaviors 7 (2006) 235–242
Factor labels and item numbers
0.72 0.73 0.63 0.65 0.61
0.75; 0.74; 0.72; 0.70; 0.60;
0.76; 0.76; 0.60; 0.54; 0.48;
0.71 0.76 0.64 0.53 0.42
0.82; 0.78; 0.78 0.80; 0.80; 0.79
Cholesterol Calories
0.67; 0.65; 0.59 0.61; 0.47; 0.47
Added fats Butter/margarine to vegetables Butter/marg/sour cream to potatoes Cheese/creamy sauce to vegetables Butter/margarine to breads/muffins Factor interpretations
0.78; 0.61; 0.58 0.66; 0.69; 0.65 0.58; 0.35; 0.36 0.56; 0.52; 0.55 Factor 7 0.65; 0.50; 0.63; 0.54;
Values affecting food purchasing decisions How safe food is to eat Nutrition Perception of nutrient intake adequacy Iron Fiber Calcium Consumption of fiber through fruit and vegetable peels Eat vegetable peels Eat fruit peels Avoidance of extra fat Trim fat off meat Remove skin from chicken Eat fried chicken Consumption of junk food Cakes, cookies, donuts Potato/corn chips
8
9
10
11
12
0.51 0.49 0.63 0.50
E.E. York-Crowe et al. / Eating Behaviors 7 (2006) 235–242
Awareness of consequences of poor nutrition Eating too much fat 0.69; Being overweight 0.66; Eating too much cholesterol 0.66; Eating too much salt/sodium 0.65;
0.57; 0.51; 0.45 0.50; 0.78; 0.83
0.71; 0.53; 0.46 0.60; 0.44; 0.38 0.58; 0.45; 0.47
0.83; 0.57; 0.72 0.83; 0.86; 0.73
0.72; 0.46; 0.37 0.72; 0.64; 0.59 0.38; 0.52; 0.53
0.71; 0.42; 0.46 0.70; 0.76; 0.77
The first column within each cell reflects the exploratory factor loadings; the factor loadings from both confirmatory analyses appear in the second and third columns of each cell.
239
240
E.E. York-Crowe et al. / Eating Behaviors 7 (2006) 235–242
The first factor, Perceived Value of Healthy Eating, consisted of 10 items measuring the extent to which participants endorsed the importance of moderating food intake and eating specific types of foods or nutrients (e.g., low saturated fat, high fiber). The second factor consisted of 6 items and assessed the extent to which participants reported using various features of food labels in making dietary selections; this factor was labeled Use of Food Labels. The third factor, Consumption of Low-fat/Low-Calorie foods, consisted of 5 items measuring participants’ use of low-fat alternatives in making food selections. Factor 4, Knowledge of Food Guide Pyramid Servings, asked participants to estimate the number of recommended daily servings of food groups classified according to the Food Guide Pyramid guidelines (Davis, Britten & Myers, 2001). Factor 4 consists of 5 items. Factor 5 asks participants to rate their belief in the extent to which their diet conforms to healthy amounts of calories, cholesterol, fat, and saturated fat intake. Factor 5 was labeled Perception of Energy Intake Adequacy. The sixth factor, Added Fats, assesses 4 eating behaviors that increase the amount of fat intake, such as adding butter to foods. Factor 7, Awareness of Consequences of Poor Nutrition, consists of 4 items assessing participants’ knowledge of negative health consequences associated with poor eating habits. The eighth factor, Values affecting Food Purchasing Decisions, consists of 2 items tapping food-purchasing decisions, and asks participants to rate the importance of nutrition and safety when making food selections. Factor 9, Perception of Nutrient Intake Adequacy, consisted of 3 items asking participants to rate the extent to which they believe they are consuming appropriate amounts of iron, fiber, and calcium. Factor 10 consisted of 2 items measuring the Consumption of Fiber through Fruit and Vegetable Peels. Factor 11 consisted of 3 items that assessed dietary practices that reduce the amount of fat intake, such as trimming skin from poultry and fat from meat; this factor was labeled Avoidance of Extra Fat. The final factor retained 2 items that asked participants to estimate the frequency of consuming sweets and chips and was labeled Consumption of Junk Food. Table 3 Summary of main effects Effect
Multivariate effect1
Sex
F(12, 5162) = 4.28
Subscale
Estimated marginal means Female 5.02b
F(1, 5173) = 25.49, p b .001
12.01a
13.06b
F(1, 5173) = 5.62, p b .020
Non-dieter 34.55b 3.52b 11.80b 4.46b
Dieter 36.06a 3.09a 10.08a 4.27a
F(1, 5173) = 10.84, p b .002 F(1, 5173) = 12.45, p b .001 F(1, 5173) = 13.80, p b .001 F(1,5173) = 5.08, p b .030
12.10b
12.99a
F(1, 5173) = 4.09, p b .05
Hispanic
7.36a 9.49a
AfricanAmerican 7.63b,c 9.01b,c
7.68c 7.68c
F(2, 5173) = 6.89, p b .001 F(2, 5173) = 10.66, p b .001
3.94a
4.77b
3.99a
F(2, 5173) = 16.31, p b .001
13.42b
12.21c
F(2, 5173) = 13.35, p b .001
6.23 b 11.90b,c
5.73c 12.22c
F(2, 5173) = 5.08, p b .007 F(2, 5173) = 7.04, p b .002
Perception of adequacy of nutrient intake Knowledge of Food Guide Pyramid servings Diet
F(12, 5162) = 2.98 Perceived value of healthy eating Consumption of junk food Use of food labels Aware of negative health consequences of poor nutrition Knowledge of Food Guide Pyramid servings
Ethnic group F(24, 10,324) = 5.80
Caucasian Food purchasing decision rules Perception of energy intake adequacy Consume fiber by fruit and/or vegetable peels Consumption of low fat/low calorie foods Avoidance of extra fat Knowledge of Food Guide Pyramid servings
BMI group
F(48, 19,887) = 2.55 Perception of energy intake adequacy
Univariate effect
Male 5.59a
11.93
a
5.89 a 13.41a
b 20 kg/m 2 20–25 kg/m 2 25–30 kg/m 2 30–40 kg/m 2 N40 kg/m2 8.21a 8.48b 9.21c 9.38d 10.10e
1 All multivariate effects were significant, p b .001. Different subscripts (a, b, c) denote significant group differences. Results from all univariate follow-up analyses are reported and discussed elsewhere (York-Crowe, 2002).
E.E. York-Crowe et al. / Eating Behaviors 7 (2006) 235–242
241
2.2. Examination of dietary practices Multivariate analysis of variance (MANCOVA) using the subscales as dependent variables revealed age and education were significant covariates ( F(12, 5162) = 90.10, p b .001; F(12, 5162) = 54.98, p b .001). Based on results of the MANCOVA, ANCOVAs were performed using age and education as covariates for each of the individual subscales. Ethnic group, sex, BMI, and dieting status were entered as independent variables, all of which were significant (all p-values b .001). In terms of gender, men reported consuming higher, more adequate, levels of iron, fiber, and calcium than women; women, believed that a person should eat more servings from the food groups in the Food Guide Pyramid than did men. In terms of dieting status, individuals who reported current dieting (dieters) were more likely to believe in the importance of healthy food, report consuming less junk food, using food labels more often when purchasing food, and endorse a belief that a person should eat more servings from the food groups in the Food Guide Pyramid. Dieters also reported a higher awareness of the consequences of the following behaviors: eating too much fat; eating too much salt or sodium; eating too much cholesterol; being overweight. Differences in ethnicity were also found. When buying food, Caucasian participants were the least concerned with nutrition and safety of food. Caucasians also report consuming higher levels of fat, cholesterol, and sugar and believed that a person should consume more servings from the food groups in the Food Guide Pyramid than both African-American and Hispanic participants. Caucasian participants also reported consuming fiber through fruit and/or vegetable peels more frequently and also consuming more low fat/low calorie food products than both African-American and Hispanic participants. African-American participants reported less healthy food preparation practices, such as trimming the fat off meat, than both Caucasians and Hispanics. Finally, contrasts across BMI groups indicated that the reported level of fat, cholesterol, and calories consumed increased with increasing BMI. The main effects are summarized in Table 3 (all p-values b.001). 3. Discussion The current study sought to shorten and condense the DHKS. This goal was accomplished using exploratory and confirmatory factor analyses, which indicated that the DHKS could be abbreviated to a 50-item, 12-factor scale assessing specific domains of dietary attitudes and behaviors.1 To demonstrate the utility of the resulting short-form DHKS, the factors were then analyzed as a function of demographic, behavioral, and body mass variables. Results from this study indicate that dietary attitudes, beliefs, and practices differ as a function of sex, diet, ethnic group, and BMI group. Compared to men and non-dieters, women and dieters reported needing greater numbers of servings from the food guide pyramid to be in accordance with current recommendations, and compared to non-dieters, dieters reported consuming junk food at a lower frequency, a stronger belief in the importance of healthy eating, and using food labels more often when purchasing food. These findings are consistent with previous research (Jeffery & French, 1996; Neumark-Sztainer et al., 2000). Taken together, these findings indicate that knowledge of healthy eating practices is associated with intentional attempts to modify dietary intake. Overall, results of the multivariate analyses were consistent with previous research (Kumanyika & Krebs-Smith, 2001). We believe there are several potential uses for the short-form DHKS. For example, the 12 factor groupings could be used individually to examine knowledge of particular constructs (e.g., buse of food labels). Secondly, subsections of the DHKS could be useful for specific target populations. For instance, the bPerceived Value of Healthy EatingQ items could be administered to men and women from various age groups and geographical locations to identify potential barriers to adopting healthy-eating behaviors. Third, the entire short-form could be used to examine current trends across various demographic groups in the U.S. related to dietary knowledge, beliefs, and practices in a more time-efficient manner than the original version. Our study has some limitations. First, we relied on self-reported height and weight data to calculate BMI. The accuracy of self-reported data has been investigated thoroughly and results indicate that the use of self-reported height and weight data is adequate when measured height and weight data are not available (Bowman & Delucia, 1992; Marietta, Welshimer & Anderson, 1999; Rowland, 1990). Second, the definition of dieting status used in this study 1
Requests for a copy of the measure should be sent to Donald A. Williamson, Ph.D. Pennington Biomedical Research Center 6400 Perkins Road, Baton Rouge LA, 70808.
242
E.E. York-Crowe et al. / Eating Behaviors 7 (2006) 235–242
was limited to individuals who endorsed current dieting to lose weight, reduce cholesterol and/or salt, and to increase dietary fiber and did not include individuals who were trying to lose weight by other means. Third, dietary recall data were not compared to self-reported intake on the DHKS; however, there is a well-known trend of over- and underreporting in dietary data, although the difference between self-reported data on the interview and self-reported intake data was not compared. Another important point to recognize is that when looking at the factor structure and item loadings, the question format may relate to the factor solution. However, not all items from the same block of questions loaded together, or at all, so it is the opinion of the authors that although the question format is important to consider, it is unlikely that it was the sole determinant of the factor structure on the short form DHKS. Finally, the format of the questions on the DHKS does not currently allow for an overall total score, because some of the questions are open-ended. However, devising a simple scoring system is a viable option, as demonstrated by Obayashi et al., who examined the reliability and validity of the scales of an older version of the DHKS (Obayashi, Bianchi & Song, 2003). Despite these shortcomings, the utilization of part or all of the short-from DHKS derived from this study could be useful to examine current dietary knowledge and practices for population-based research (e.g., examination of trends, sociodemographics, and practices of adults in the U.S.). The development and use of assessment instruments, such as the short-form DHKS, could be especially useful for identifying trends and potential barriers to the adoption of healthy eating habits to lose weight or to modify specific features of an individual’s diet. References Bowman, R. L, & Delucia, J. L. (1992). Accuracy of self-reported weight: A meta-analysis. Behavior Therapy, 23, 637 – 655. Davis, C. A., Britten, P., & Myers, E. F. (2001). Past, present, and future of the food guide pyramid. Journal of the American Dietetic Association, 101, 881 – 885. Jeffery, R. W., & French, S. A. (1996). Socioeconomic status and weight control practices among 20- to 45-year-old women. American Journal of Public Health, 86, 1005 – 1010. Kuchler, F., & Lin, B. H. (2002). The influence of individual choices and attitudes on adiposity. International Journal of Obesity and Related Metabolic Disorders, 26, 1017 – 1022. Kumanyika, S. K., & Krebs-Smith, S. M. (2001). Preventive nutrition issues in ethnic and socioeconomic groups in the United States. Nutrition, 18, 325 – 356. Marietta, A. B., Welshimer, K. J., & Anderson, S. L. (1999). Knowledge, attitudes, and behaviors of college students regarding the 1990 Nutrition Labeling Education Act food labels. Journal of the American Dietetic Association, 99, 445 – 449. Neumark-Sztainer, D., Rock, C. L., Thornquist, M. D., Cheskin, L. J., Neuhouser, M. L., & Barnett, M. J. (2000). Weight-control behaviors among adults and adolescents: Associations with dietary intake. Preventive Medicine, 30, 381 – 391. Obayashi, S., Bianchi, L. J., & Song, W. O. (2003). Reliability and validity of nutrition knowledge, social–psychological factors, and food label use scales from the 1995 Diet and Health Knowledge Survey. Journal of Nutrition Education and Behavior, 35, 83 – 92. Rowland, M. (1990). Self reported weight and height. American Journal of Clinical Nutrition, 52, 1125 – 1133. U.S. Department of Agriculture, Agricultural Research Service. (2000). Popular diets: A scientific review. Washington, DC7 Office of Research, Education, and Economics. U.S. Departments of Agriculture and Health and Human Services (USDA/DHHS). (1990). Nutrition and your health, dietary guidelines for Americans (3rd ed.). Home and Garden Bulletin No. 232. World Health Organization. (2000). Obesity: Preventing and managing the global epidemic. Report of a WHO consultation, vol. 894 (pp. 1 – 253). Geneva7 World Health Organization Technical Report Series. York-Crowe, E. E. (2002). Who’s not dieting in America and who should be? Results from the 1994–1996 Diet and Health Knowledge Survey (DHKS 1994–1996). Unpublished master’s thesis, Louisiana State University.