reactive model: A new approach to examining eating decisions among female college dieters and nondieters

Eating Behaviors 23 (2016) 33–40

Contents lists available at ScienceDirect

Eating Behaviors

The reasoned/reactive model: A new approach to examining eating decisions among female college dieters and nondieters Holly Ruhl ⁎, Shayla C. Holub, Elaine A. Dolan The University of Texas at Dallas, 800 W. Campbell Rd., GR 41, Richardson, TX 75080, USA

a r t i c l e

i n f o

Article history: Received 30 January 2016 Received in revised form 27 June 2016 Accepted 13 July 2016 Available online 15 July 2016 Keywords: Dietary restraint Theory of reasoned action Prototype/willingness model Nutrition knowledge Affective associations Reasoned/reactive model

a b s t r a c t Female college students are prone to unhealthy eating patterns that can impact long-term health. This study examined female students' healthy and unhealthy eating behaviors with three decision-making models. Specifically, the theory of reasoned action, prototype/willingness model, and new reasoned/reactive model were compared to determine how reasoned (logical) and reactive (impulsive) factors relate to dietary decisions. Females (N = 583, Mage = 20.89 years) completed measures on reasoned cognitions about foods (attitudes, subjective norms, nutrition knowledge, intentions to eat foods), reactive cognitions about foods (prototypes, affect, willingness to eat foods), dieting, and food consumption. Structural equation modeling (SEM) revealed the new reasoned/reactive model to be the preeminent model for examining eating behaviors. This model showed that attitudes were related to intentions and willingness to eat healthy and unhealthy foods. Affect was related to willingness to eat healthy and unhealthy foods, whereas nutrition knowledge was related to intentions and willingness to eat healthy foods only. Intentions and willingness were related to healthy and unhealthy food consumption. Dieting status played a moderating role in the model and revealed mean-level differences between dieters and nondieters. This study highlights the importance of specific factors in relation to female students' eating decisions and unveils a comprehensive model for examining health behaviors. © 2016 Elsevier Ltd. All rights reserved.

1. Introduction Many college students consume more foods with added fats and sugars and fewer fruits and vegetables than recommended by dietary guidelines (Anding, Suminski, & Boss, 2001; Racette, Deusinger, Strube, Highstein, & Deusinger, 2005). Eating patterns formed during college may relate to obesity, diabetes, and heart disease (Mannino, Lee, Mitchell, Smiciklas-Wright, & Birch, 2004). Because females diet more and hold more negative cognitions about eating and weight than males, it is important to investigate factors related to eating decisions in order to improve diet and health in college-age women (Kelly-Weeder, 2011; Tapper & Pothos, 2010). This study investigated the utility of three decision-making models for examining female students' eating behaviors, differences in eating decisions among female dieters and nondieters, and the relative importance of specific factors in relation to intentions and willingness to eat foods. 1.1. Examining eating behaviors with decision-making models The theory of reasoned action (TRA) posits that a behavior (e.g., healthy eating) is predicted by reasoned intentions to engage in the ⁎ Corresponding author at: Behavioral and Brain Sciences, The University of Texas at Dallas, 800 W. Campbell Rd., GR 41, Richardson, TX 75080, USA. E-mail address: [email protected] (H. Ruhl).

http://dx.doi.org/10.1016/j.eatbeh.2016.07.011 1471-0153/© 2016 Elsevier Ltd. All rights reserved.

behavior (Fig. 1; Fishbein & Ajzen, 1975). Intentions are influenced by subjective norms and attitudes. Subjective norms are perceptions of social pressure (e.g., “My family thinks I should eat fruit”). Attitudes are thoughts about a behavior (e.g., “I think eating fruit is beneficial”). Although widely used, the TRA is simplistic and additional factors may contribute to eating behaviors (Brewer, Blake, Rankin, & Douglass, 1999; Dohnke, Steinhilber, & Fuchs, 2015; Shepherd & Stockley, 1987). The prototype/willingness (P/W) model extends the TRA by positing that decisions are also based on reactive factors, namely behavioral willingness and prototypes (Fig. 1; Gibbons, Gerrard, Blanton, & Russell, 1998). Specifically, in situations that encourage unhealthy eating, individuals may show spontaneous willingness to eat certain foods, despite having no rational intention to eat them. The model also posits that individuals hold prototypes about people who engage in specific behaviors (e.g., the stereotypical unhealthy eater), which impact individuals' own willingness to engage in that behavior. For example, people who hold negative prototypes of unhealthy eaters will be less willing to eat unhealthy foods (Gibbons et al., 1998). Although some research has examined prototypes and willingness separately in relation to eating (Gerrits et al., 2010; Ohtomo, Hirose, & Midden, 2011), very little research has holistically examined eating with the P/W model (Dohnke et al., 2015). Although the TRA assumes that behaviors are reasoned (Fishbein & Ajzen, 1975), it ignores one highly reasoned component: knowledge. Similarly, the P/W model acknowledges reactivity (Gibbons et al.,

34

H. Ruhl et al. / Eating Behaviors 23 (2016) 33–40

Fig. 1. Theoretical decision-making model for examining eating behaviors. Rectangles are observed variables, circles are residual terms, single-headed arrows are regression paths, and double-headed arrows are correlations. White observed variables originate from the theory of reasoned action (TRA), striped variables originate from the prototype/willingness (P/W) model, and solid variables originate from the reasoned/reactive (R/R) model. Because prototypes are reactive, they do not theoretically influence intentions to engage in a behavior.

1998) but ignores one fundamentally reactive factor: affect. Thus, we extended these models to create a new theoretical model that includes knowledge and affective associations—the reasoned/reactive (R/R) model (Fig. 1). 1.2. Factors related to intentions and willingness Research supports examining reasoned factors (attitudes, subjective norms, nutrition knowledge) in relation to intentions and willingness to eat foods. Attitudes about foods are related to intentions to eat foods (Brewer et al., 1999; Shepherd & Stockley, 1987). Subjective norms are also related to intentions, although inconsistently (Brewer et al., 1999; Conner, Hugh-Jones, & Berg, 2011; Shepherd & Stockley, 1987; Wong & Mullan, 2009). One study found that attitudes and subjective norms were not related to behavioral willingness, but this should be explored further (Ohtomo et al., 2011). Nutrition knowledge is related to food consumption (Dickson-Spillmann, Siegrist, & Keller, 2011; Kolodinsky, Harvey-Berino, Berlin, Johnson, & Reynolds, 2007). However, findings are mixed regarding the relationship between nutrition knowledge and intentions (Shepherd & Stockley, 1987; Shepherd & Towler, 1992). No studies have examined the relationship between nutrition knowledge and reactive willingness. Examining these reasoned factors with a comprehensive model may shed light on these inconsistent relationships. Research also supports examining reactive factors (prototypes, affective associations) in relation to intentions and willingness. Holding negative prototypes about unhealthy eaters is related to less unhealthy eating (Gerrits et al., 2010). Dohnke et al. (2015) found bivariate relationships between prototypes and willingness to eat healthy and unhealthy foods; however, in the context of the P/W model, prototypes were not related to willingness. Positive affective associations with foods are related to more consumption (Kiviniemi & Brown-Kramer, 2015; Kiviniemi & Duangdao, 2009; Walsh & Kiviniemi, 2014). Affect about foods may also be related to intentions and willingness, but evidence is lacking (Kiviniemi & Brown-Kramer, 2015). Examining these reactive factors may prove important in understanding intentions and willingness to eat. 1.3. Dieters' eating behaviors Dieters have more negative attitudes and affect about unhealthy foods and more nutrition knowledge than nondieters (Irmak, Vallen, & Rosen Robinson, 2011; Maas, Keijsers, Rinck, Tanis, & Becker, 2015; York-Crowe, White, Paeratakul, & Williamson, 2006). Research has yet to find relationships between dieting and subjective norms, prototypes,

intentions, or willingness (Gerrits et al., 2010). Further, research has yet to examine whether dieting moderates the relationships between eating cognitions and behaviors. Moderation would indicate how the decision-making processes of dieters and nondieters differ, offering insight for interventions. Therefore, this study will examine mean differences between female dieters and nondieters and whether dieting moderates relations between eating cognitions and behaviors. 1.4. The current study This study had three primary aims. Aim 1 was to compare the utility of the TRA, P/W, and R/R models in examining healthy and unhealthy eating behaviors among female college students. Aim 2 was to examine whether the relations outlined in the best model differ for female dieters and nondieters (moderation) and whether there are mean differences found for these two groups. Aim 3 was to examine the relative importance of factors in the final model. 2. Method 2.1. Participants Participants were 583 female college students (276 dieters, 307 nondieters) from a public university in Texas (Mage = 20.89 years, SD = 1.85). Over one-third (39%) were Caucasian (29% Asian, 14% Hispanic, 6% African American, 6% Middle Eastern, 6% “Mixed/Other”). Approximately 6% were underweight (BMI under 18.5), 64% were normal weight (BMI of 18.5 to 24.9), 19% were overweight (BMI of 25 to 29.9), and 11% were obese (BMI over 30; Centers for Disease Control and Prevention, 2011). 2.2. Procedure Students were recruited through classes, flyers, and student organizations. They provided informed consent and completed confidential surveys on lab computers. Afterward, height and weight were measured by trained research assistants to calculate BMI (kg/m2). Participants were offered research exposure credits or entry into a raffle to win one of 25 gift cards worth $50.00. This study was approved by the university's Institutional Review Board. 2.3. Measures Items on fruits (e.g., apples, bananas, oranges) and vegetables (e.g., carrots, broccoli, squash) were used to measure cognitions about

H. Ruhl et al. / Eating Behaviors 23 (2016) 33–40

“healthy” foods (Boeing et al., 2012). Items on foods with added fats and sugars (e.g., chocolate, potato chips, ice cream) were used to measure cognitions about “unhealthy” foods (Gidding et al., 2005). Items on other food types were included to avoid priming effects (e.g., responding negatively to all unhealthy food items), but were not analyzed here. 2.3.1. Intentions Intentions were examined with items adapted from research on the TRA (Table 1; Brewer et al., 1999). Intentions to eat healthy foods were based on an average of intentions to eat fruits and vegetables (2 items). Intentions to eat unhealthy foods were based on intentions to eat foods with added fats and sugars (1 item). Higher scores indicated more intentions. 2.3.2. Behavioral willingness Willingness to eat foods in situations that might elicit unhealthy eating was examined with items adapted from research on the P/W model (Table 2; Gibbons et al., 1998). Items were averaged to assess willingness to eat healthy foods (6 items, α = 0.80) and unhealthy foods (3 items, α = 0.69). Higher scores indicated more willingness. 2.3.3. Consumption A food frequency questionnaire found to be valid and reliable for female college students in Texas assessed consumption of 195 foods over a two-month period on a scale from 1 (“Never or b1 month) to 9 (“2+ per day”; Chacko George, Milani, Hanss-Nuss, Kim, & Freeland-Graves, 2004). Scores were summed to assess overall consumption of healthy foods (54 items, α = 0.93) and unhealthy foods (28 items, α = 0.85).

Table 1 Measuring reasoned cognitions about healthy and unhealthy foods based on the TRA. Intentions Do you intend to eat the following foods in the next week? 1. Fruits (e.g., apples, bananas, oranges) 2. Vegetables (e.g., carrots, broccoli, squash) 3. Foods with Added Fats and Sugars (e.g., chocolate, potato chips, ice cream) 1. Definitely will not 2. Probably will not 3. Don’t know 4. Probably will 5. Definitely will Attitudes 1. I think that eating fruits (e.g., apples, bananas, oranges) is: 2. I think that eating vegetables (e.g., carrots, broccoli, squash) is: 3. I think that eating foods with added fats and sugars (e.g., chocolate, potato chips, ice cream) is: 1. Extremely Harmful 1. Extremely Unpleasant 1. Extremely Unnecessary 2. Somewhat Harmful 2. Somewhat Unpleasant 2. Somewhat Unnecessary 3. Slightly Harmful 3. Slightly Unpleasant 3. Slightly Unnecessary 4. Neither Harmful nor 4. Neither Unpleasant 4. Neither Unnecessary nor Beneficial nor Pleasant Necessary 5. Slightly Beneficial 5. Slightly Pleasant 5. Slightly Necessary 6. Somewhat Beneficial 6. Somewhat Pleasant 6. Somewhat Necessary 7. Extremely Beneficial 7. Extremely Pleasant 7. Extremely Necessary Subjective Norms 1. I believe that most people who are important to me think that I should eat fruits (e.g., apples, bananas, oranges). 2. I believe that most people who are important to me think that I should eat vegetables (e.g., carrots, broccoli, squash). 3. I believe that most people who are important to me think that I should eat foods with added fats and sugars (e.g., chocolate, potato chips, ice cream). 1. Strongly Disagree 2. Disagree 3. Somewhat Disagree 4. Neither Agree nor Disagree 5. Somewhat Agree 6. Agree 7. Strongly Agree

35

Table 2 Measuring reactive cognitions about healthy and unhealthy foods based on the P/W model. Willingness 1. Suppose you were presented with a situation in which you were short on time. 2. Suppose you were presented with a situation in which you were short on money. 3. Suppose you were presented with a situation in which you were with friends. How likely is it that you would eat: a. Fruits (e.g., apples, bananas, oranges) b. Vegetables (e.g., carrots, broccoli, squash) c. Foods with Added Fats and Sugars (e.g., chocolate, potato chips, ice cream) 1. Not at all likely 2. Unlikely 3. Somewhat unlikely 4. Undecided 5. Somewhat likely 6. Likely 7. Very likely Prototypes When trying to describe someone, people generally use characteristics of that person. For example, if you describe someone your age who always gets good grades, you might say that this person is smart, serious and bookish. 1. Now, if we would ask you to describe someone your age who eats fruits (e.g., apples, bananas, oranges), which characteristics would you use? 2. Now, if we would ask you to describe someone your age who eats vegetables (e.g., carrots, broccoli, squash), which characteristics would you use? 3. Now, if we would ask you to describe someone your age who eats foods with added fats and sugars (e.g., chocolate, potato chips, ice cream), which characteristics would you use? 1 Foolish Irresponsible Undisciplined Focused on present Dissatisfied Insecure Sloppy Unkempt Chubby Thinks body is unimportant Not sporty Lazy

2

3

4

5

6

7 Wise Responsible Disciplined Focused on future Satisfied Self-confident Meticulous Well-groomed Slim Thinks body is important Sporty Active

2.3.4. Attitudes Attitudes were assessed with items adapted from research on the TRA (Table 1; Wong & Mullan, 2009). These items assessed the extent to which participants think eating certain foods is pleasant, beneficial, and necessary. Items were averaged to assess overall attitudes about healthy foods (6 items, α = 0.63) and unhealthy foods (3 items, α = 0.53).

2.3.5. Subjective norms Subjective norms were assessed with items adapted from research on the TRA (Table 1; Shepherd & Stockley, 1987). Two items were averaged to assess subjective norms to eat healthy foods. One item assessed subjective norms to eat unhealthy foods.

2.3.6. Prototypes Prototypes were measured with items based on research assessing prototypes of healthy and unhealthy eaters (Table 2; Gerrits, de Ridder, de Wit, & Kuijer, 2009). Participants were asked to evaluate a prototypic person who engages in the behavior being studied (e.g., eating fruits) on 12 adjective dimensions. Ratings were averaged to assess prototypes of healthy eaters (24 items, α = 0.94) and unhealthy eaters (12 items, α = 0.91).

36

H. Ruhl et al. / Eating Behaviors 23 (2016) 33–40

2.3.8. Affective associations To measure affective associations, participants rated how they feel when they think about eating certain foods using 12 positive and 10 negative affect terms (Table 3; Aikman, Crites, & Fabrigar, 2006). Items were averaged to assess affective associations with healthy foods (44 items, α = 0.94) and unhealthy foods (22 items, α = 0.91).

Table 3 Measuring affective associations with healthy and unhealthy foods. 1. Please rate FRUITS on the scales below. Indicate how true or descriptive each statement is of your beliefs about FRUITS by circling the appropriate description. There are no right or wrong answers—We are interested in your beliefs about FRUITS. The thought of eating fruits (e.g., apples, bananas, oranges) makes me feel: 2. Please rate VEGETABLES on the scales below. Indicate how true or descriptive each statement is of your beliefs about VEGETABLES by circling the appropriate description. There are no right or wrong answers—We are interested in your beliefs about VEGETABLES. The thought of eating vegetables (e.g., carrots, broccoli, squash) makes me feel: 3. Please rate FOODS WITH ADDED FATS AND SUGARS on the scales below. Indicate how true or descriptive each statement is of your beliefs about FOODS WITH ADDED FATS AND SUGARS by circling the appropriate description. There are no right or wrong answers—We are interested in your beliefs about FOODS WITH ADDED FATS AND SUGARS. The thought of eating foods with added fats and sugars (e.g., chocolate, potato chips, ice cream) makes me feel: Lively At ease Joyful Relaxed Calm Comforted Enthusiastic Excited Refreshed Content Satisfied Rewarded

Guilty (R) Ashamed (R) Depressed (R) Concerned (R) Disturbed (R) Disgusted (R) Sluggish (R) Nauseated (R) Bored (R) Sick (R)

2.3.9. Dieting Dieting status was assessed with the restraint subscale of the Dutch Eating Behaviors Questionnaire (DEBQ; van Strien, Frijters, Bergers, & Defares, 1986). This subscale consists of 10 items (α = 0.91) on a scale from 1 (“Never”) to 5 (“Very often”), with a “N/A” option for applicable items. An example is, “Do you deliberately eat less in order not to become heavier?” Participants above the median (2.70) were classified as dieters; participants equal to or below the median were classified as nondieters based on van Strien's (1997) research. 2.4. Data analytic plan

1. Disagree strongly 2. Disagree 3. Undecided 4. Agree 5. Agree strongly

In order to compare the utility of the TRA, P/W, and R/R models for examining eating behaviors among female college students (Aim 1), structural equation modeling (SEM) was used to assess which nested model was the best fit while maintaining the most parsimony (Fig. 1). All effects were unconstrained across dieters and nondieters. The simplest framework was first analyzed, in which all parameters not in the TRA were fixed to zero (Model 1). Model 2, including willingness and prototypes, was then compared to Model 1. The better of these two models, based on a chi-square difference test, was compared to Model 3, which included nutrition knowledge and affective associations. This process was conducted separately for healthy and unhealthy foods. Aim 2 examined differences between female dieters and nondieters. To test moderation, the best unconstrained model from Aim 1 was compared to a corresponding model in which all parameters, variances, and covariances were constrained to be equal for dieters and nondieters (Model 4). If the unconstrained model was a better fit based on a chisquare difference test, this suggested differences between dieters and nondieters. In this case, each regression coefficient was tested for differences by constraining each coefficient to be equal for dieters and nondieters one by one. If the constraint significantly decreased the

Note: R indicates that item was reverse coded.

2.3.7. Nutrition knowledge Nutrition knowledge was assessed with the General Nutrition Knowledge Questionnaire (GNKQ; Parmenter & Wardle, 1999). Three subscales measured participants' knowledge of current dietary recommendations (4 items), which foods provide specific nutrients (21 items), and how to choose between foods to identify the healthiest ones (10 items). Scores were based on percentage of total items answered correctly (α = 0.82).

Table 4 Descriptive statistics and mean differences between dieters and nondieters for key study variables. Overall

Dieters

Nondieters

Min

Max

M

SD

M

SD

M

SD

t

p

Cohen's d

Healthy foods Attitudes Subjective norms Prototypes Affective associations Intentions Willingness Consumption

3.33 1.00 3.33 2.59 1.00 1.00 57.00

7.00 7.00 7.00 5.00 5.00 7.00 348.00

6.51 6.18 5.79 4.04 4.41 4.72 135.68

0.53 1.06 0.76 0.47 0.77 1.33 41.81

6.52 6.16 5.87 4.02 4.52 4.81 139.20

0.53 1.12 0.73 0.46 0.67 1.26 40.52

6.50 6.19 5.72 4.05 4.31 4.65 132.52

0.54 1.00 0.78 0.47 0.84 1.39 42.76

0.28 −0.37 2.33 −0.62 3.22 1.49 1.93

0.78 0.71 0.02 0.54 0.001 0.14 0.05

0.02 −0.03 0.19 −0.05 0.27 0.12 0.16

Unhealthy foods Attitudes Subjective norms Prototypes Affective associations Intentions Willingness Consumption Nutrition knowledge

1.00 1.00 1.00 1.18 1.00 1.00 29.00 25.56

6.67 7.00 5.67 5.00 5.00 7.00 169.00 86.67

3.68 2.73 2.74 3.28 4.13 5.35 65.86 57.93

0.98 1.39 0.86 0.69 0.98 1.34 18.74 10.63

3.47 2.57 2.60 3.06 3.91 5.18 63.67 59.96

0.95 1.39 0.86 0.65 1.06 1.35 17.79 10.15

3.87 2.87 2.87 3.48 4.32 5.50 67.83 56.10

0.97 1.38 0.85 0.65 0.85 1.31 19.37 10.73

−5.07 −2.57 −3.84 −7.87 −5.20 −2.95 −2.69 4.45

0.001 0.01 0.001 0.001 0.001 0.003 0.007 0.001

−0.42 −0.21 −0.32 −0.65 −0.43 −0.24 −0.22 0.37

Demographics Age in years BMI

18.01 14.89

25.99 50.46

20.89 23.70

1.85 4.90

20.99 24.73

1.94 5.14

20.79 22.79

1.77 4.49

1.31 4.83

0.19 0.001

0.11 0.40

Note. Higher scores indicate more positive attitudes, subjective norms, prototypes, and affective associations regarding both healthy and unhealthy foods. Degrees of freedom for t-tests = 581.

H. Ruhl et al. / Eating Behaviors 23 (2016) 33–40

37

Table 5 Pearson correlations between key study variables, eating behaviors, and demographics. Variable 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

1

Attitudes Subjective norms Prototypes Nutrition knowledge Affective associations Intentions Willingness Consumption Dieting Age BMI

2

3

4

5

6

7

8

9

10

11

0.22⁎⁎⁎

0.25⁎⁎⁎ 0.20⁎⁎⁎

0.06 0.09⁎ 0.00

0.61⁎⁎⁎ 0.18⁎⁎⁎ 0.27⁎⁎⁎ 0.09⁎

0.55⁎⁎⁎ 0.16⁎⁎⁎ 0.13⁎⁎⁎ 0.21⁎⁎⁎ 0.49⁎⁎⁎

0.51⁎⁎⁎ 0.12⁎⁎ 0.12⁎⁎ 0.15⁎⁎⁎ 0.55⁎⁎⁎ 0.63⁎⁎⁎

0.39⁎⁎⁎ 0.12⁎⁎ 0.18⁎⁎⁎ 0.18⁎⁎⁎ 0.37⁎⁎⁎ 0.47⁎⁎⁎ 0.48⁎⁎⁎

0.01 −0.02 0.10⁎ 0.18⁎⁎⁎ −0.03 0.13⁎⁎ 0.06 0.08

0.10⁎ 0.02 0.08 0.07 0.08 0.03 0.04 0.07 0.05

0.05 0.20⁎⁎⁎

0.13⁎⁎

0.00 0.03 0.01 0.01 −0.03 0.01 −0.04 −0.02 0.20⁎⁎⁎ 0.13⁎⁎

0.43⁎⁎⁎ 0.32⁎⁎⁎ −0.12⁎⁎ 0.54⁎⁎⁎ 0.40⁎⁎⁎ 0.29⁎⁎⁎ 0.33⁎⁎⁎ −0.21⁎⁎⁎

0.27⁎⁎⁎ −0.09⁎ 0.22⁎⁎⁎ 0.18⁎⁎⁎ 0.05 0.12⁎⁎ −0.11⁎

−0.16⁎⁎⁎ 0.34⁎⁎⁎ 0.17⁎⁎⁎ 0.06 0.03 −0.16⁎⁎⁎

−0.12⁎⁎ −0.14⁎⁎⁎ −0.10⁎ −0.20⁎⁎⁎ 0.18⁎⁎⁎

0.36⁎⁎⁎ 0.30⁎⁎⁎ 0.30⁎⁎⁎ −0.31⁎⁎⁎

−0.08 −0.07

−0.06 −0.17⁎⁎⁎

−0.06 −0.06

0.07 0.01

−0.07 −0.09⁎

0.59⁎⁎⁎ 0.40⁎⁎⁎ −0.21⁎⁎⁎ −0.12⁎⁎ −0.02

0.45⁎⁎⁎ −0.12⁎⁎ −0.02 0.07

−0.11⁎⁎ −0.09⁎ −0.00

Note. Unhealthy food variables below the diagonal and healthy food variables above the diagonal. Correlations with dichotomous “Dieting” variable are point-biserial correlations. ⁎ p b 0.05. ⁎⁎ p b 0.01. ⁎⁎⁎ p b 0.001.

model fit, the parameter was allowed to vary for dieters and nondieters. Independent-samples t-tests assessed mean-level differences between female dieters and nondieters, and Cohen's d assessed effect sizes (small = 0.2; medium = 0.5; large = 0.8). To examine the relative importance of factors in relation to intentions and willingness to eat foods (Aim 3), standardized coefficients of the final model from Aim 2 were compared. Evaluation of overall model fit was based on the Comparative Fit Index (CFI) and Root Mean Square Error of Approximation (RMSEA), with CFIs greater than or equal to 0.95 or RMSEAs less than or equal to 0.05 indicating acceptable fit (Hu & Bentler, 1998). Some missing data was present in the BMI variable, as four participants requested not to be weighed; however, this variable was not used in primary analyses. 3. Results and discussion Descriptive statistics are shown in Table 4. Bivariate correlations are shown in Table 5. 3.1. The utility of three decision-making models For Aim 1, the TRA models for both healthy and unhealthy foods fit the data poorly (Table 6). The P/W models were significantly better than the TRA models but were still poor fits. The R/R models for healthy and unhealthy foods were significantly better than the P/W models and fit the data well. Thus, the R/R models were retained, suggesting that a model including nutrition knowledge and affective associations more

Table 6 Series of model comparisons to determine the best models for examining healthy and unhealthy eating behaviors. χ2

df

Δχ2

Healthy TRA (M1) P/W (M2) R/R (M3) Diet constrained (M4)

893.07 437.44 30.31 74.28

48 34 455.63⁎⁎⁎ 14 12 407.13⁎⁎⁎ 22 42 43.97⁎ 30

Unhealthy TRA (M1) P/W (M2) R/R (M3) Diet constrained (M4)

677.51 304.53 56.46 105.09

48 34 372.98⁎⁎⁎ 14 12 248.07⁎⁎⁎ 22 42 48.63⁎ 30

Δdf Comparison CFI

RMSEA

M2 vs. M1 M3 vs. M2 M4 vs. M3

0.30 0.66 0.99 0.97

0.17 0.14 0.05 0.04

M2 vs. M1 M3 vs. M2 M4 vs. M3

0.32 0.71 0.95 0.93

0.15 0.12 0.08 0.05

Note. TRA = theory of reasoned action, P/W = prototype/willingness model, R/R = reasoned/reactive model. Models with a CFI ≥ 0.95 or RMSEA ≤ 0.05 were considered to have an acceptable fit (Hu & Bentler, 1998). ⁎ p b 0.05. ⁎⁎⁎ p b 0.001.

sufficiently explains female students' eating behaviors. Thus, the R/R model appears to be an effective tool for examining factors involved in this multifaceted decision-making process. 3.2. Differences between dieters and nondieters 3.2.1. Healthy foods For Aim 2, the final unconstrained model for healthy foods was a better fit than the constrained model (Table 6), suggesting differences between dieters and nondieters (Table 7). Tests for specific differences revealed that attitudes were more strongly related to intentions for nondieters than dieters (Δχ2(1) = 8.15, p = 0.004). Female dieters' intentions to eat healthy foods may not be as strongly related to their attitudes about how beneficial, necessary, and pleasant these foods are because they are more concerned with the foods' role in weight gain (Carels, Konrad, & Harper, 2007). Although dieting moderated the relationship between subjective norms and willingness to eat healthy foods (Δχ2(1) = 5.97, p = 0.01), the coefficients were non-significant

Table 7 Path coefficients of final reasoned/reactive models for examining healthy and unhealthy eating behaviors. Healthy

Unhealthy

Attitudes → Intentions → Willingness

0.20⁎⁎⁎/0.36⁎⁎⁎ 0.30⁎⁎⁎

0.16⁎⁎⁎ 0.22⁎⁎⁎

Subjective norms → Intentions → Willingness

0.03 −0.09/0.07

0.03 −0.08

Prototypes → Willingness

−0.06

−0.07

Nutrition knowledge → Intentions → Willingness

0.11⁎⁎⁎ 0.09⁎⁎

−0.05 −0.06

Affective associations → Intentions → Willingness

0.07 0.39⁎⁎⁎

0.07 0.19⁎⁎⁎

Willingness → Intentions → Behavior

0.44⁎⁎⁎ 0.31⁎⁎⁎

0.47⁎⁎⁎ 0.33⁎⁎⁎

Intentions → Behavior

0.24⁎⁎⁎

0.22⁎⁎⁎

Note. Independent variables are bold. All estimates are standardized. Where one coefficient is presented, the coefficient is constrained to be equal for dieters and nondieters. Where two coefficients are presented, dieters are listed first. ⁎⁎ p b 0.01. ⁎⁎⁎ p b 0.001.

38

H. Ruhl et al. / Eating Behaviors 23 (2016) 33–40

(Table 7). No other moderation effects were found. The final model was an excellent fit (χ2(22) = 39.81, p = 0.01; CFI = 0.99; RMSEA = 0.04). Regarding mean differences (Table 4), female dieters held more positive prototypes of healthy eaters (small effect). Those who perceive healthy eaters positively likely want to be perceived as healthy themselves, which may impact dieting behaviors. Female dieters may also believe they are healthy eaters and show an in-group bias. Dieters had more nutrition knowledge than nondieters (small-medium effect), likely because dieters utilize nutrition information in making food choices conducive to weight loss. Dieters also had more intentions to eat healthy foods (small effect). Despite these intentions, female dieters did not consume more healthy foods than nondieters, indicating that dieting is not always congruent with healthy eating.

3.2.2. Unhealthy foods For Aim 2, the unconstrained model was a better fit (Table 6). However, when coefficients were iteratively constrained to be equal, fit did not significantly decrease, suggesting no moderation based on dieting status. Thus, all regression coefficients were constrained to be equal for dieters and nondieters (Table 7). The final model was a good fit (χ2(24) = 70.98, p b 0.001; CFI = 0.95; RMSEA = 0.06). Regarding mean differences, dieters reported more negative attitudes (small-medium effect) and affect (medium-large effect), fewer subjective norms (small effect), more negative prototypes (small effect), less intention (small-medium effect) and willingness (small effect), and less consumption (small effect) of unhealthy foods (Table 4). Indeed, dieters often hold negative cognitions regarding unhealthy foods and negative perceptions of unhealthy eaters (Barker, Tandy, & Stookey, 1999; Durkin, Hendry, & Stritzke, 2013). Female dieters likely associate unhealthy foods with foiled weight loss attempts, guilt, and depression (Durkin et al., 2013; Oakes, 2005), which is ultimately reflected in their consumption of less unhealthy food than nondieters.

3.3. The relative importance of factors on intentions and willingness This study is the first to examine the relative importance of attitudes, subjective norms, prototypes, nutrition knowledge, and affective associations in relation to intentions and willingness to eat healthy and unhealthy foods. The R/R model highlights the importance of intentions when making eating decisions, with intentions being related to consumption of healthy and unhealthy foods (Table 7). Attitudes were the factor most strongly related to intentions to eat healthy and unhealthy foods. Female students are unlikely to eat foods they perceive as unpleasant, harmful, or unnecessary (Shepherd & Stockley, 1987; Wong & Mullan, 2009). Second to attitudes, nutrition knowledge was related to intentions to eat healthy, but not unhealthy, foods, indicating female students may eat unhealthily despite having good nutrition knowledge (Dickson-Spillmann et al., 2011). In addition to intentions, spontaneous willingness is also related to females' consumption of both healthy and unhealthy foods (Table 7). For healthy and unhealthy foods, attitudes had a relatively strong relationship with willingness. For healthy foods, nutrition knowledge had a smaller, but significant, association with willingness. Having good nutrition knowledge may help female students make healthy choices even when faced with situations that encourage unhealthy eating. For reactive factors, affective associations had the strongest relationship to willingness to eat healthy foods and were second to attitudes in relation to willingness to eat unhealthy foods. Willingness is likely influenced by the rewarding, satisfying, or comforting aspects of foods. This link may explain recent findings that affective associations were related to food consumption but not intentions (Kiviniemi & Brown-Kramer, 2015). Interestingly, these findings suggest that intentions are only related to attitudes and nutrition knowledge, whereas willingness is related to these reasoned factors as well as reactive affect.

3.4. Applications Attitudes, affective associations, and nutrition knowledge should be considered when helping female students make dietary changes. Attitudes about healthy foods might be improved through exposure to healthy foods, by raising awareness of their benefits and increasing preferences for them (Berg, Jonsson, & Conner, 2000; Hoefkens, Pieniak, Van Camp, & Verbeke, 2012). Improving attitudes may be especially beneficial for female nondieters, as their attitudes about healthy foods are even more strongly related to intentions to eat healthy foods than dieters'. Affective associations could be improved by showing women that, prepared in different ways, healthy foods can be refreshing and satisfying. Similarly, helping women reflect on negative associations with unhealthy foods (e.g., guilt, nausea) may reduce unhealthy eating. Lastly, nutrition knowledge could be improved through university classes and student organizations (Kicklighter, Koonce, Rosenbloom, & Commander, 2011; Peterson, Duncan, Null, Roth, & Gill, 2010). Female dieters and nondieters appear to have somewhat different decision-making processes regarding healthy eating. Although dieters reported less consumption of unhealthy foods than nondieters, they did not report more consumption of healthy foods, despite more intentions to eat these foods. Female dieters may find it challenging to follow through with intentions to eat healthy foods because consumption of these foods is not obviously associated with weight loss. Interventions should inform dieters of the indirect role of healthy foods in weight loss to improve attitudes about healthy foods, including reducing caloric intake from foods high in added fats and sugars, aiding in weight loss maintenance, and promoting satiety (Ogden, 2000; Rolls, Ello-Martin, & Tohill, 2004). 3.5. Limitations Self-reports are susceptible to shared-method variance. However, research suggests reasoned and reactive factors are related to food consumption measured with multiple methods (e.g., food sales, food diaries; Buscher, Martin, & Crocker, 2001; Gerrits et al., 2009). Nevertheless, this study relied on a cross-sectional survey and causal relationships cannot be established. Therefore, reversed causality may exist in the study. For instance, healthy food consumption may impact attitudes about healthy foods through exposure. Thus, future research should examine causal relationships with experimental or longitudinal methods. Additionally, the use of a median split for classifying dieters and nondieters may not have produced two distinct, homogenous groups. This method was used in order to include all participants in analyses and to have a sufficient sample for producing reliable parameter estimates. However, future research should consider alternative methods, such as classifying based on tertiles of a dieting measure (van Strien, Herman, & Anschutz, 2012). Another limitation is low reliability for some measures (e.g., behavioral willingness). Low internal consistency for this construct likely arose because individuals may be more willing to eat unhealthy foods in certain situations than others. Future research may benefit from examining the relative importance of different situations that might elicit unhealthy eating, as well as the relative importance of different affective associations, attitudes, and prototypes in relation to eating. Lastly, this study may have attracted participants with a pre-existing interest in health. Thus, the results may not be representative of all female college students. Future research utilizing random sampling is needed to support the study's findings. 3.6. Conclusion Both reasoned and reactive factors are related to eating behaviors. Specifically, the reasoned/reactive model indicates that attitudes, nutrition knowledge, and affective associations relate to female students'

H. Ruhl et al. / Eating Behaviors 23 (2016) 33–40

intentions and willingness to eat foods. Additionally, findings indicate notable differences between female dieters and nondieters. It is our hope that the findings from the R/R model will assist in developing interventions focused on improving the dietary habits of female college students. Furthermore, because the R/R model accounts for a broad range of factors, future research on other health and non-health behaviors may also benefit from the use of this new decision-making model. Role of funding sources This research was funded by a dissertation grant provided by the Psychological Sciences program and a graduate fellowship provided by Carol and Maynard Redeker at the University of Texas at Dallas. Funding sources played no role in the study design, collection, analysis or interpretation of the data, writing the manuscript, or the decision to submit the paper for publication. However, faculty members in the Psychological Sciences program at the university were involved in refining the study design and providing feedback throughout the progression of the study. Contributors This research was a dissertation study completed by the first author, Holly Ruhl. The second author, Shayla Holub, was the student's faculty advisor and was involved in the conception and design of the study, as well as drafting of the manuscript. The third author, Elaine Dolan, conducted literature searches and provided feedback on drafts of the manuscript. All authors contributed to and have approved the final manuscript. Conflict of interest The authors declare that they have no conflicts of interest. Acknowledgements This research was funded by a dissertation grant provided by the School of Behavioral and Brain Sciences and a graduate fellowship provided by Carol and Maynard Redeker. We greatly appreciate their support of this research.

References Aikman, S. N., Crites, S. L., & Fabrigar, L. R. (2006). Beyond affect and cognition: Identification of the informational bases of food attitudes. Journal of Applied Social Psychology, 36(2), 340–382. http://dx.doi.org/10.1111/j.0021-9029.2006.00011.x. Anding, J. D., Suminski, R. R., & Boss, L. (2001). Dietary intake, body mass index, exercise, and alcohol: Are college women following the dietary guidelines for Americans? Journal of American College Health, 49(4), 167–171. http://dx.doi.org/10.1080/ 07448480109596299. Barker, M. E., Tandy, M. M., & Stookey, J. D. (1999). How are consumers of low-fat and high-fat diets perceived by those with lower and higher fat intake? Appetite, 33(3), 309–317. http://dx.doi.org/10.1006/appe.1999.0248. Berg, C., Jonsson, I., & Conner, M. (2000). Understanding choice of milk and bread for breakfast among Swedish children aged 11–15 years: An application of the theory of planned behaviour. Appetite, 34(1), 5–19. http://dx.doi.org/10.1006/appe.1999. 0269. Boeing, H., Bechthold, A., Bub, A., Ellinger, S., Haller, D., Kroke, A., ... Watzl, B. (2012). Critical review: Vegetables and fruit in the prevention of chronic diseases. European Journal of Nutrition, 51(6), 637–663. http://dx.doi.org/10.1007/s00394-012-0380-y. Brewer, J., Blake, A., Rankin, S., & Douglass, L. (1999). Theory of reasoned action predicts milk consumption in women. Journal of the American Dietetic Association, 99(1), 39–44. Buscher, L., Martin, K., & Crocker, S. (2001). Point-of-purchase messages framed in terms of cost, convenience, taste, and energy improve healthful snack selection in a college foodservice setting. Journal of the American Dietetic Association, 101(8), 909–913. Carels, R., Konrad, K., & Harper, J. (2007). Individual differences in food perceptions and calorie estimation: An examination of dieting status, weight, and gender. Appetite, 49(2), 450–458. http://dx.doi.org/10.1016/j.appet.2007.02.009. Centers for Disease Control and Prevention (2011). About BMI for adults. (Retrieved from http://www.cdc.gov/healthyweight/assessing/bmi/adult_bmi/index.html). Chacko George, G., Milani, T. J., Hanss-Nuss, H., Kim, M., & Freeland-Graves, J. H. (2004). Development and validation of a semi-quantitative food frequency questionnaire for young adult women in the southwestern United States. Nutrition Research, 24(1), 29–43. http://dx.doi.org/10.1016/j.nutres.2003.09.006. Conner, M., Hugh-Jones, S., & Berg, C. (2011). Using the two-factor theory of planned behaviour to predict adolescent breakfast choices. Educational and child psychology. 28(4). (pp. 37–50). Dickson-Spillmann, M., Siegrist, M., & Keller, C. (2011). Development and validation of a short, consumer-oriented nutrition knowledge questionnaire. Appetite, 56(3), 617–620. http://dx.doi.org/10.1016/j.appet.2011.01.034. Dohnke, B., Steinhilber, A., & Fuchs, T. (2015). Adolescents' eating behaviour in general and in the peer context: Testing the prototype-willingness model. Psychology & Health, 30(4), 381–399. http://dx.doi.org/10.1080/08870446.2014.974604. Durkin, K., Hendry, A., & Stritzke, W. K. (2013). Mixed selection. Effects of body images, dietary restraint, and persuasive messages on females' orientations towards chocolate. Appetite, 60(1), 95–102. http://dx.doi.org/10.1016/j.appet.2012.09.025. Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to theory and research. Reading, MA: Addison-Wesley.

39

Gerrits, J. H., de Ridder, D. D., de Wit, J. F., & Kuijer, R. G. (2009). Cool and independent or foolish and undisciplined? Adolescents' prototypes of (un)healthy eaters and their association with eating behaviour. Appetite, 53(3), 407–413. http://dx.doi.org/10. 1016/j.appet.2009.08.008. Gerrits, J., O'Hara, R., Piko, B., Gibbons, F., de Ridder, D., Keresztes, N., ... de Wit, J. (2010). Self-control, diet concerns and eater prototypes influence fatty foods consumption of adolescents in three countries. Health Education Research, 25(6), 1031–1041. http:// dx.doi.org/10.1093/her/cyq055. Gibbons, F. X., Gerrard, M., Blanton, H., & Russell, D. W. (1998). Reasoned action and social reaction: Willingness and intention as independent predictors of health risk. Journal of Personality and Social Psychology, 74(5), 1164–1180. http://dx.doi.org/10.1037/ 0022-3514.74.5.1164. Gidding, S. S., Dennison, B. A., Birch, L. L., Daniels, S. R., Gilman, M. W., Lichtenstein, A. H., ... Van Horn, L. (2005). Dietary recommendations for children and adolescents: A guide for practitioners. Pediatrics, 117(2), 544–559. http://dx.doi.org/10.1542/peds.20052374. Hoefkens, C., Pieniak, Z., Van Camp, J., & Verbeke, W. (2012). Explaining the effects of a point-of-purchase nutrition-information intervention in university canteens: A structural equation modelling analysis. International Journal of Behavioral Nutrition and Physical Activity, 9, 111. http://dx.doi.org/10.1186/1479-5868-9-111. Hu, L., & Bentler, P. M. (1998). Fit indices in covariance structure modeling: Sensitivity to underparameterized model misspecification. Psychological Methods, 3(4), 424–453. http://dx.doi.org/10.1037/1082-989X.3.4.424. Irmak, C., Vallen, B., & Rosen Robinson, S. (2011). The impact of product name on dieters' and nondieters' food evaluations and consumption. Journal of Consumer Research, 38(2), 390–405. http://dx.doi.org/10.1086/660044. Kelly-Weeder, S. (2011). Binge drinking and disordered eating in college students. Journal of the American Academy of Nurse Practitioners, 23(1), 33–41. http://dx.doi.org/10. 1111/j.1745-7599.2010.00568.x. Kicklighter, J., Koonce, V., Rosenbloom, C., & Commander, N. (2011). College freshmen perceptions of effective and ineffective aspects of nutrition education. Journal of American College Health, 59(2), 98–104. http://dx.doi.org/10.1080/07448481.2010. 483709. Kiviniemi, M. T., & Brown-Kramer, C. R. (2015). Planning versus action: Different decision-making processes predict plans to change one's diet versus actual dietary behavior. Journal of Health Psychology, 20(5), 556–568. http://dx.doi.org/10.1177/ 1359105315576605. Kiviniemi, M. T., & Duangdao, K. M. (2009). Affective associations mediate the influence of cost–benefit beliefs on fruit and vegetable consumption. Appetite, 52(3), 771–775. http://dx.doi.org/10.1016/j.appet.2009.02.006. Kolodinsky, J., Harvey-Berino, J., Berlin, L., Johnson, R. K., & Reynolds, T. (2007). Knowledge of current dietary guidelines and food choice by college students: Better eaters have higher knowledge of dietary guidance. Journal of the American Dietetic Association, 107(8), 1409–1413. http://dx.doi.org/10.1016/j.jada.2007.05.016. Maas, J., Keijsers, G. J., Rinck, M., Tanis, J., & Becker, E. S. (2015). Does a dieting goal affect automatic cognitive processes and their trainability? Cognitive Therapy and Research, 39(3), 378–389. http://dx.doi.org/10.1007/s10608-014-9658-0. Mannino, M. L., Lee, Y., Mitchell, D. C., Smiciklas-Wright, H., & Birch, L. L. (2004). The quality of girls' diets declines and tracks across middle childhood. International Journal of Behavioral Nutrition and Physical Activity, 1(1), 5–11. http://dx.doi.org/10.1186/14795868-1-5. Oakes, M. E. (2005). Stereotypical thinking about foods and perceived capacity to promote weight gain. Appetite, 44(3), 317–324. http://dx.doi.org/10.1016/j.appet.2005. 03.010. Ogden, J. (2000). The correlates of long-term weight loss: A group comparison study of obesity. International Journal of Obesity, 24(8), 1018–1025. http://dx.doi.org/10. 1038/sj.ijo.0801354. Ohtomo, S., Hirose, Y., & Midden, C. H. (2011). Cultural differences of a dual-motivation model on health risk behaviour. Journal of Risk Research, 14(1), 85–96. http://dx.doi. org/10.1080/13669877.2010.503936. Parmenter, K., & Wardle, J. (1999). Development of a general nutrition knowledge questionnaire for adults. European Journal of Clinical Nutrition, 53(4), 298–308. Peterson, S., Duncan, D., Null, D., Roth, S., & Gill, L. (2010). Positive changes in perceptions and selections of healthful foods by college students after a short-term point-of-selection intervention at a dining hall. Journal of American College Health, 58(5), 425–431. http://dx.doi.org/10.1080/07448480903540457. Racette, S. B., Deusinger, S. S., Strube, M. J., Highstein, G. R., & Deusinger, R. H. (2005). Weight changes, exercise, and dietary patterns during freshman and sophomore years of college. Journal of American College Health, 53(6), 245–251. Rolls, B. J., Ello-Martin, J. A., & Tohill, B. C. (2004). What can intervention studies tell us about the relationship between fruit and vegetable consumption and weight management? Nutrition Reviews, 62(1), 1–17. http://dx.doi.org/10.1301/nr. 2004.jan.1-17. Shepherd, R., & Stockley, L. (1987). Nutrition knowledge, attitudes, and fat consumption. Journal of the American Dietetic Association, 87(5), 615. Shepherd, R., & Towler, G. (1992). Nutrition knowledge, attitudes and fat intake: Application of the theory of reasoned action. Journal of Human Nutrition and Dietetics, 5, 387–397. http://dx.doi.org/10.1111/j.1365-277X.1992.tb00178.x. Tapper, K., & Pothos, E. M. (2010). Development and validation of a food preoccupation questionnaire. Eating Behaviors, 11(1), 45–53. http://dx.doi.org/10.1016/j.eatbeh. 2009.09.003. van Strien, T., Frijters, J. E., Bergers, G. P., & Defares, P. B. (1986). The Dutch eating behavior questionnaire (DEBQ) for assessment of restrained, emotional, and external eating behavior. International Journal of Eating Disorders, 5(2), 295–315. http://dx.doi.org/10.1002/1098-108X(198602)5:2b295::AIDEAT2260050209N3.0.CO;2-T.

40

H. Ruhl et al. / Eating Behaviors 23 (2016) 33–40

van Strien, T., Herman, C. P., & Anschutz, D. (2012). The predictive validity of the DEBQ external eating scale for eating in response to food commercials while watching television. International Journal of Eating Disorders, 45(2), 257–262. http://dx.doi.org/10. 1002/eat.20940. Walsh, E. M., & Kiviniemi, M. T. (2014). Changing how I feel about the food: Experimentally manipulated affective associations with fruits change fruit choice behaviors. Journal of Behavioral Medicine, 37(2), 322–331. http://dx.doi.org/10.1007/s10865012-9490-5.

Wong, C. L., & Mullan, B. A. (2009). Predicting breakfast consumption: An application of the theory of planned behaviour and the investigation of past behaviour and executive function. British Journal of Health Psychology, 14(3), 489–504. http://dx.doi.org/ 10.1348/135910708X360719. York-Crowe, E. E., White, M. A., Paeratakul, S., & Williamson, D. A. (2006). The diet and health knowledge survey: Development of a short interview format. Eating Behaviors, 7(3), 235–242. http://dx.doi.org/10.1016/j.eatbeh.2005.08.010.