299

Health Education Research Psychosocial predictors of fruit and vegetable consumption among elementary school children S.B. Domel, W.O. Thompson, H.C. Davis, T. Baranowski, S.B. Leonard and J. Baranowski Health Educ. Res. 11:299-308, 1996. doi:10.1093/her/11.3.299

The full text of this article, along with updated information and services is available online at http://her.oxfordjournals.org Cited by

This article has been cited by 10 articles at 27 July 2008 . View these citations at http://her.oxfordjournals.org#otherarticles

Reprints

Reprints of this article can be ordered at http://www.oxfordjournals.org/corporate_services/reprints.html

Email and RSS alerting

Sign up for email alerts, and subscribe to this journal’s RSS feeds at http://her.oxfordjournals.org

PowerPoint® image downloads

Images from this journal can be downloaded with one click as a PowerPoint slide.

Journal information

Additional information about Health Education Research, including how to subscribe can be found at http://her.oxfordjournals.org

Published on behalf of

Oxford University Press http://www.oxfordjournals.org/

Downloaded from http://her.oxfordjournals.org by on 27 July 2008

HEALTH EDUCATION RESEARCH Theory & Practice

Vol.11 no.3 1996 Pages 299-308

Psychosocial predictors of fruit and vegetable consumption among elementary school children S.B.Domel, W.O.Thompson1, H.C.Davis1, T.Baranowski2, S.B.Leonard3 and J.Baranowski2 Abstract A self-efficacy questionnaire for fruit and vegetable (F&V) consumption among fourth and fifth grade students was developed, pilot tested with 140 students from one school and field tested with 252 students from two schools. The questionnaire included 34 items generated by inventorying behaviors targeted for change in the F&V school curriculum. The four subscales derived from principal components analysis were labeled 'after-school F&V snacks'; 'breakfast and lunch F&V, and paying for F&V; 'independent shopping for F&V; and 'assisted shopping for F&V; they accounted for 43.8% of the total variance. Subscale test-re-test reliabilities and internal consistencies were acceptable. Criterion validities against F&V consumption from food records were low. Relationships were stronger with preferences and outcome expectations. Results from stepwise regression analyses indicated that preferences were the only significant predictors of fruit and total F&V consumption, as well as the main predictor of vegetable consumption; however, the models accounted for less than 13% of the variance. Nutrition education programs which target preferences may be more effective in increasing Georgia Prevention Institute, Department of Pediatrics, Medical College of Georgia, Augusta, GA 30912-3710, 'Office or Biostatistics, Medical College of Georgia, Augusta, GA 30912-4900, 'Department or Behavioral Science, University of Texas MJXAnderson Cancer Center, Houston, TX 77030 and ^Section of Endocrinology and Nutrition, Department of Medidne, Medical College of Georgia, Augusta, GA 30912-3102, USA

© Oxford University Press

F&V consumption among elementary school children than programs which target selfefficacy and outcome expectations. However, since preferences accounted for only small proportions of the variances, further research should consider other issues such as availability.

Introduction Increased fruit and vegetable (F&V) consumption is an aspect of eating which has received considerable attention recently in various nutrition guidelines to decrease the risk of several chronic diseases such as cardiovascular disease and certain cancers (American Heart Association, 1988; Butrum et al., 1988; Surgeon General, 1988; Committee on Diet and Health, 1989; USDA and USDHHS, 1990). One of the Nation's Year 2000 Health Goals (USDHHS, 1991) specifies a minimum of five daily servings of F&V; the Food Guide Pyramid (USDA and HNIS, 1992) recommends two to four daily servings of fruit (F) and three to five daily servings of vegetables (V). A national program entitled '5 A Day For Better Health' (Subar et al, 1992; Havas et al., 1995) was recently initiated to help Americans eat more F&V since most adults and children consume much less than the recommended levels (Patterson et al., 1990; Block, 1991; Subar et al., 1992; Wolfe and Campbell, 1993; DOmel et al., 1994a). Behaviors initiated during childhood may last into adulthood and thereby impact the risk of chronic diseases (Kelder et al., 1994). Eating is a behavior. According to Bandura's Social Cognitive Theory (1986), behavior may be explained and predicted by several key concepts 299

S.B.Domel et al. including self-efficacy and outcome expectations. Within social cognitive theory, skills provide the capability to perform the behavior, outcome expectations provide the motivation for the behavior and self-efficacy provides the confidence that barriers can be overcome (Baranowski, 1990). Self-efficacy concerns the beliefs a person has about how capable s/he is of performing a particular behavior in particular situations (Strecher et al., 1986). Bandura (1982, 1986) proposed that behavior change will occur only when the person has a substantial level of self-efficacy for, or the confidence that s/he can perform, the new behavior. Since self-efficacy reflects beliefs about abilities, the perceptions, not necessarily the 'true' abilities, influence behavior (Strecher et al., 1986). Strong relationships were reported between self-efficacy and health behavior change and maintenance in a review of studies regarding health practices such as weight control and alcohol abuse (Strecher et al., 1986). In the 'Go For Health' project, self-efficacy for heart-healthy foods was studied among third and fourth grade students through 18 items that asked how sure students were of choosing a hearthealthy alternative in a variety of scenarios; an a coefficient of 0.76 was reported (Parcel et al, 1989). Outcome expectations are the beliefs about whether a certain behavior will lead to certain outcomes (Strecher et al., 1986). Outcome expectations have been shown to relate to alcohol consumption among adolescents and adults (see Goldman et al., 1987 for a review), and to lowand high-fat food consumption among college students (Bowen et al, 1992). Food preferences are also thought to be determinants of eating behavior (Birch and Sullivan, 1991; Contento, 1991). Preference involves affect or liking for something and/or choosing one thing over something else (Birch and Sullivan, 1991). Birch (1979) reported that preferences of nursery school children for various sandwiches were effective predictors of consumption (r = 0.80, P < 0.01). Calfas etal. (1991) reported 66% agreement between stated preferences and food choices from

300

pairs of photographed foods among 4- to 8-yearold children. Research regarding the reliability and validity of questionnaires for preferences and outcome expectations regarding F&V consumption among fourth and fifth grade students has recently been conducted (Domel et al., 1993a, 1995). The purpose of this paper is two-fold: (1) to describe the development, pilot testing and field application of a self-efficacy questionnaire for F&V consumption among fourth and fifth grade students, and to discuss questionnaire subscales and assessment of reliability and validity; and (2) to report psychosocial predictors, specifically self-efficacy, outcome expectations and preferences of F&V consumption among fourth and fifth grade students. This study was conducted with data collected for a larger study which involved the development and pilot testing of a 6 week, school-based curriculum to increase F&V consumption among fourth and fifth grade students (Domel et al., 1993b).

Methods

Overview The self-efficacy questionnaire for F&V consumption was developed and pilot tested at one school and field tested at two schools in an intervention trial; the three elementary schools were randomly assigned to either the developmental or field application conditions. Approximately half of the students were eligible for free or reduced-price lunches. The study received approval from the institutional Human Assurance Committee; informed consent forms were signed by each participant and one of his/her parents.

Development of self-efficacy questionnaire The self-efficacy questionnaire included 34 items generated by inventorying behaviors targeted for change in the F&V school curriculum; topics included purchasing F&V, selecting F&V over other items for after-school snacks, including F at breakfast and adding F&V at lunch. The three response options listed for each item included 'not

Psychosocial predictors of fruit and vegetable intake at all confident', 'a little confident' and 'very confident'.

Pilot testing Approximately 73% of the 246 students from all fourth and fifth grade classes (five of each) from school 1 participated in the pilot testing. Data collectors followed a written protocol to classroom administer the questionnaire two times about 14 days apart (to assess test-re-test reliability); the procedure for classroom administration has been described previously (Domel et al., 1993a).

Questionnaire revisions Cronbach's a coefficients for the 34 items were 0.88 and 0.92 within the first and second administrations, respectively. Test-re-test reliability (Pearson product-moment correlation) between the first and second administrations was 0.70. Based on these results and comments from the data collectors, minor wording revisions were made in two items, and the word 'confident' was changed to 'sure' in the response options and group headings before field application.

Field application Approximately 92% of the 378 fourth and fifth grade students from all fourth and fifth grade classes (eight of each) from schools 2 and 3 participated in the field application. The revised questionnaire was classroom administered in a manner similar to that used during pilot testing.

Analyses Responses were coded as 0 = not at all confident (sure), 1 = a little confident (sure) and 2 = very confident (sure). The few students of other racialethnic groups (4.3%) were added to the white, nonHispanic group to form a 'non-African-American' group since the vast majority of the sample was white, non-Hispanic (40.3%) or AfricanAmerican (55.4%). Principal components analysis was conducted to identify subscales. Because only minor revisions were made between pilot and field testing, the data from all three schools were combined in order to

use the largest possible data set A scree plot of the eigenvalues was used to determine the number of principal components to retain; varimax rotation was then used to achieve simple structure. Data were also randomly split in half and re-analyzed to determine whether the items loaded similarly on the same number of components. Items with loadings of 0.4 or higher on more than one component or with loadings of less than 0.4 on any component were excluded from further analyses. The items that loaded at 0.4 or higher for each of the principal components were unit weighted and averaged to compute self-efficacy subscale scores for each subject A multivariate analyses of variance (MANOVA) was conducted on the set of self-efficacy subscales by gender, grade (fourth, fifth), ethnicity (nonAfrican-American, African-American) and school (1,2, 3). This was followed by univariate analyses of variance (ANOVA) for each subscale separately. Internal consistencies were calculated using Cronbach's a coefficient to assess reliability for each subscale for schools 2 and 3. The pilot data from school 1 were re-analyzed by subscale; this included Pearson product—moment correlations to determine test—re-test reliabilities and Cronbach's a to assess internal consistencies. Test-re-test reliabilities by subscale were also calculated for school 3 over a 7 week period; they were not calculated for school 2 since an educational intervention had occurred. To evaluate criterion validity between the questionnaire and F&V consumption (Contento, 1991) for schools 2 and 3 combined (field sample), Pearson product-moment correlations were computed between the F&V self-efficacy subscales and actual F&V consumption determined from daily food records completed by students for 1 week prior to completing the questionnaire; consumption variables included servings of F (F and F&V juices), V (V and legumes) and F&V combined The food record procedure and validation through school lunch observations have been described previously (Domel et al., 1994b); Pearson correlations between components of the school lunch

301

S.B.Domel et al. portion of the food records and observations ranged from 0.16 to 0.85 with a median of 0.66. The procedure for coding F&V servings from the food records has been described elsewhere (Domel et al., 1993b). To relate F&V self-efficacy with F&V preferences for the field sample, Pearson correlations were computed between the self-efficacy subscales and preferences for F, V and F&V snacks as assessed through a F&V preferences questionnaire administered 1 day prior to the self-efficacy questionnaire. The 31 items assessed how much the students liked 10 common F and 10 common V, and whether the students preferred F&V to 11 other after-school snacks. Cronbach's a coefficients for the three preference subscales (F, V and F&V snacks) ranged from 0.55 to 0.77; Pearson correlations for each of the three preference subscales with F&V consumption ranged from 0.01 to 0.29 with a median of 0.17 (Domel et al., 1993a). To relate F&V self-efficacy with F&V outcome expectations, Pearson correlations were computed between the F&V self-efficacy subscales and two F&V outcome expectation subscales as assessed through a F&V outcome expectations questionnaire administered immediately after the self-efficacy questionnaire. The 17 items were applied to F and repeated for V. Principal components analysis indicated two outcome expectation subscales: social, and health and physical ability. Cronbach's a coefficients for the subscales ranged from 0.79 to 0.89; Pearson correlations for each of the subscales with F&V consumption ranged from 0.11 to 0.16 (Domel et al., 1995). Correlations with the F&V self-efficacy subscales were computed within cells and transformed using Fisher's r to z transformation; a weighted average was then taken and transformed back to r. These transformations were required since mean values differed by cell membership, which can inflate correlation coefficients. Significance was set at a of 0.05; significance tests were one-tailed where applicable. To determine psychosocial predictors of F&V consumption among fourth and fifth grade students, a stepwise regression was conducted with demo-

302

graphics (gender, ethnicity, grade), F&V preferences subscales, F&V outcome expectations subscales and F&V self-efficacy subscales as the independent variables, and F consumption as the dependent variable. The analysis was then conducted with V consumption as the dependent variable-, and finally with total F&V consumption as the dependent variable. F&V were analyzed separately as well as combined for direct assessment of the Year 2000 F&V health goal (USDHHS, 1991) and the '5 A Day For Better Health' program goal (Subar et al., 1992) which combine F&V, and the Food Guide Pyramid recommendations (USDA and HNIS, 1992) which separate F from V.

Results The sample consisted of 392 students (49% male, 54% fourth graders, 56% African-American) with 140, 164 and 88 students from schools 1, 2 and 3, respectively. The samples by school were approximately evenly split by gender, grade and racialethnic group except for at school 1 where 67% were African-American and at school 3 where 64% were fourth graders. Approximately 16% of the students were absent from class during questionnaire administration. Approximately 7% of the students were eliminated from analyses for not answering all of the items and another 3% for answering every item 'not at all confident/sure', which suggested they were not paying attention to the individual questions. The four subscales derived from principal components analysis were labeled 'after-school F&V snacks'; 'breakfast and lunch F&V, and paying for F & V ; 'independent shopping for F&V; and 'assisted shopping for F&V; they accounted for 43.8% of the total variance. Similar results were indicated when principal components analysis with varimax rotation was conducted on half of the sample randomly selected and then repeated on the remaining half. Overall subscale means across items were 1.29 (SD = 0.55) for 'after school F&V snacks'; 1.30 (SD = 0.49) for 'breakfast and lunch F&V, and paying for F&V; 1.17 (SD = 0.53) for 'independ-

Psychosocial predictors of fruit and vegetable intake

Table L Coefficient a for four F&V self-efficacy subscales by school (separately and combined) F&V self-efficacy subscales

School 1 1st administration 2nd administration School 2 School 3 All three schools combined

After school F&V snacks

Breakfast and lunch F&V, and paying for F&V

Independent shopping for F&V

Assisted shopping for F&V

0.81 0.86 0.86 0.88 0.87

0.68 0.77 0.77 0.76 0.78

0.76 0.84 0.76 0.68 0.74

0.74 0.79 0.71 0.64 0.72

Table II. Pearson product-moment correlations between four F&V self-efficacy subscales

After school F&V snacks Breakfast and lunch F&V, and paying for F&V Independent shopping for F&V

Breakfast and lunch F&V, and paying for F&V

Independent shopping for F&V

Assisted shopping for F&V

0.57

0.21 0.29

0.44 0.45 0.34

ent shopping for F&V; and 1.69 (SD = 0.39) for 'assisted shopping for F&V. Means for the 'assisted shopping for F&V subscale were higher than those for the other three subscales regardless of gender, grade and ethnicity. Cronbach's a coefficients (Table I) for the four subscales were high and ranged from 0.72 to 0.87 for all three schools combined, which indicated that the subscale items were measuring the same underlying constructs. Pearson product-moment correlations (Table II) between the four subscales were low and ranged from 0.21 to 0.57, which indicated that they were measuring four different constructs. Test-re-test reliabilities (Table HI) at school 1 over a 2 week period were acceptable; however, at school 3 over a 7 week period, they were lower than at school 1 over a 2 week period. Results from the MANOVA indicated an overall significant effect for the school by grade by ethnicity by gender interaction (P < 0.03). Since this term included all of the independent variables, the univariate ANOVA's were conducted using the full model with each of the four subscales as dependent variables. Results from the ANOVA for the 'after school F&V snacks' subscale indicated a significant school effect (P < 0.001). Means at schools 2

Table m . Test-re-test reliabilities for four F&V self-efficacy subscales for school I and school 3*

After school F&V snacki Breakfast and lunch F&V, and paying for F&V Independent shopping for F&V Assisted shopping for F&V

School l b

School 3C

0.62 0.67

0.51 0.63

0.64 0.52

0.53 0.35

Test-re-test reliabilities were not calculated for School 2 since an educational intervention occurred between the two administrations. 'Two weeks separated the two administrations. c Seven weeks separated the two administrations.

and 3 were the highest (1.46 and 1.44, respectively) and school 1 (0.99) the lowest. For the 'breakfast and lunch F&V, and paying for F&V subscale, ANOVA results indicated a significant ethnicity effect (P < 0.004) and a significant school effect (P < 0.001) which were subsumed within a significant gender by ethnicity by school interaction (P < 0.01). At school 1, African-American boys and non-African-American girls scored the highest (1.11 and 1.10, respectively). At school 2, African-American boys scored

303

S.B.Domel et al.

Tame IV. Weighted Pearson correlations (one-tailed) between four F&.V self-efficacy subscaUs, F and V consumption, F and V preferences, and two F and V outcome expectation subscaUs for field sample (n = 206) F and V self-efficacy subscales

Consumption fruits vegetables F&V Preferences fruits vegetables F&V snacks Outcome expectation subscales social health/physical ability

After school F and V snacks

Breakfast and lunch F&V, and paying for F and V

Independent shopping for F and V

Assisted shopping for F and V

0.06 0.06 0.06

0.07 0.17* 0.13*

0.09 0.06 0.08

-0.02 0.08 0.04

0.20** 0.23*** 0.49*«*

0.23*** 0.19** 0.12

0.12 0.03 0.15*

0.18** 0.22** 0.26***

0.06 0.24***

0.14* 0.25***

0.20** 0.13*

0.02 0.24***

*P < 0.05; **P < 0.01; •••/> < 0.001.

the highest (1.65), followed by African-American girls (1.55). At school 3, African-American girls scored the highest (1.58), followed by AfricanAmerican boys (1.44). Means for all groups at school 1 were lower than those at schools 2 and 3. Results from the ANOVA for the 'independent shopping for F&V subscale indicated only a significant ethnicity effect (P < 0.04); means for African-Americans were higher than for nonAfrican-Americans (1.24 versus 1.08). Results from ANOVA for the 'assisted shopping for F&V subscale were non-significant Pearson correlations between the F&V selfefficacy subscales and F&V consumption, F&V preferences and F&V outcome expectations for the field sample are found in Table IV. Of the 252 students in the field sample, 206 (137 and 69 from schools 2 and 3, respectively) completed all three of the questionnaires and the food record. This sample was fairly evenly split by gender, grade and ethnicity. Correlations between the self-efficacy subscales and F&V consumption ranged from -0.02 to 0.17. Between the self-efficacy subscales and preferences, correlations ranged from 0.03 to 0.49; the highest correlation coefficient was between preferences for F&V snacks and selfefficacy for 'after school F&V snacks'. Correla-

304

Tabte V. Pearson correlations (unweighted) between demographic and psychosocial variables and F&V consumption Consumption variables Fruit Demographic variables gender 0.05 grade -0.05 ethnicity 0.06 Preference subscales fruit 0.20** vegetables F&V snacks Self-efficacy subscales after school F&V snacks 0.08 breakfast and lunch F&V, and paying for F&V 0.09 independent shopping for F&V 0.11 assisted shopping for F&V 0.00 Outcome expectation subscales social 0.08 health/physical ability 0.14*

Vegetable

F&V

0.01 0.10 0.07

0.04 0.04 0.09

0.27***

021*** 032*** 0.15*

0.08

0.10

0.20**

0.18**

0.06 0.08

0.10 0.05

0.15* 0.14*

0.15* 0.18**

*P < 0.05; **P < 0.01; ***P < 0.001.

tions between the self-efficacy subscales and outcome expectations ranged from 0.02 to 0.25. Table V includes correlations between each

PsychosociaJ predictors of fruit and vegetable intake

Table VL Regression results predicting FA. V consumption using psychosocial and demographicvariables (n = 206) Variables in equation

Fruit consumption 2

R increase

Significance

Vegetable preferences

Vegetable consumption 2

F&V consumption

R increase

Significance

R2 increase

Significance

0.07

0.01

0.10

0.01

(Mepl) Fruit preferences

0.04

(step 1)

0.01

0.02

0.03 (step 2)

Self-efficacy for breakfast and lunch F&V, and paying for F&V

0.02

0.04 (step 2)

Adjusted R2

0.03

demographic (gender, grade and ethnicity) and psychosocial predictor (subscales for preferences, self-efficacy and outcome expectations) and each of the dependent measures (consumption for F, V and F&V combined) for stepwise regression analyses. Table VI includes R2 change and respective P values from the stepwise regression. The adjusted R2's for F, V and total F&V consumption were 0.03, 0.09 and 0.12, respectively. The only significant predictor for F consumption was F preferences. The first significant predictor for V consumption was V preferences; the second predictor, self-efficacy for 'breakfast and lunch F&V and paying for F&V, increased the R2 by only 0.02. For total F&V consumption, V preferences was the first significant predictor, the second predictor, F preferences, added only 0.02 to the R2. The predictive value of the independent variables was much stronger when V consumption, rather than F consumption, was the dependent variable.

Discussion While internal consistencies were high, test-re-test reliability decreased as time increased from 2 to 7 weeks which indicates instability in the children's response to F&V self-efficacy items. One possible explanation would be that self-efficacy performs much like a mood variable among children. Means for the 'after school F&V snacks' subscale were much lower at school 1 than schools 2

0.09

0.12

and 3. Investigation of individual items indicated that means for the majority of items in this subscale (15 of 17) were significantly lower (ANOVA, P < 0.05) at school 1 than at schools 2 and 3. Although means for all of the F&V self-efficacy subscales indicated that the students reported being somewhere between 'a little confident/sure' and 'very confident/ sure' that they could do the various tasks concerning eating more F&V, overall F&V consumption was low (Domel et ai, 1993b). In addition, criterion validity for the F&V selfefficacy subscales was low; while two of the 12 correlation coefficients with F&V consumption were significantly greater than zero, they accounted for only small proportions of the variance. One possible explanation for the low correlations of F&V self-efficacy to F&V consumption is that the F&V available and served are not completely under the children's control (Hearn et al., 1996), thereby minimizing the possible effect of perceived selfefficacy. Since there are numerous determinants of eating (Michela and Contento, 1986), a second explanation would be that self-efficacy may not be a primary determinant of F&V consumption among children. Alternatively, the instrument itself simply may have low validity. The relationships among F&V self-efficacy, F&V preferences and F&V outcome expectations were stronger than those with F&V consumption. While 10 of the 12 correlation coefficients between F&V self-efficacy and F&V preferences were sig-

305

S.B.Domel et al. nificantly greater than zero, they were nevertheless low; the highest correlation was between preferences for F&V snacks and the 'after school F&V snacks' subscale. First, consistent with above theorizing, the child probably exerts more influence over snacks than any of the meals. Alternatively, it is possible that enhanced motivation for F&V, as revealed through preferences, enhanced the perceptions of self-efficacy (e.g. an enhanced willingness to overcome more potential barriers); that higher self-efficacy had a dissonance reduction type effect on preferences (Scher and Cooper, 1989); or some third variable was affecting both self-efficacy and preferences, e.g. exposure to F&V (Domel et al., 1993a). Six of the eight correlation coefficients between self-efficacy and outcome expectations were significantly greater than zero; however, they were also moderately low, thereby accounting for only small proportions of the variance. The questionnaire might be improved by including a more systematic sampling of question types. For example, a matrix could be developed from which questions could be sampled and systematically varied including different meals (breakfast, lunch, dinner, snack), preparation method (frozen, canned, raw), location (home, school, fast food restaurant), identification of food type (favorite F or V, specific F or V, favorite cookie or candy bar) and position of food ('instead of or 'in addition to'). We believe that the 'instead o f format (e.g. selection of F&V items over an alternative) poses more realistic choices in some situations while the 'in addition to' format poses more realistic choices in other situations; thus, both types of formats merit further research. An issue which could impact the relationship between F&V self-efficacy and F&V consumption involves many teachers, parents and care-givers who may unconsciously attempt to deactivate a child's individual responsibility for changing eating behaviors, e.g. by requiring the child to eat F&V served. This conflicts with the self-efficacy model which attempts to activate the person to assume individual responsibility for changing (Strecher et al., 1986). According to Satter (1983), a division

306

of feeding responsibility is needed in which the adult is responsible for providing food that is appropriate for the child and offering it in a positive environment; the child is responsible for deciding how much of, or even if, the food offered is eaten. Perhaps more emphasis is needed concerning this division of feeding responsibility in nutrition education programs for adults who are involved in feeding children. More emphasis may be needed in school nutrition education programs to help students assume age-appropriate responsibility for changing eating behavior, as well as to differentiate between feeling confident about what to do to make the changes, increasing the desire to do so and actually doing it In the investigation of psychosocial predictors of F&V consumption, preferences were consistently related to F&V consumption among children; however, the variances accounted for were all less than 13%. These findings are similar to those of Resnicow et al. (19%). There are many determinants of eating (Michela and Contento, 1986). Since the foods served are not completely under the children's control, availability may be a strong predictor (Baranowski et al., 1993; Michela and Contento, 1986; Heam et al., 1996). For example, exposure to F&V has been related to children's food preferences and to consumption (Birch, 1979; Domel et al, 1993a). This emphasis on an environmental determinant which may not be under the children's control is consistent with the recent efforts of Fishbein's theory of planned change (Ajzen, 1985). An implication of these analyses is that psychosocial variables should be more strongly related to consumption when the foods are more readily available.

Conclusions Although reliability for the F&V self-efficacy questionnaire was acceptable, criterion validity against F&V consumption was low. Stronger relationships were found between F&V self-efficacy and F&V preferences, and between F&V self-efficacy and F&V outcome expectations, indicating that how much children like F&V as well as what they

Psychosocial predictors of fruit and vegetable intake expect will happen as a result of eating F&V impact the confidence they have in eating more F&V. Preferences were consistent predictors of F&V consumption, especially vegetable consumption, among fourth and fifth grade children. Thus, it appears that nutrition education programs which target F&V preferences may be more effective at increasing F&V consumption among elementary school children than programs which target F&V self-efficacy and F&V outcome expectations. While preferences were consistent predictors of F&V consumption, they accounted for only small proportions of the variance. Research is needed to clarify the role of F&V availability as a moderator of F&V consumption. Further research should be directed at making F&V more available to children or exposing them to a larger variety of F&V prepared a variety of ways to document the role of availability and exposure in increasing F&V preferences and F&V consumption among this population.

Acknowledgements The authors recognize the important contributions of Drs Rebecca Mullis and Tim Byers, and thank them for their encouragement and support of this project. Sincere appreciation is also expressed to Patricia Riley, Lisa Bryant and Sandra Young for helping with data collection, and to Carrie B. Harris for secretarial support. This study was part of a collaborative project among the International Apple Institute, Porter Novelli Omnicom PR Network, the Centers for Disease Control— Division of Nutrition, the Richmond County Board of Education, and the Georgia Institute for the Prevention of Human Disease and Accidents (Georgia Prevention Institute) of the Medical College of Georgia. This study was made possible by funds from the International Apple Institute.

References Ajzen, I. (1985) From intentions to actions: a theory of planned action. In Kuhl, J. and Beckman, J. (eds). Action Control: From Cognition to Behavior. Springer, New York.

American Heart Association (1988) Dietary Guidelines for Healthy American Adults: a statement for Physicians and Health Professionals by the Nutrition Committee, American Heart Association. Circulation, 77, 721A-724A. Bandura, A. (1982) Self-efficacy mechanism in human agency. American Psychologist. 37, 122-147. Bandura, A. (1986) Social Foundations of Thought and Action: A Social Cognitive Theory. Prentice-Hall, Englewood CUffs, NJ. Baranowski, T. (1990) Reciprocal determinism at the stages of behavior change: an integration of community, personal and behavioral perspectives. International Quarterly of Community Health Education, 10, 297-327. Baranowski, T., Domel, S., Gould, IL, Baranowski, J., Leonard, S., Treiber, F. and Mullis, R. (1993) Increasing fruit and vegetable consumption among 4th and 5th grade students: results from focus groups using reciprocal determinism. Journal of Nutrition Education, 25, 114-120. Birch, L. L. (1979) Preschool children's food preferences and consumption patterns. Journal of Nutrition Education, 11, 189-192. Birch, L. L. and Sullivan, S. A. (1991) Measuring children's food preferences. Journal of School Health, 61, 212-214. Block, G. (1991) Dietary guidelines and the results of food consumption surveys. American Journal of Clinical Nutrition, 53, 356S-357S. Bowen, D. J., Tomoyasu, N., Anderson, M, Carney, M. and Kristal, A. (1992) Effects of expectartciet and personalized feedback on fat consumption, taste, and preferences. Journal of Applied Social Psychology, 22, 1061-1079. Butrum, R. R., Clifford, C. K. and Lanza, E. (1988) NCI dietary guidelines: rationale. American Journal of Clinical Nutrition, 48, 888-S95. Calfat, It J., Sallis, J. F. and Nader, P. R. (1991) The development of scales to measure knowledge and preference for diet and physical activity behavior in 4- to 8-year old children. Journal of Developmental and Behavioral Pediatrics, 12, 185-190. Committee on Diet and Health, Food and Nutrition Board. (1989) Diet and Health. Implications for Reducing Chronic Disease Risk. National Research Council, Washington, DC, pp. 670-672. Contento, I. R. (1991) Children's dietary knowledge, skills, and attitudes: measurement issues. Journal of School Health, 61, 208-211. Domel, S. B., Baranowski, T, Davis, H., Leonard, S. B., Riley, P. and Baranowski, J. (1993a) Measuring fruit and vegetable preferences among fourth andfifthgrade students. Preventive Medicine, 22, 866-879. Domel, S. B., Baranowski, T., Davis, H., Thompson, W. O., Leonard, S. B., Riley, P., Baranowski, J., Dudovitz, B. and Smyth, M. (1993b) Development and evaluation of a school intervention to increase fruit and vegetable consumption among 4th and 5th grade students. Journal of Nutrition Education, 25, 345-349. Domel, S. B, Baranowski, T., Davit, H., Leonard, S. B., Riley, P. and Baranowski, J. (1994a) Fruit and vegetable food frequencies by fourth and fifth grade students: validity and reliability. Journal of the American College of Nutrition, 13, 33-39. Domel, S. B., Baranowski. T., Leonard, S. B., Davis, H., Riley,

307

S.B.Domel et al.

P. and Baranowski, J. (1994b) Accuracy of fourth and fifth grade students' food records compared to school lunch observations. American Journal of Clinical Nutrition, 59 (Suppl.), 218S-22OS. Domel, S. B., Baranowslci, T, Davis, H. C , Thompson, W. O., Leonard, S. B. and Baranowski, J. (1995) A measure of outcome expectations for fruit and vegetable consumption among fourth and fifth grade children: reliability and validity. Health Education Research, 10, 65-72. Goldman, M. SM Brown, S. A. and Christiansen, B. A. (1987) Expectancy theory: thinking about drinking. In Blane, H. T. and Leonard, K. E. (eds), Psychological-Theories of Drinking and Alcoholism. Guilford Press, New York, pp. 181-226. Havas, S., Heimendinger, J., Damron, D., Nicklas, T. A., Cowan, A., Beresford, S. A., Sorensen, G, Buller, D., Bishop, D., Baranowksi, T. and Reynolds, K. (1995) 5 A Day for Better Health—nine community research projects to increase fruit and vegetable consumption. Public Health Reports, 110, 68-79. Heam, M. D., Baranowski, T., Baranowski, J., Doyle, C, Lin, L. S., Smith, M., Wang, D. T. and Resnicow, K. (1996) Environmental determinants of behavior among children: availability and accessibility of fruits and vegetables. Journal of Health Education, in press Kelder, S. D., Perry, C. U, Klepp. K. I. and Lytle, L. L. (1994) Longitudinal tracking of adolescent smoking, physical activity, and food choice behaviors. American Journal of Public Health, 84, 1121-1126. Michela, J. L. and Contento, I. R. (1986) Cognitive, motivational, social, and environmental influences on children's food choices. Health Psychology, 5, 209-230. Parcel, G. S., Simons-Morton, B., O'Hara, N. M., Baranowski, T. and Wilson, B. (1989) School promotion of healthful diet and physical activity, impact on learning outcomes and selfreported behavior. Health Education Quarterly, 16, 181-189. Patterson, B. H., Block, G., Rosenberger. W. E, Pee, D. and Kahle, L. L. (1990) Fruit and vegetables in the American diet: data from the NHANES II Survey. American Journal of Public Health, 80, 1443-1449.

308

Resnicow, K., Baranowski, T, Heam, M. D., Lin, L. S., Smith, M., Wang, D. T., Baranowski, J. and Doyle, C. (19%) Psychosocial correlates of fruit and vegetable consumption, Health Psychology, in press. Satter, E. M. (1983; expanded edition, 1986) Child of Mine: Feeding with Love and Good Sense. Bull, Palo Alto, CA. Scher, S. J. and Cooper, J. (1989) Motivational basis of dissonance: the singular role of behavioral consequences. Journal of Personality and Social Psychology, 56, 899-906. Strecher, V. J., McEvoy De Vellis, B., Becker, M. H. and Rosenstock, L M. (1986) The role of self-efficacy in achieving health behavior change. Health Education Quarterly, 13, 73-91. Subar, A. S., Heimendinger, J., Krebs-Smith, S. M_ Patterson, B. H., Kessler, R. and Pivonka, E. (1992) 5 A Day for Better Health: A Baseline Study of Americans' Fruit and Vegetable Consumption. National Cancer Institute, RockviUe, MD. Surgeon General (1988) Report on Nutrition and Health. DHHS (PHS) publ. 88-50210. US Department of Health and Human Services, Public Health Service, Washington, DC, pp. 9-13. US Department of Agriculture and Human Nutrition Information Service (1992) The Food Guide Pyramid. Home and Garden Bulletin 252. US Department of Agriculture and US Department of Health and Human Services (1990) Nutrition and Your Health: Dietary Guidelines for Americans, 3rd edn. Washington, DC. US Department of Health and Human Services (1991) HealthyPeople 2000: National Health Promotion and Disease Prevention Objectives. Public Health Service, Washington, DC, pp. 117-119. Wolfe, W. S. and Campbell, C. C. (1993) Food pattern, diet quality, and related characteristics of schoolchildren in New York State. Journal of the American Dietetic Association, 93, 1280-1284.

Received on January I, 1995: accepted on December 30, 1995