- Email: [email protected]
Adolescents' Reported Usage of Selected Foods in Relation to Their Perceptions and Social Norms for Those Foods
Appetite, 1996, 27, 109–117
Adolescents’ Reported Usage of Selected Foods in Relation to Their Perceptions and Social Norms for Those Foods
DAVID R. WOODWARD and JUDY A. BOON Biochemistry Department, University of Tasmania, Hobart, Tasmania 7001, Australia
FIONA J. CUMMING School of Nutrition and Public Health, Deakin University, Geelong, Victoria 3217, Australia
PETER J. BALL, HEATHER M. WILLIAMS and HELEN HORNSBY Psychology Department, University of Tasmania, Hobart, Tasmania 7001, Australia
This paper reports a study of factors associated with the consumption of 22 diverse foods among 2082 students aged 12–15 years in Tasmania, Australia. Students provided data by completing a printed questionnaire. Multiple linear regression was used to assess how well self-reported usual (past) frequency of consumption of a food could be estimated from their perceptions of the qualities of the food (their liking for it, its healthfulness) and their descriptive social norms for the food (its usage by parents and its usage by friends). For all foods, multiple R was significant (p<0·0001), with values ranging from 0·28–0·64 (median for the 22 foods being 0·49). Standardized regression coefficients for liking and parental usage were highly significant (p<0·0001) for all foods. However, consistently lower coefficients were found for friends’ usage (11 with p<0·01, but only three with p<0·0001) and healthfulness (five with p<0·01, but only one with p<0·0001). The absolute and relative magnitudes of the regression coefficients varied substantially between the foods. Liking was a stronger predictor than parental usage for 11 of the foods, and parental usage out-ranked liking for the remaining 11. Regression coefficients for friends’ usage and healthfulness were smaller than those for liking and parental usage for all foods studied. 1996 Academic Press Limited
Funding for this study was generously provided by the National Heart Foundation of Australia. We acknowledge with gratitude the assistance of the Tasmanian Department of Education and the Arts, and of the various schools that participated; Jane Kidd and several other health educators; research assistants Ingrid Ganley, Maurice Gourley, Janet Haines, Maria Hennessey, Pauline O’Connor, Mark Stanton; research collaborators Elma Wobma (Landbouw-universiteit, Wageningen), Tim and Greg Woodward, Michael and Melissa Boon, Jessica Ball, Susan Doust, Kate Williams and Terry Dwyer. The protocol for this study was approved by the University of Tasmania’s Committee on Ethical Aspects of Human Experimentation. Correspondence to: Dr David R. Woodward, Biochemistry Department, University of Tasmania, Hobart, Tasmania 7001, Australia. 0195–6663/96/050109+09 $25.00/0
1996 Academic Press Limited
110
D. R. WOODWARD ET AL.
I Understanding how people make dietary decisions would make us better equipped to intervene effectively when we wished to encourage health-promoting food choices or discourage choices that increase the risk of disease. The present study focussed on junior high school students (11–16 years old). This age-group is of considerable interest for health education purposes, because they are gradually forming their adult behaviour patterns, which are likely to persist. Moreover, they are still in fulltime schooling, and easily reached by classroom instruction on health matters. Our aim was to assess the relative importance of selected factors in the conceptualization of dietary choices by secondary students, using a retrospective design. Dietary choices were defined in terms of the self-reported usual frequency of consumption of 22 diverse foods. A food-based approach seemed more appropriate than a nutrient-based approach, as students (like most people) usually seek primarily to eat foods rather than to ingest nutrients. Having chosen a food-based approach, we examined each food separately, on the hypothesis that the pattern of associations may vary substantially between foods. The concepts we chose to explore included perceptions of the qualities of the food itself (the students’ liking for each food, and their perception of its healthfulness) and their descriptive social norms for the food (expressed as the frequency of consumption of each food by parents, and—separately—by friends). All these concepts have previously been suggested in the literature as being involved in teenage dietary choices, though with varying amounts of evidence. However, no previous study has looked at their assessment simultaneously.
M A detailed report on subjects and methods has already appeared (Williams, Woodward, Ball, Cumming, Hornsby & Boon, 1993). However, for convenience, the important aspects are summarized. Respondents Tasmania is an island state of Australia. It has a strongly decentralized population of just under half a million. Attendance at school is compulsory until a child’s sixteenth birthday, but students may attend either a “state” school (operated by the state government) or a “private” (non-government) school. The target population was students in the junior secondary grades 7 (modal age 12 years), 8 (age 13), 9 (age 14) and 10 (age 15) of Tasmanian schools. As at July 1990, the size of this target population was 27 412, distributed across 87 schools (Australian Bureau of Statistics, 1991). A two-stage cluster sampling technique was used. The first stage was the selection of schools for study. Twenty five of the 87 eligible schools were selected by random sampling in which the probability of a school being selected was proportional to its enrolment in these grades. Ten schools declined to take part, and were replaced by other schools similarly selected. In the second stage, the principal of each participating school was asked to select from each grade one class group of students of mixed ability who had not received unusual amounts of health/nutrition education. Thus,
ADOLESCENT FOOD CHOICES
111
at each school, the sample comprised approximately 20 students from each of the four grades—approximately 80 per school.
Food Terms Some of the food terms used in this paper may not be readily recognized or correctly interpreted outside Australia and so a translation is provided below. In addition, certain foods were described in the questionnaire in terms that limited their generality: such qualifications are also noted below. “Potato” was described in the instrument as “boiled mashed potato”, a frequent method of preparation in Tasmania, and would therefore exclude potato prepared in other ways. “Hot chips” are known in many parts of the world as “French fries”, and “biscuits” are known in North America as “cookies”. “Soft drinks” are flavoured carbonated beverages, and “cordials” are concentrated flavoured syrups that are diluted with water; the two were linked together on the instrument because the term “cordial” is often used for both in Tasmania. Milk sold in Tasmania is “full-cream” (3·8% fat—the most widely-sold type), “reduced-fat” (1·2% fat) or “skim” (<0·1% fat). Margarine sold in Tasmania is labelled “polyunsaturated” if, and only if, polyunsaturated fatty acids comprise at least 40% of total fatty acids. At the time of our study, margarine not so labelled contained predominantly saturated fatty acids. Specially prepared posters were used to enable students to discriminate between the two types of milk and the two types of margarine mentioned in the instrument. These showed full-size colour pictures of all commercial milk and margarine containers sold in this State. Full-cream and “low-fat” (= “reduced fat” and “skim”) containers were carefully distinguished on these posters, as were polyunsaturated margarines and other margarines. Flavoured milk, as well as “plain” milk, was included in the full-cream milk category: at the time of the survey, the only flavoured milks on sale in Tasmania were full-cream. The instrument referred to a “glass of . . . milk”, thus referring only to milk used as a beverage.
Questionnaire Instrument The development of the instrument was based on “focus groups” with junior secondary students, and discussions with experienced school health educators. This qualitative phase identified key concepts that seemed to be significant in decisionmaking about foods: two we considered to be perceptions (liking for the food, and its perceived heathfulness) and two were important reference groups through whom social norms were conveyed (parents and friends). The final instrument was a 12-page printed booklet, which for each of 22 commonly-available foods asked about the respondent’s usual (past) frequency of usage of the food, perceptions of the qualities of the food itself (how much the student liked the food, and how healthy the student thought it was to eat the food), and descriptive social norms for the food (how often they thought it was eaten by the adults living in the home, and by their friends). It also included a smaller number of other questions, mainly on socio-demographic background. Its reliability was assessed by test–retest with an interval of two weeks with 41 students: the mean within-individual across-items correlation was 0·72 (p<0·01).
112
D. R. WOODWARD ET AL.
Procedure The protocol for the survey was approved by the University of Tasmania’s Committee on Ethical Aspects of Human Experimentation. The printed instrument (described above) was administered to each class group, under the supervision of team members. After a standard spoken introduction, students were allowed 30 min to fill in their own copy of the instrument, with help from team members where needed. Names of respondents were not recorded. Because of possible seasonal variations, it should be noted that the study took place in mid-winter. During data entry, extensive checking indicated an error rate no greater than 0·5% of data entries. Statistical analyses were carried out using SPSS/PC+, version 4.0.
R Respondents A total of 2099 students took part in the study. Results from 17 students were excluded from analysis due to literacy problems or manifestly unreliable results. The remaining 2082 respondents, whose data are presented here, appeared representative of the total population of Tasmanian students in grades 7–10 (Australian Bureau of Statistics, 1991). Thus, 52% of the sample were male (compared to 51% of the target population); 25% (25%) were in each of the four grades studied; students from schools in the southern administrative region comprised 50% (46%) of the total. Students’ Reported Usage and Perceptions of Foods The instrument which students were asked to complete covered 22 foods. All were reasonably common food items: they included cereals, fruits and vegetables, meats, dairy products, fatty spreads and snack items. For most of these, mean consumption frequency reported was between one and four days per week, although higher frequencies were found for bread, breakfast cereals, and soft drinks (Table 1). Mean liking ratings (from 1=“hate it” to 5=“love it”) ranged from 2·7 for nonpolyunsaturated margarine to 4·4 for orange juice, with a median of 3·8 (Table 1). Mean healthfulness ratings (from 1=“very unhealthy” to 5=“very healthy”) ranged from 1.8 for hot chips to 4·8 for apple, with a median of 3·7. Ratings of perceived parental usage (from 1=“very rarely” to 5=“very often”) ranged from 2·2 for meat pie to 4·3 for bread, with a median of 3·4. Those for perceived friends’ usage (same scale) ranged from 2·8 for low-fat milk to 4·3 for bread, with a median of 3·7. Associations Between Reported Consumption and Perceptions of Foods Multiple linear regression was used to explore how well reported consumption frequency of a particular food could be predicted from the four above ratings (entered simultaneously). R values were highly significant (p<0·0001) for every food (Table 2), ranging from 0·28–0·64, with a median of 0·49.
113
ADOLESCENT FOOD CHOICES
T 1 Reported consumption frequency (days of usage in an average week), and ratings of liking, healthfulness, parental usage and friends’ usage, for the selected foods among Tasmanian junior high school students Food Spreads Butter Margarine, polyunsaturated Margarine, other Snacks Meat pie Soft drinks Hot chips Meats Steak Sausages Chicken Lamb Fruits and vegetables Apple Orange juice Potato Tomato Bread and cereals Bread Cake Biscuit Breakfast cereal Dairy products Milk, full-cream Milk, low-fat Cheese Ice-cream
Usage M SD
Liking M SD
Health M SD
Parents M SD
Friends M SD
2·7 2·8
3·0 1·2
2·6 1·1
2·9 1·5
3·1 1·2
3·6 2·9 2·1 2·7
2·9 1·1 2·7 1·1
3·4 1·1 2·5 1·0
3·3 1·4 2·7 1·4
3·3 1·2 3·2 1·2
1·1 1·2 4·4 2·4 1·5 1·3
3·6 1·3 4·2 0·9 4·2 1·0
2·3 1·0 2·2 1·1 1·8 1·0
2·2 1·1 2·7 1·4 2·4 1·2
3·6 1·2 4·2 0·9 3·8 1·1
1·7 1·3 1·5 1·6
1·2 1·1 1·1 1·1
3·9 3·4 4·3 3·8
1·0 1·2 0·9 1·0
3·7 3·0 3·8 3·7
0·9 1·0 0·9 0·9
3·9 3·5 3·7 3·8
1·0 1·1 0·9 1·0
3·7 3·5 3·8 3·7
0·9 1·0 0·9 0·9
3·8 3·3 4·0 1·7
2·2 2·5 1·9 1·6
4·2 4·4 3·7 3·2
0·9 0·9 1·1 1·4
4·8 4·7 4·0 4·4
0·5 0·6 0·9 0·8
3·4 3·4 4·2 3·7
1·2 1·3 0·9 1·0
3·8 3·7 3·7 3·0
1·1 1·0 1·0 1·1
5·9 1·8 3·5 4·6
1·6 1·6 2·2 2·6
3·8 4·0 3·9 3·8
0·8 0·9 0·9 1·1
4·2 2·0 2·5 4·1
0·7 0·9 0·9 0·8
4·3 2·8 3·1 3·5
0·8 1·1 1·2 1·4
4·3 3·7 3·9 4·0
0·8 1·0 0·9 1·0
4·0 1·8 2·8 2·4
2·7 2·6 1·9 1·9
4·0 2·8 3·8 4·1
1·1 1·4 1·1 1·0
3·6 4·4 4·0 2·3
1·0 0·9 0·9 1·1
3·0 2·6 3·6 2·8
1·4 1·5 1·0 1·3
3·9 2·8 3·6 3·9
1·1 1·3 1·0 1·0
See Food Terms in Method for details of the foods rated. Scores for liking, health, parents and friends ranged from 1 (respectively “hate it”, “very unhealthy”, “very rarely”, “very rarely”) to 5 (respectively “love it”, “very healthy”, “very often”, “very often”).
Liking was a statistically significant predictor (p<0·0001) for every food, coefficients ranging from 0·16 for bread to 0·55 for breakfast cereal, with a median of 0·27. All coefficients were positive, indicating that well-liked foods were reportedly eaten more often. Perceived parental usage was also a significant (p<0·0001) predictor for every food, coefficients ranging from 0·09 to 0·46, with a median of 0·25. All were positive, i.e., foods thought to be eaten frequently by parents tended to be reported as eaten frequently by our respondents. Friends’ usage was a statistically significant (p<0·01) predictor of frequency scores for 11 of the foods; but only three foods reached p<0·0001. Coefficients ranged from −0·01–0·13, with a median of 0·06; only values of 0·06 or greater achieved p<0·01.
114
D. R. WOODWARD ET AL.
T 2 Prediction of frequency of consumption of specific foods from ratings of liking, healthfulness, parental usage and friends’ usage, using multiple linear regression Standardized regression coefficients for
Spreads Butter Margarine, polyunsaturated Margarine, other Snacks Meat pie Soft drinks Hot chips Meats Steak Sausages Chicken Lamb Fruits and vegetables Apple Orange juice Potato Tomato Bread and cereals Bread Cake Biscuit Breakfast cereal Dairy products Milk, full-cream Milk, low-fat Cheese Ice-cream
Multiple R
Liking
Health
Parents
Friends
0·59∗∗∗
0·21∗∗∗
0·01
0·43∗∗∗
0·09∗∗
0·64∗∗∗ 0·59∗∗∗
0·30∗∗∗ 0·21∗∗∗
−0·04 0·06
0·46∗∗∗ 0·42∗∗∗
0·00 0·02
0·50∗∗∗ 0·48∗∗∗ 0·47∗∗∗
0·36∗∗∗ 0·32∗∗∗ 0·18∗∗∗
0·05 0·11∗∗∗ 0·08∗∗
0·14∗∗∗ 0·18∗∗∗ 0·30∗∗∗
0·13∗∗∗ 0·09∗∗ 0·12∗∗∗
0·39∗∗∗ 0·42∗∗∗ 0·32∗∗∗ 0·35∗∗∗
0·22∗∗∗ 0·24∗∗∗ 0·16∗∗∗ 0·18∗∗∗
0·01 0·05 0·02 0·02
0·27∗∗∗ 0·26∗∗∗ 0·24∗∗∗ 0·23∗∗∗
0·49∗∗∗ 0·49∗∗∗ 0·46∗∗∗ 0·57∗∗∗
0·44∗∗∗ 0·22∗∗∗ 0·22∗∗∗ 0·51∗∗∗
0·01 −0·03 −0·02 0·02
0·12∗∗∗ 0·37∗∗∗ 0·33∗∗∗ 0·13∗∗∗
0·05 0·06∗ 0·08∗∗ 0·03
0·28∗∗∗ 0·44∗∗∗ 0·45∗∗∗ 0·59∗∗∗
0·16∗∗∗ 0·19∗∗∗ 0·30∗∗∗ 0·55∗∗∗
0·04 0·09∗∗ 0·06 0·00
0·12∗∗∗ 0·28∗∗∗ 0·19∗∗∗ 0·09∗∗∗
0·11∗∗∗ 0·08∗∗ 0·08∗∗ 0·07∗∗
0·53∗∗∗ 0·62∗∗∗ 0·51∗∗∗ 0·51∗∗∗
0·38∗∗∗ 0·46∗∗∗ 0·47∗∗∗ 0·30∗∗∗
0·09∗∗ 0·04 −0·01 0·07∗
0·22∗∗∗ 0·27∗∗∗ 0·12∗∗∗ 0·28∗∗∗
0·03 0·00 0·02 0·08∗∗
0·01 0·00 −0·01 0·03
The regression equation included liking, perceived healthfulness, parental usage and friends’ usage as simultaneous predictors· Separate equations were calculated for each food. ∗ p<0·01; ∗∗ p<0·001; ∗∗∗ p<0·0001.
All coefficients with p<0·01 were positive, i.e., foods thought to be eaten frequently by friends tended to be eaten frequently as reported by our respondents. Perceived healthfulness was a statistically significant (p<0·01) predictor for only five foods, of which only one reached p<0·0001. Coefficients varied markedly between foods, ranging from −0·04–0·11, with a median of 0·03; only values of 0·07 or greater achieved p<0·01. All statistically significant coefficients were positive, indicating that foods perceived as healthy tended to be eaten frequently by our respondents (or perhaps that foods perceived as unhealthy tended to be avoided).
ADOLESCENT FOOD CHOICES
115
Variations in Association Among Foods Liking and parental usage were the two strongest predictors of reported frequency of consumption for all the foods studied, but the relative strength of association varied among foods. Liking had the stronger association for 11 foods, including all four of the dairy product items, three of the four bread and cereal items, and two of the three snack items. Parental usage had the stronger association for the remaining eleven foods, including all four meat items and all three spreads. Friends’ usage was a significant predictor mainly for items likely to be eaten away from formal meals, such as meat pie, bread, and hot potato chips. Healthfulness was a significant predictor mainly for “unhealthy” foods, such as cakes, ice cream, soft drinks and hot chips. Possible Deficiencies in the Data To ascertain whether associations between predictor variables could have distorted the results, we estimated the variance inflation factor (VIF) for every predictor term in each regression. The 88 VIFs ranged from 1·04–1·42, suggesting that multicollinearity was not a problem. The results presented above came from respondents who gave analysable responses for all variables discussed for a particular food: these ranged from 1295 (62%) to 1771 (85%), the median being 1656 (80%). The rating value most commonly omitted by respondents was the estimate of friends’ usage frequency. Lest exclusion of these respondents providing incomplete data might have biased estimates, we repeated the regression analyses, replacing missing values by the corresponding sample means. This altered parameter estimates only slightly, no multiple R values or regression coefficient changing by more than 0·03. Statistical significance was affected in only two instances: for biscuits and sausages, p values for the healthfulness regression coefficients declined from just above 0·01 to just below 0·01.
D Association of Reported Food Usage with Food Perceptions A strong link has been found between liking for a food and the reported frequency of its consumption. This is consistent with previous studies of adults (Randall & Sanjur, 1981; Serville & Bleyer, 1976; Tuorila & Pangborn, 1988), adolescents (Contento, Michela & Goldberg, 1988; Williams et al., 1993) and children (Birch, 1979; Worsley, Baghorst, Worsley, Coonan & Peters, 1984) which have concluded that hedonic factors are an important determinant of continued frequent consumption of a given food. It was found for most foods, however, that the perceived heathfulness of a food is not a good predictor of its reported consumption frequency. Other authors have noted that health issues may influence dietary choice in adults (Serville & Bleyer, 1976; Tuorila & Pangborn, 1988), adolescents (Contento et al., 1988) and children (Worsley et al., 1984), but in general their data suggest a relatively weak effect, consistent with our results. A strong linkage was observed between perceived parental frequency of usage of a food and its frequency of consumption reported by their adolescent offspring.
116
D. R. WOODWARD ET AL.
(We do not claim that these assessments of parental usage necessarily represent the actual frequencies of parental usage; indeed, we deliberately sought to tap the students’ “perceptions” of parental usage, on the grounds that it is perceived usage that it likely to be linked to the offspring’s own usage.) The findings are consistent with those of other authors for children (Hertzler, 1983; Laskarzewski et al., 1980; Pliner & Pelchat, 1986; Worsley et al., 1984) and adolescents (Contento et al., 1988; Laskarzewski et al., 1980), and with reports of moderate correlations between food preference ratings between university students and their parents (Pliner, 1983; Rozin, Fallon & Mandell, 1984). However, a less consistent pattern of linkage was found between perceived friends’ frequency of food consumption and that of respondents. While peer influence on choices has been documented to some extent in young children (Hertzler, 1983) and adults (Cosper & Wakefield, 1975; Schafer & Keith, 1981), there appear to be few studies of it among adolescents (Contento et al., 1988), with Farthing (1991) drawing attention to the paucity of research data on this issue. The relatively weak peer influences detected contrast with the strong peer influences reported for, e.g., smoking (Morgan & Grube, 1991; O’Connell et al., 1981). In common with other authors cited in this section, we assume that liking and other predictors are causes of actual consumption, reflected in the frequency scores. However, cross-sectional designs cannot rule out Bem’s (1967, 1972) suggestion that people infer their attitudes from their behaviour, rather than determine their behaviour by their attitudes. Repeating our investigation using a prospective design and validated consumption assessments could help to confirm the directionality of causation. Different Foods—Different Patterns of Influence The relative strength of associations of food perceptions to reported intake frequency varied substantially from food to food. This is not a new finding: for example, Tuorila and Pangborn (1988), in discussing their study on high-fat foods, comment, “all four foods studied elicited individual sets of beliefs and belief structures, and none of them were rated similar to the generic ‘high-fat foods’.” However, our study included many more foods than in its predecessors and so documented the diversity of foods’ linkages in more detail. This diversity has implications for nutrition promotion, as attempts to motivate changes in consumption are more likely to be successful if they take account of concepts linked to perceived consumption frequency. Perceived parental usage and the student’s liking for the food seemed to be the key concepts in reported frequency of choice. The linkage of parental usage to consumption frequency presumably reflects the impact of parental example, parental advice and/or parental control over what is served at mealtimes. Liking, on the other hand, reflects the direct personal response of the individual to the food. Thus, one can suggest that, for those foods where parental usage is a dominant predictor, the individual is accepting parental control, while for those foods where liking is the dominant predictor, the individual is asserting her or his own identity. In the former cases, attempts to motivate a change in consumption frequency would seem most appropriately targeted towards the teenager; in the latter situation, such attempts would seem likely to fail without parental support. By comparison, the impact of concepts of healthfulness and friends’ usage seems modest, and restricted to a limited number of foods. Attempts to motivate change
ADOLESCENT FOOD CHOICES
117
in consumption frequency could in principle be based on emphasizing health aspects or by developing resistance to peer pressure. Our data suggest that such approaches are, at most, appropriate to only a minority of foods. R Australian Bureau of Statistics. (1991). Schools Australia 1990. Canberra: Australian Bureau of Statistics. Bem, D. (1967). Self-perception: an alternative interpretation of cognitive dissonance phenomena. Psychological Review, 74, 183–200. Bem, D. (1972). Self-perception theory. Advances in Psychological Theory, 6, 1–62. Birch, L. L. (1979). Pre-school children’s preferences and consumption patterns. Journal of Nutrition Education, 11, 189–92. Contento, I. R., Michela, J. L. & Goldberg, C. J. (1988). Food choice among adolescents: population segmentation by motivations. Journal of Nutrition Education, 20, 289–98. Cosper, B. A. & Wakefield, L. M. (1975). Food choices of women. Journal of the American Dietetic Association, 66, 152–5. Farthing, M. C. (1991). Current eating patterns of adolescents in the United States. Nutrition Today, 26(2), 35–9. Hertzler, A. A. (1983). Children’s food patterns—a review. II. Family and group behavior. Journal of the American Dietetic Association, 83, 555–60. Laskarzewski, P., Morrison, J. A., Khoury, P., Kelly, K., Glatfelter, L., Larsen, R. & Glueck, C. J. (1980). Parent-child nutrient intake interrelationships in school children ages 6–19: The Princeton School District Study. American Journal of Clinical Nutrition, 33, 2350–5. Morgan, M. & Grube, J. W. (1991). Closeness and peer group influence. British Journal of Social Psychology, 30, 159–69. O’Connell, D. L., Alexander, H. M., Dobson, A. J., Lloyd, D. M., Hardes, G. R., Springthorpe, H. J. & Leeder, S. R. (1981). Cigarette smoking and drug use in schoolchildren. II. Factors associated with smoking. International Journal of Epidemiology, 10, 223–31. Pliner, P. (1983). Family resemblance in food preferences. Journal of Nutrition Education, 15, 137–40. Pliner, P. & Pelchat, M. L. (1986). Similarities in food preferences between children and their siblings and parents. Appetite, 7, 333–42. Randall, E. & Sanjur, D. (1981). Food preferences—their conceptualization and relationship to consumption. Ecology of Food and Nutrition, 11, 151–61. Rozin, P., Fallon, A. E. & Mandell, R. (1984). Family resemblance in attitudes to food. Developmental Psychology, 20, 309–14. Schafer, R. B. & Keith, P. M. (1981). Influences on food decisions across the family life cycle. Journal of the American Dietetic Association, 78, 152–5. Serville, Y. & Bleyer, R. (1976). L’apport des enqueˆtes de motivation pour la connaissance du comportement alimentaire. Annales de la Nutrition et de l’Alimentation, 309, 195–210. Tuorila, H. & Pangborn, R. M. (1988). Prediction of reported consumption of selected fatcontaining foods. Appetite, 11, 81–95. Williams, H. M., Woodward, D. R., Ball, P. J., Cumming, F. J., Hornsby, H. & Boon, J. A. (1993). Food perceptions and food consumption among Tasmanian high school students. Australian Journal of Nutrition and Dietetics, 50, 156–63. Worsley, A., Baghurst, P., Worsley, A. J., Coonan, W. & Peters, M. (1984). Australian tenyear-olds’ perceptions of food. 1. Sex differences. Ecology of Food and Nutrition, 15, 231–46.
Received 12 August 1994, revision 21 February 1996
Adolescents’ Reported Usage of Selected Foods in Relation to Their Perceptions and Social Norms for Those Foods
DAVID R. WOODWARD and JUDY A. BOON Biochemistry Department, University of Tasmania, Hobart, Tasmania 7001, Australia
FIONA J. CUMMING School of Nutrition and Public Health, Deakin University, Geelong, Victoria 3217, Australia
PETER J. BALL, HEATHER M. WILLIAMS and HELEN HORNSBY Psychology Department, University of Tasmania, Hobart, Tasmania 7001, Australia
This paper reports a study of factors associated with the consumption of 22 diverse foods among 2082 students aged 12–15 years in Tasmania, Australia. Students provided data by completing a printed questionnaire. Multiple linear regression was used to assess how well self-reported usual (past) frequency of consumption of a food could be estimated from their perceptions of the qualities of the food (their liking for it, its healthfulness) and their descriptive social norms for the food (its usage by parents and its usage by friends). For all foods, multiple R was significant (p<0·0001), with values ranging from 0·28–0·64 (median for the 22 foods being 0·49). Standardized regression coefficients for liking and parental usage were highly significant (p<0·0001) for all foods. However, consistently lower coefficients were found for friends’ usage (11 with p<0·01, but only three with p<0·0001) and healthfulness (five with p<0·01, but only one with p<0·0001). The absolute and relative magnitudes of the regression coefficients varied substantially between the foods. Liking was a stronger predictor than parental usage for 11 of the foods, and parental usage out-ranked liking for the remaining 11. Regression coefficients for friends’ usage and healthfulness were smaller than those for liking and parental usage for all foods studied. 1996 Academic Press Limited
Funding for this study was generously provided by the National Heart Foundation of Australia. We acknowledge with gratitude the assistance of the Tasmanian Department of Education and the Arts, and of the various schools that participated; Jane Kidd and several other health educators; research assistants Ingrid Ganley, Maurice Gourley, Janet Haines, Maria Hennessey, Pauline O’Connor, Mark Stanton; research collaborators Elma Wobma (Landbouw-universiteit, Wageningen), Tim and Greg Woodward, Michael and Melissa Boon, Jessica Ball, Susan Doust, Kate Williams and Terry Dwyer. The protocol for this study was approved by the University of Tasmania’s Committee on Ethical Aspects of Human Experimentation. Correspondence to: Dr David R. Woodward, Biochemistry Department, University of Tasmania, Hobart, Tasmania 7001, Australia. 0195–6663/96/050109+09 $25.00/0
1996 Academic Press Limited
110
D. R. WOODWARD ET AL.
I Understanding how people make dietary decisions would make us better equipped to intervene effectively when we wished to encourage health-promoting food choices or discourage choices that increase the risk of disease. The present study focussed on junior high school students (11–16 years old). This age-group is of considerable interest for health education purposes, because they are gradually forming their adult behaviour patterns, which are likely to persist. Moreover, they are still in fulltime schooling, and easily reached by classroom instruction on health matters. Our aim was to assess the relative importance of selected factors in the conceptualization of dietary choices by secondary students, using a retrospective design. Dietary choices were defined in terms of the self-reported usual frequency of consumption of 22 diverse foods. A food-based approach seemed more appropriate than a nutrient-based approach, as students (like most people) usually seek primarily to eat foods rather than to ingest nutrients. Having chosen a food-based approach, we examined each food separately, on the hypothesis that the pattern of associations may vary substantially between foods. The concepts we chose to explore included perceptions of the qualities of the food itself (the students’ liking for each food, and their perception of its healthfulness) and their descriptive social norms for the food (expressed as the frequency of consumption of each food by parents, and—separately—by friends). All these concepts have previously been suggested in the literature as being involved in teenage dietary choices, though with varying amounts of evidence. However, no previous study has looked at their assessment simultaneously.
M A detailed report on subjects and methods has already appeared (Williams, Woodward, Ball, Cumming, Hornsby & Boon, 1993). However, for convenience, the important aspects are summarized. Respondents Tasmania is an island state of Australia. It has a strongly decentralized population of just under half a million. Attendance at school is compulsory until a child’s sixteenth birthday, but students may attend either a “state” school (operated by the state government) or a “private” (non-government) school. The target population was students in the junior secondary grades 7 (modal age 12 years), 8 (age 13), 9 (age 14) and 10 (age 15) of Tasmanian schools. As at July 1990, the size of this target population was 27 412, distributed across 87 schools (Australian Bureau of Statistics, 1991). A two-stage cluster sampling technique was used. The first stage was the selection of schools for study. Twenty five of the 87 eligible schools were selected by random sampling in which the probability of a school being selected was proportional to its enrolment in these grades. Ten schools declined to take part, and were replaced by other schools similarly selected. In the second stage, the principal of each participating school was asked to select from each grade one class group of students of mixed ability who had not received unusual amounts of health/nutrition education. Thus,
ADOLESCENT FOOD CHOICES
111
at each school, the sample comprised approximately 20 students from each of the four grades—approximately 80 per school.
Food Terms Some of the food terms used in this paper may not be readily recognized or correctly interpreted outside Australia and so a translation is provided below. In addition, certain foods were described in the questionnaire in terms that limited their generality: such qualifications are also noted below. “Potato” was described in the instrument as “boiled mashed potato”, a frequent method of preparation in Tasmania, and would therefore exclude potato prepared in other ways. “Hot chips” are known in many parts of the world as “French fries”, and “biscuits” are known in North America as “cookies”. “Soft drinks” are flavoured carbonated beverages, and “cordials” are concentrated flavoured syrups that are diluted with water; the two were linked together on the instrument because the term “cordial” is often used for both in Tasmania. Milk sold in Tasmania is “full-cream” (3·8% fat—the most widely-sold type), “reduced-fat” (1·2% fat) or “skim” (<0·1% fat). Margarine sold in Tasmania is labelled “polyunsaturated” if, and only if, polyunsaturated fatty acids comprise at least 40% of total fatty acids. At the time of our study, margarine not so labelled contained predominantly saturated fatty acids. Specially prepared posters were used to enable students to discriminate between the two types of milk and the two types of margarine mentioned in the instrument. These showed full-size colour pictures of all commercial milk and margarine containers sold in this State. Full-cream and “low-fat” (= “reduced fat” and “skim”) containers were carefully distinguished on these posters, as were polyunsaturated margarines and other margarines. Flavoured milk, as well as “plain” milk, was included in the full-cream milk category: at the time of the survey, the only flavoured milks on sale in Tasmania were full-cream. The instrument referred to a “glass of . . . milk”, thus referring only to milk used as a beverage.
Questionnaire Instrument The development of the instrument was based on “focus groups” with junior secondary students, and discussions with experienced school health educators. This qualitative phase identified key concepts that seemed to be significant in decisionmaking about foods: two we considered to be perceptions (liking for the food, and its perceived heathfulness) and two were important reference groups through whom social norms were conveyed (parents and friends). The final instrument was a 12-page printed booklet, which for each of 22 commonly-available foods asked about the respondent’s usual (past) frequency of usage of the food, perceptions of the qualities of the food itself (how much the student liked the food, and how healthy the student thought it was to eat the food), and descriptive social norms for the food (how often they thought it was eaten by the adults living in the home, and by their friends). It also included a smaller number of other questions, mainly on socio-demographic background. Its reliability was assessed by test–retest with an interval of two weeks with 41 students: the mean within-individual across-items correlation was 0·72 (p<0·01).
112
D. R. WOODWARD ET AL.
Procedure The protocol for the survey was approved by the University of Tasmania’s Committee on Ethical Aspects of Human Experimentation. The printed instrument (described above) was administered to each class group, under the supervision of team members. After a standard spoken introduction, students were allowed 30 min to fill in their own copy of the instrument, with help from team members where needed. Names of respondents were not recorded. Because of possible seasonal variations, it should be noted that the study took place in mid-winter. During data entry, extensive checking indicated an error rate no greater than 0·5% of data entries. Statistical analyses were carried out using SPSS/PC+, version 4.0.
R Respondents A total of 2099 students took part in the study. Results from 17 students were excluded from analysis due to literacy problems or manifestly unreliable results. The remaining 2082 respondents, whose data are presented here, appeared representative of the total population of Tasmanian students in grades 7–10 (Australian Bureau of Statistics, 1991). Thus, 52% of the sample were male (compared to 51% of the target population); 25% (25%) were in each of the four grades studied; students from schools in the southern administrative region comprised 50% (46%) of the total. Students’ Reported Usage and Perceptions of Foods The instrument which students were asked to complete covered 22 foods. All were reasonably common food items: they included cereals, fruits and vegetables, meats, dairy products, fatty spreads and snack items. For most of these, mean consumption frequency reported was between one and four days per week, although higher frequencies were found for bread, breakfast cereals, and soft drinks (Table 1). Mean liking ratings (from 1=“hate it” to 5=“love it”) ranged from 2·7 for nonpolyunsaturated margarine to 4·4 for orange juice, with a median of 3·8 (Table 1). Mean healthfulness ratings (from 1=“very unhealthy” to 5=“very healthy”) ranged from 1.8 for hot chips to 4·8 for apple, with a median of 3·7. Ratings of perceived parental usage (from 1=“very rarely” to 5=“very often”) ranged from 2·2 for meat pie to 4·3 for bread, with a median of 3·4. Those for perceived friends’ usage (same scale) ranged from 2·8 for low-fat milk to 4·3 for bread, with a median of 3·7. Associations Between Reported Consumption and Perceptions of Foods Multiple linear regression was used to explore how well reported consumption frequency of a particular food could be predicted from the four above ratings (entered simultaneously). R values were highly significant (p<0·0001) for every food (Table 2), ranging from 0·28–0·64, with a median of 0·49.
113
ADOLESCENT FOOD CHOICES
T 1 Reported consumption frequency (days of usage in an average week), and ratings of liking, healthfulness, parental usage and friends’ usage, for the selected foods among Tasmanian junior high school students Food Spreads Butter Margarine, polyunsaturated Margarine, other Snacks Meat pie Soft drinks Hot chips Meats Steak Sausages Chicken Lamb Fruits and vegetables Apple Orange juice Potato Tomato Bread and cereals Bread Cake Biscuit Breakfast cereal Dairy products Milk, full-cream Milk, low-fat Cheese Ice-cream
Usage M SD
Liking M SD
Health M SD
Parents M SD
Friends M SD
2·7 2·8
3·0 1·2
2·6 1·1
2·9 1·5
3·1 1·2
3·6 2·9 2·1 2·7
2·9 1·1 2·7 1·1
3·4 1·1 2·5 1·0
3·3 1·4 2·7 1·4
3·3 1·2 3·2 1·2
1·1 1·2 4·4 2·4 1·5 1·3
3·6 1·3 4·2 0·9 4·2 1·0
2·3 1·0 2·2 1·1 1·8 1·0
2·2 1·1 2·7 1·4 2·4 1·2
3·6 1·2 4·2 0·9 3·8 1·1
1·7 1·3 1·5 1·6
1·2 1·1 1·1 1·1
3·9 3·4 4·3 3·8
1·0 1·2 0·9 1·0
3·7 3·0 3·8 3·7
0·9 1·0 0·9 0·9
3·9 3·5 3·7 3·8
1·0 1·1 0·9 1·0
3·7 3·5 3·8 3·7
0·9 1·0 0·9 0·9
3·8 3·3 4·0 1·7
2·2 2·5 1·9 1·6
4·2 4·4 3·7 3·2
0·9 0·9 1·1 1·4
4·8 4·7 4·0 4·4
0·5 0·6 0·9 0·8
3·4 3·4 4·2 3·7
1·2 1·3 0·9 1·0
3·8 3·7 3·7 3·0
1·1 1·0 1·0 1·1
5·9 1·8 3·5 4·6
1·6 1·6 2·2 2·6
3·8 4·0 3·9 3·8
0·8 0·9 0·9 1·1
4·2 2·0 2·5 4·1
0·7 0·9 0·9 0·8
4·3 2·8 3·1 3·5
0·8 1·1 1·2 1·4
4·3 3·7 3·9 4·0
0·8 1·0 0·9 1·0
4·0 1·8 2·8 2·4
2·7 2·6 1·9 1·9
4·0 2·8 3·8 4·1
1·1 1·4 1·1 1·0
3·6 4·4 4·0 2·3
1·0 0·9 0·9 1·1
3·0 2·6 3·6 2·8
1·4 1·5 1·0 1·3
3·9 2·8 3·6 3·9
1·1 1·3 1·0 1·0
See Food Terms in Method for details of the foods rated. Scores for liking, health, parents and friends ranged from 1 (respectively “hate it”, “very unhealthy”, “very rarely”, “very rarely”) to 5 (respectively “love it”, “very healthy”, “very often”, “very often”).
Liking was a statistically significant predictor (p<0·0001) for every food, coefficients ranging from 0·16 for bread to 0·55 for breakfast cereal, with a median of 0·27. All coefficients were positive, indicating that well-liked foods were reportedly eaten more often. Perceived parental usage was also a significant (p<0·0001) predictor for every food, coefficients ranging from 0·09 to 0·46, with a median of 0·25. All were positive, i.e., foods thought to be eaten frequently by parents tended to be reported as eaten frequently by our respondents. Friends’ usage was a statistically significant (p<0·01) predictor of frequency scores for 11 of the foods; but only three foods reached p<0·0001. Coefficients ranged from −0·01–0·13, with a median of 0·06; only values of 0·06 or greater achieved p<0·01.
114
D. R. WOODWARD ET AL.
T 2 Prediction of frequency of consumption of specific foods from ratings of liking, healthfulness, parental usage and friends’ usage, using multiple linear regression Standardized regression coefficients for
Spreads Butter Margarine, polyunsaturated Margarine, other Snacks Meat pie Soft drinks Hot chips Meats Steak Sausages Chicken Lamb Fruits and vegetables Apple Orange juice Potato Tomato Bread and cereals Bread Cake Biscuit Breakfast cereal Dairy products Milk, full-cream Milk, low-fat Cheese Ice-cream
Multiple R
Liking
Health
Parents
Friends
0·59∗∗∗
0·21∗∗∗
0·01
0·43∗∗∗
0·09∗∗
0·64∗∗∗ 0·59∗∗∗
0·30∗∗∗ 0·21∗∗∗
−0·04 0·06
0·46∗∗∗ 0·42∗∗∗
0·00 0·02
0·50∗∗∗ 0·48∗∗∗ 0·47∗∗∗
0·36∗∗∗ 0·32∗∗∗ 0·18∗∗∗
0·05 0·11∗∗∗ 0·08∗∗
0·14∗∗∗ 0·18∗∗∗ 0·30∗∗∗
0·13∗∗∗ 0·09∗∗ 0·12∗∗∗
0·39∗∗∗ 0·42∗∗∗ 0·32∗∗∗ 0·35∗∗∗
0·22∗∗∗ 0·24∗∗∗ 0·16∗∗∗ 0·18∗∗∗
0·01 0·05 0·02 0·02
0·27∗∗∗ 0·26∗∗∗ 0·24∗∗∗ 0·23∗∗∗
0·49∗∗∗ 0·49∗∗∗ 0·46∗∗∗ 0·57∗∗∗
0·44∗∗∗ 0·22∗∗∗ 0·22∗∗∗ 0·51∗∗∗
0·01 −0·03 −0·02 0·02
0·12∗∗∗ 0·37∗∗∗ 0·33∗∗∗ 0·13∗∗∗
0·05 0·06∗ 0·08∗∗ 0·03
0·28∗∗∗ 0·44∗∗∗ 0·45∗∗∗ 0·59∗∗∗
0·16∗∗∗ 0·19∗∗∗ 0·30∗∗∗ 0·55∗∗∗
0·04 0·09∗∗ 0·06 0·00
0·12∗∗∗ 0·28∗∗∗ 0·19∗∗∗ 0·09∗∗∗
0·11∗∗∗ 0·08∗∗ 0·08∗∗ 0·07∗∗
0·53∗∗∗ 0·62∗∗∗ 0·51∗∗∗ 0·51∗∗∗
0·38∗∗∗ 0·46∗∗∗ 0·47∗∗∗ 0·30∗∗∗
0·09∗∗ 0·04 −0·01 0·07∗
0·22∗∗∗ 0·27∗∗∗ 0·12∗∗∗ 0·28∗∗∗
0·03 0·00 0·02 0·08∗∗
0·01 0·00 −0·01 0·03
The regression equation included liking, perceived healthfulness, parental usage and friends’ usage as simultaneous predictors· Separate equations were calculated for each food. ∗ p<0·01; ∗∗ p<0·001; ∗∗∗ p<0·0001.
All coefficients with p<0·01 were positive, i.e., foods thought to be eaten frequently by friends tended to be eaten frequently as reported by our respondents. Perceived healthfulness was a statistically significant (p<0·01) predictor for only five foods, of which only one reached p<0·0001. Coefficients varied markedly between foods, ranging from −0·04–0·11, with a median of 0·03; only values of 0·07 or greater achieved p<0·01. All statistically significant coefficients were positive, indicating that foods perceived as healthy tended to be eaten frequently by our respondents (or perhaps that foods perceived as unhealthy tended to be avoided).
ADOLESCENT FOOD CHOICES
115
Variations in Association Among Foods Liking and parental usage were the two strongest predictors of reported frequency of consumption for all the foods studied, but the relative strength of association varied among foods. Liking had the stronger association for 11 foods, including all four of the dairy product items, three of the four bread and cereal items, and two of the three snack items. Parental usage had the stronger association for the remaining eleven foods, including all four meat items and all three spreads. Friends’ usage was a significant predictor mainly for items likely to be eaten away from formal meals, such as meat pie, bread, and hot potato chips. Healthfulness was a significant predictor mainly for “unhealthy” foods, such as cakes, ice cream, soft drinks and hot chips. Possible Deficiencies in the Data To ascertain whether associations between predictor variables could have distorted the results, we estimated the variance inflation factor (VIF) for every predictor term in each regression. The 88 VIFs ranged from 1·04–1·42, suggesting that multicollinearity was not a problem. The results presented above came from respondents who gave analysable responses for all variables discussed for a particular food: these ranged from 1295 (62%) to 1771 (85%), the median being 1656 (80%). The rating value most commonly omitted by respondents was the estimate of friends’ usage frequency. Lest exclusion of these respondents providing incomplete data might have biased estimates, we repeated the regression analyses, replacing missing values by the corresponding sample means. This altered parameter estimates only slightly, no multiple R values or regression coefficient changing by more than 0·03. Statistical significance was affected in only two instances: for biscuits and sausages, p values for the healthfulness regression coefficients declined from just above 0·01 to just below 0·01.
D Association of Reported Food Usage with Food Perceptions A strong link has been found between liking for a food and the reported frequency of its consumption. This is consistent with previous studies of adults (Randall & Sanjur, 1981; Serville & Bleyer, 1976; Tuorila & Pangborn, 1988), adolescents (Contento, Michela & Goldberg, 1988; Williams et al., 1993) and children (Birch, 1979; Worsley, Baghorst, Worsley, Coonan & Peters, 1984) which have concluded that hedonic factors are an important determinant of continued frequent consumption of a given food. It was found for most foods, however, that the perceived heathfulness of a food is not a good predictor of its reported consumption frequency. Other authors have noted that health issues may influence dietary choice in adults (Serville & Bleyer, 1976; Tuorila & Pangborn, 1988), adolescents (Contento et al., 1988) and children (Worsley et al., 1984), but in general their data suggest a relatively weak effect, consistent with our results. A strong linkage was observed between perceived parental frequency of usage of a food and its frequency of consumption reported by their adolescent offspring.
116
D. R. WOODWARD ET AL.
(We do not claim that these assessments of parental usage necessarily represent the actual frequencies of parental usage; indeed, we deliberately sought to tap the students’ “perceptions” of parental usage, on the grounds that it is perceived usage that it likely to be linked to the offspring’s own usage.) The findings are consistent with those of other authors for children (Hertzler, 1983; Laskarzewski et al., 1980; Pliner & Pelchat, 1986; Worsley et al., 1984) and adolescents (Contento et al., 1988; Laskarzewski et al., 1980), and with reports of moderate correlations between food preference ratings between university students and their parents (Pliner, 1983; Rozin, Fallon & Mandell, 1984). However, a less consistent pattern of linkage was found between perceived friends’ frequency of food consumption and that of respondents. While peer influence on choices has been documented to some extent in young children (Hertzler, 1983) and adults (Cosper & Wakefield, 1975; Schafer & Keith, 1981), there appear to be few studies of it among adolescents (Contento et al., 1988), with Farthing (1991) drawing attention to the paucity of research data on this issue. The relatively weak peer influences detected contrast with the strong peer influences reported for, e.g., smoking (Morgan & Grube, 1991; O’Connell et al., 1981). In common with other authors cited in this section, we assume that liking and other predictors are causes of actual consumption, reflected in the frequency scores. However, cross-sectional designs cannot rule out Bem’s (1967, 1972) suggestion that people infer their attitudes from their behaviour, rather than determine their behaviour by their attitudes. Repeating our investigation using a prospective design and validated consumption assessments could help to confirm the directionality of causation. Different Foods—Different Patterns of Influence The relative strength of associations of food perceptions to reported intake frequency varied substantially from food to food. This is not a new finding: for example, Tuorila and Pangborn (1988), in discussing their study on high-fat foods, comment, “all four foods studied elicited individual sets of beliefs and belief structures, and none of them were rated similar to the generic ‘high-fat foods’.” However, our study included many more foods than in its predecessors and so documented the diversity of foods’ linkages in more detail. This diversity has implications for nutrition promotion, as attempts to motivate changes in consumption are more likely to be successful if they take account of concepts linked to perceived consumption frequency. Perceived parental usage and the student’s liking for the food seemed to be the key concepts in reported frequency of choice. The linkage of parental usage to consumption frequency presumably reflects the impact of parental example, parental advice and/or parental control over what is served at mealtimes. Liking, on the other hand, reflects the direct personal response of the individual to the food. Thus, one can suggest that, for those foods where parental usage is a dominant predictor, the individual is accepting parental control, while for those foods where liking is the dominant predictor, the individual is asserting her or his own identity. In the former cases, attempts to motivate a change in consumption frequency would seem most appropriately targeted towards the teenager; in the latter situation, such attempts would seem likely to fail without parental support. By comparison, the impact of concepts of healthfulness and friends’ usage seems modest, and restricted to a limited number of foods. Attempts to motivate change
ADOLESCENT FOOD CHOICES
117
in consumption frequency could in principle be based on emphasizing health aspects or by developing resistance to peer pressure. Our data suggest that such approaches are, at most, appropriate to only a minority of foods. R Australian Bureau of Statistics. (1991). Schools Australia 1990. Canberra: Australian Bureau of Statistics. Bem, D. (1967). Self-perception: an alternative interpretation of cognitive dissonance phenomena. Psychological Review, 74, 183–200. Bem, D. (1972). Self-perception theory. Advances in Psychological Theory, 6, 1–62. Birch, L. L. (1979). Pre-school children’s preferences and consumption patterns. Journal of Nutrition Education, 11, 189–92. Contento, I. R., Michela, J. L. & Goldberg, C. J. (1988). Food choice among adolescents: population segmentation by motivations. Journal of Nutrition Education, 20, 289–98. Cosper, B. A. & Wakefield, L. M. (1975). Food choices of women. Journal of the American Dietetic Association, 66, 152–5. Farthing, M. C. (1991). Current eating patterns of adolescents in the United States. Nutrition Today, 26(2), 35–9. Hertzler, A. A. (1983). Children’s food patterns—a review. II. Family and group behavior. Journal of the American Dietetic Association, 83, 555–60. Laskarzewski, P., Morrison, J. A., Khoury, P., Kelly, K., Glatfelter, L., Larsen, R. & Glueck, C. J. (1980). Parent-child nutrient intake interrelationships in school children ages 6–19: The Princeton School District Study. American Journal of Clinical Nutrition, 33, 2350–5. Morgan, M. & Grube, J. W. (1991). Closeness and peer group influence. British Journal of Social Psychology, 30, 159–69. O’Connell, D. L., Alexander, H. M., Dobson, A. J., Lloyd, D. M., Hardes, G. R., Springthorpe, H. J. & Leeder, S. R. (1981). Cigarette smoking and drug use in schoolchildren. II. Factors associated with smoking. International Journal of Epidemiology, 10, 223–31. Pliner, P. (1983). Family resemblance in food preferences. Journal of Nutrition Education, 15, 137–40. Pliner, P. & Pelchat, M. L. (1986). Similarities in food preferences between children and their siblings and parents. Appetite, 7, 333–42. Randall, E. & Sanjur, D. (1981). Food preferences—their conceptualization and relationship to consumption. Ecology of Food and Nutrition, 11, 151–61. Rozin, P., Fallon, A. E. & Mandell, R. (1984). Family resemblance in attitudes to food. Developmental Psychology, 20, 309–14. Schafer, R. B. & Keith, P. M. (1981). Influences on food decisions across the family life cycle. Journal of the American Dietetic Association, 78, 152–5. Serville, Y. & Bleyer, R. (1976). L’apport des enqueˆtes de motivation pour la connaissance du comportement alimentaire. Annales de la Nutrition et de l’Alimentation, 309, 195–210. Tuorila, H. & Pangborn, R. M. (1988). Prediction of reported consumption of selected fatcontaining foods. Appetite, 11, 81–95. Williams, H. M., Woodward, D. R., Ball, P. J., Cumming, F. J., Hornsby, H. & Boon, J. A. (1993). Food perceptions and food consumption among Tasmanian high school students. Australian Journal of Nutrition and Dietetics, 50, 156–63. Worsley, A., Baghurst, P., Worsley, A. J., Coonan, W. & Peters, M. (1984). Australian tenyear-olds’ perceptions of food. 1. Sex differences. Ecology of Food and Nutrition, 15, 231–46.
Received 12 August 1994, revision 21 February 1996