Ethnic differences in perceptions of body size in middle-aged European, Maori and Paci®c People living in New Zealand

PA Metcalf 1,2*, RKR Scragg2, P Willoughby3, S Finau4 and D Tipene-Leach4

1Department of Statistics, University of Auckland, Private Bag 92019, Auckland, New Zealand; 2Department of Community Health, University of Auckland, Private Bag 92019, Auckland, New Zealand; 3Telstra New Zealand, PO Box 963, Auckland, New Zealand; and 4Department of Maori and Paci®c Health, University of Auckland, Private Bag 92019, Auckland, New Zealand

OBJECTIVES: The aim of this study was to compare perceptions of body size in European, Maori and Paci®c Islands people with measured body mass index (BMI), waist-to-hip ratio and change in BMI since age 21 y. Socio-demographic factors that in¯uenced perceptions of body size were also investigated. DESIGN: Cross-sectional survey. METHODS: Participants were 5554 workers, aged � 40 y, recruited from companies in New Zealand during 1988 ± 1990. RESULTS: Prevalences of BMI> 25 kg=m2 were: Europeans, 64.7% men, 47.2% women; Maori, 93.2% men, 80.6% women; and Paci®c Islanders, 94.1% men, 92.9% women. Similarly, prevalences of BMI > 30 kg=m2 were: Europeans, 14.4% men, 14.6% women; Maori, 55.0% men, 41.9% women; and Paci®c Islanders, 55.1% men, 71.7% women. At each perception of body size category, Maori and Paci®c Islands men and women had a higher BMI than European men and women, respectively. BMI increased with increasing perception of body size in all gender and ethnic groups. Since age 21, increases in BMI were highest in Paci®c Islands people and increased with increasing perceptions of body size category in all ethnic and gender groups. BMI adjusted odds (95% CI) of being in a lower perception category for body size were 1.70 (1.38 ± 2.12) in Maori and 8.99 (7.30 ± 11.09) in Paci®c people compared to Europeans, 1.27 (1.13 ± 1.42) times higher for people with no tertiary education, 1.41 (1.25 ± 1.59) times higher in people with low socioeconomic status, and 0.94 (0.92 ± 0.95) for change in BMI since age 21. CONCLUSION: Nutritional programs aimed at reducing levels of obesity should be ethnic-speci®c, addressing food and health in the context of their culture, and also take into account the socioeconomic status of the group. On the population level, obesity reduction programs may be more bene®cial if they are aimed at the maintenance of weight at age 21. InternationalJournal of Obesity (2000) 24, 593±599

Keywords: body mass index; perception of body size; socioeconomic status; education; Polynesia

Introduction

Obesity is associated with higher all-cause mortality1

and an increased risk of diseases,2 such as coronary heart disease,3 ± 5 hypertension,6,7 non-insulin depen- dent diabetes,8 ± 10 and some cancers.2 Reduction of the disease burden of obesity in the population will be assisted by the identi®cation of factors that in¯uence body weight which can be used for the development of intervention programs.

Western culture has a prevailing perception that a low body weight is attractive.11 Cultural and social factors such as the diet and fashion industries play an important role in shaping the idealized ®gure.12 Other cultures have different value systems, for example Paci®c Islands people, who ®nd a fuller ®gured body

more attractive13,14 as it represents high status, power, authority and wealth.15 Compared with Whites living in the USA, the prevalence of obesity is greater amongst African-Americans who pay less attention to weight reduction and dieting because they are less concerned about being overweight.16 Both cultural and social factors in¯uence dietary habits. BMI has been shown to be lower with increasing levels of education and socioeconomic status.17 ± 20

The aim of this study was to compare perceptions of body size in European, Maori and Paci®c Islands people with measured body mass index (BMI), waist- to-hip ratio and change in BMI since age 21. The roles of socioeconomic status, and level of education on perception of body size were also examined.

Methods

Between May 1988 and April 1990, 5678 workers aged 40 y and over were interviewed at 41 work sites in Auckland and ®ve work sites in Tokoroa (response

*Correspondence: PA Metcalf, Department of Statistics, University of Auckland, Private Bag 92019, Auckland, New Zealand. E-mail: metcalf@stat.auckland.ac.nz. Received 30 June 1999; revised 21 October 1999; accepted 3 December 1999

International Journal of Obesity (2000) 24, 593±599 ß 2000 Macmillan Publishers Ltd All rights reserved 0307±0565/00 $15.00 www.nature.com/ijo

rate 67%). The main criterion used to select compa- nies was size (more than 50 staff of all ages). Within Auckland, emphasis was placed on work sites in south and central Auckland to ensure suf®cient numbers of Maori and Paci®c Island participants. Workers repre- sented the following industries classi®ed according to the New Zealand Standard Industrial Classi®cation:21

agricultural services, hunting, forestry and ®shing, 3; manufacturing, 26; electricity, gas and water, 1; trans- port, storage and communications, 4; banking and ®nancial services, 1; and community, social and personal services, 11.

All participants completed a self-administered questionnaire regarding socio-demographic status. Ethnicity was self-de®ned and 100 Asians excluded because of their small numbers. Paci®c Islands people comprised: 53.9% Samoans; 26.6% Cook Islanders; 10.5% Tongans; 6.7% Niueans; and 2.3% other Paci- ®c Islanders. Participants were asked to `describe your present weight’ and to circle one of the following: 1, `Very underweight’ 2, `Moderately underweight’ 3, `Slightly underweight’ 4, `Just the right weight’ 5, `Slightly overweight’ 6, `Moderately overweight’ 7, `Very overweight’. Responses in the ®rst three cate- gories were combined as `Underweight’ because of small numbers, and responses renumbered on a scale of 1 ± 5. Of the 5578 remaining participants, 5554 (99.6%) answered this question. Participants were also asked `Do you keep your weight at the right level?’.

Weight was measured to the nearest 0.2 kg, height to the nearest 0.5 cm, and waist and hips to the nearest 0.5 cm by trained interviewers. BMI was calculated as weight in kg divided by height in m2. Participants were asked to record their weight at age 21. BMI at age 21 was calculated and the difference (BMI at interview 7BMI at age 21) was also calculated.

Education was classi®ed as no tertiary education versus tertiary education (Trade School, Technical Institute, Professional Institution, or University). Socioeconomic status was classi®ed according to the Elley ± Irving occupation based scale22 using the higher of the participant or their partner (1ˆ highest to 6ˆ lowest). This scale was categorised as low socioeconomic status (levels 4 ± 6) vs high socioeco- nomic status (levels 1 ± 3).

Statistical analyses were carried out using the SAS statistical package.23 Figures were generated using the Splus statistical package.24 Categorical variables were compared using the chi-squared test. For continuous variables, analysis of covariance was used to ex- amine the relationship with perception of body size categories after adjusting for potential confounding variables. Geometric mean body mass index was calculated as the loge of body mass index and a tolerance factor as exp(1.96�s.e.). A 95% con®dence interval can be calculated as mean=tolerance factor to mean�tolerance factor. A proportional odds model23 was built to examine socio-demographic factors that in¯uenced perception of body size and the 95% con®dence limits for the odds ratios were based on the pro®le likelihood. A proportional odds model ®ts a common slope cumulative model, which is a parallel lines regression model based on the cumulative prob- abilities. The cumulative logit model has the form

Pr ‰ y � i j xŠ ˆ exp…ai ‡ bT x†=…1‡ exp…ai ‡ bT x††; 1 � i � k

where a1, . . . , ak are k intercept parameters, and b is the vector of slope parameters. Study numbers may vary due to missing information.

Results

The socio-demographic characteristics of the study population are shown in Table 1. Overall, there were more males than females in each ethnic group. How- ever, there was a lower proportion of males in the Paci®c group compared to the European and Maori groups. Household income, no tertiary education and Elley ± Irving socioeconomic status (EISES) were lowest in Maori and Paci®c Islands people. Mean age was highest in Europeans. Mean age and gender adjusted body mass index (BMI), change in BMI since age 21, and waist-to-hip ratio were signi®cantly higher in Maori and Paci®c Islands people compared with Europeans.

Table 1 Socioeconomic-demographic characteristics of the study population by ethnic group. Mean (s.e.) or percentage by ethnic group

European Maori Pacific P-value

n 4464 433 657 Male gender 74.1% 71.4% 61.3% <0.001 Household income, <$30,000 22.7% 46.2% 67.9% <0.001 No tertiary education 41.2% 77.7% 81.9% <0.001 Low EISES (>3) 31.1% 63.3% 79.9% <0.001 Agea (y) 49.1 (0.09) 48.3 (0.30)* 47.4 (0.24)** BMIb,c (kg=m2) 26.0 (1.004) 30.3 (1.013)** 31.9 (1.011)** D BMIc since age 21 3.5 (0.06) 5.5 (0.19)** 8.1 (0.15)** Waist-to-hip ratioc 0.88 (0.001) 0.90 (0.004)** 0.91 (0.003)**

*0.01<P< ˆ0.001; **P<0.001, Maori and Paci®c people compared to Europeans. EISESˆElley ± Irving socio-economic scale (1ˆhigh, 6ˆ low); BMIˆ Body mass index; Dˆ change in. aAdjusted for gender; bGeometric mean (tolerance factor); c Adjusted for age and gender.

Ethnic differences in perceptions of body size PA Metcalf et al

594

International Journal of Obesity

Table 2 shows the percentage of participants in each category of perception of body size by ethnic group. There were ethnic differences in perception of body size (P< 0.001), with lower proportions of Maori describing themselves as `Underweight’ or `Just the right weight’. The highest proportions of Europeans and Maori described themselves as `Slightly overweight’, while the highest proportions of Paci®c Islands people described themselves as `Just the right weight’ or `Slightly overweight’.

Prevalences of overweight (BMI 25 ± 30 kg=m2) were Europeans: 50.3% men, 32.6% women; Maori: 38.2% men, 38.7% women; and Paci®c Islanders: 39.0% men, 21.2% women. Similarly, prevalences of obesity (BMI> 30 kg=m2) were Europeans: 14.4% men, 14.6% women; Maori: 55.0% men, 41.9% women; and Paci®c Islanders: 55.1% men, 71.7% women.

The percentages of participants who were over- weight and obese in each perception of body weight category by ethnic group are shown in Table 3. Higher percentages of Europeans were overweight in the `Slightly overweight’ and `Moderately overweight’ categories of perception of body size, but a lower percentage were obese in these perception categories compared with Maori and Paci®c Islands participants. In contrast, higher proportions of Maori and Paci®c Islands participants were overweight or obese in the `Underweight’ and `Right Weight’ perception cate- gories compared to Europeans.

At each perception of body size category, Maori and Paci®c Islands men (Figure 1A) and women (Figure 1B) had a higher BMI than European men and women, respectively. In general, BMI levels increased with increasing perception of body size in all gender and ethnic groups. There were also ethnic differences in waist-to-hip ratios in the same direction as for BMI in men and women, however, the associa-

tion with perception of body size was less marked (data not shown).

Increases in BMI since age 21 were highest in Paci®c Islands men (Figure 2A) and women (Figure 2B) compared to Maori and Europeans and also showed an association of increasing change in BMI since age 21 with increasing perceptions of body size categories.

More Europeans in the `Right weight’ category said they kept their weight at the right level than Maori and Paci®c Islands people (Figure 3). More Paci®c Islands and European people in the `Slightly overweight’

Table 2 Percentage of participants in each perception of body weight category by ethnic group

European Maori Pacific

n 4464 433 657 `Underweight’ 7.5 3.7 5.3 `Right weight’ 23.1 18.7 33.9 `Slightly overweight’ 43.8 35.6 30.6 `Moderately overweight’ 20.0 20.8 13.6 `Very overweight’ 5.8 21.3 16.6

Chi-square P-value< 0.001.

Table 3 Percentage of participants overweight (BMI 25 ± 30 kg=m2) and obese (BMI >30 kg=m2) in each perception of body weight category by ethnic group

European Maori Pacific

n 4464 433 657 Perception Overweight Obese Overweight Obese Overweight Obese `Underweight’ 11.4% 5.4% 31.3% 12.5% 45.7% 37.1% `Right weight’ 24.8% 0.8% 64.2% 11.1% 51.1% 37.7% `Slightly overweight’ 58.7% 7.8% 48.7% 43.5% 29.4% 66.2% `Moderately overweight’ 61.1% 30.8% 24.4% 72.2% 19.1% 80.9% `Very overweight’ 24.9% 73.9% 13.0% 85.9% 4.6% 93.6%

Figure 1 Geometric mean BMI by perception of body size category in 4110 men (A) and 1568 women (B). Bars representing 95% con®dence limits are one-sided when they overlap.

Ethnic differences in perceptions of body size PA Metcalf et al

595

International Journal of Obesity

category said they kept their weight at the right level, but fewer Europeans in the `Moderately overweight’ and `Very overweight’ categories said they kept their weight at the right level.

For both educational and socioeconomic groups, there was a trend (P< 0.001) towards the lower strata having more participants who perceived that their body sizes were `Underweight’ or the `Right weight’, less reported that they were `Slightly’ or `Moderately’ overweight, but more perceived being `Very overweight’ (Table 4). These results suggest that different socioeconomic strata have different views as to their perceptions of body size.

Mean changes in BMI since age 21 were greatest in the `Very overweight’ low and high education groups and low and high socioeconomic status (Table 5). There was an increasing gain in weight from the `Underweight’ to `Very overweight’ perception cate- gories in all education and socioeconomic status groups. Participants with low education and low socioeconomic status in the `Underweight’ group had put on signi®cantly more weight since age 21 compared to their high education and socioeconomic status groups. Similarly, participants with low educa- tion and low socioeconomic status had put on more weight since age 21 in the `Slightly overweight’ and `Moderately overweight’ groups.

A proportional odds model was built to identify factors in¯uencing perception of body size after adjusting for gender and BMI (Table 6). Age was not a signi®cant predictor of perception of body size and was therefore excluded from the model. BMI was more strongly related to perception of body size than waist-to-hip ratio, which was excluded from the model due to multicollinearity. BMI was centred and entered as a quadratic due to the non-linear relationship with perception of body size. Holding all other variables in the model constant, the odds of being in a lower perception of body size was almost 1.70 times higher in Maori compared to Europeans, 8.99 times higher for Paci®c people compared to Europeans, 1.27 times higher for people with no tertiary education compared with those who had tertiary education, 1.41 times higher for people in the lower socio-economic strata, and only 0.94 fold for each kg=m2 increase in BMI since age 21. The latter result suggests that those people gaining the most weight since age 21 appear to overestimate their body size.

Figure 2 Mean change in BMI since age 21 y by perception of body size category in 4000 men (A) and 1519 women (B). Bars representing 95% con®dence limits are one-sided when they overlap.

Figure 3 Proportions of participants in each perception cate- gory, by ethnic group, who `keep their weight at the right level’. Signi®cant differences are noted.

Table 4 Percentage of participants in each perception of body weight category by education and socioeconomic status group

Education Group Socioeconomic status

Low High Low High

n 2767 2874 2237 3417 `Underweight’ 7.8 6.1 6.8 7.0 `Right weight’ 24.1 23.9 26.2 22.6 `Slightly overweight’ 38.8 44.2 38.3 43.7 `Moderately overweight’ 19.1 19.4 18.5 19.7 `Very overweight’ 10.2 6.4 10.2 7.0

P<0.001 P<0.001

Column totals sum to 100%.

Ethnic differences in perceptions of body size PA Metcalf et al

596

International Journal of Obesity

Discussion

These cross-sectional data demonstrate that percep- tion of body size varies between ethnic groups, level of education and socioeconomic status. In particular, Maori and Paci®c people have a higher BMI than Europeans at all categories of perception of body size. Further, weight gains since age 21 were highest in Maori and Paci®c people. More Paci®c Islands and Maori people in the `Moderately overweight’ and `Very overweight’ perception categories thought that they kept their weight at the right level. This study is unique as it has much larger numbers, examines an older age group and contains a large proportion of males compared to previous studies.25,26

Age differences in body perception have been described previously27 and suggest that a gradual reduction in ideal body size is being experienced in Western societies. Here, there was no signi®cant effect of age on perception of body size, which may have been due to the relatively narrow age range of participants who were middle-aged, or that age differ- ences may be less notable these days.

A study of Maori health over the period 1955 ± 197528 found age-standardized rates of obesity (de®ned as 120% of ideal body weight) were 37% for Maori males and 57% for Maori females. This compared with 7% and 24% for non-Maori males and females, respectively. Although these are different measurements, and our sample may not be as repre- sentative, the ®ndings of the current study were higher

than these results. Further, there was a trend towards a higher change in BMI since age 21 in all ethnic groups, suggesting that weight increases had occurred prior to or during middle age.

Obesity in Paci®c people living in New Zealand has been acknowledged as playing a pivotal role in the development of coronary heart disease, non-insulin dependent diabetes mellitus and hypertension29 in `modernizing’ Paci®c people. Paci®c people are also at higher risk of developing obesity-related diseases than are Europeans.29 However, consistent with these results were our ®ndings that many Paci®c people in New Zealand are overweight, and that Paci®c people in New Zealand tend to develop obesity in middle to later years more readily than other New Zealanders. Feasts play an important social role in the lives of Paci®c peoples as food, social obligations and tradi- tion are closely linked.30 Therefore, it has been recommended that nutritional programs address food and health in the context of their society.30

Limitations

The appropriateness of Caucasian-based weight ranges for population groups in the Paci®c has been questioned.31 One suggestion has been to raise these ranges to include the Caucasian overweight cate- gory.31 At any given level of BMI, Cook Islanders are leaner than Australians32 and have less body fat for a given BMI.26 For example, a 25 y old Cook Islands woman with a BMI of 22 had 13 kg of fat, whereas a young Australian woman with the same BMI had 18 kg of fat. Young Polynesian women also have a lower resting metabolic rate33 and reduced energy expenditure34 compared to European women. Reduced energy expenditure and excessive energy intake have been hypothesized as causing obesity.34

A further study of these same young women found that a BMI of 30 kg=m2 for the New Zealand Eur- opean group corresponded to the same percentage body fat of 34 kg=m2 in Polynesians.35

A further limitation is the accuracy with which weight at age 21 is reported. It is likely that over- estimates of weight at age 21 were reported, thus dampening the change in BMI since age 21 ®ndings.

Table 5 Mean (s.e.) change in BMI since age 21 in kg=m2 adjusted for age, gender, and ethnic group by education and socioeconomic status group

Education group Socioeconomic status

Low High Low High

n 2767 2874 2237 3417 `Underweight’ 2.42 (0.23) 1.38 (0.26)* 2.70 (0.28) 1.48 (0.22)** `Right weight’ 2.39 (0.14) 2.06 (0.13) 2.29 (0.15) 2.17 (0.12) `Slightly overweight’ 4.34 (0.11) 3.77 (0.10)** 4.51 (0.12) 3.75 (0.09)** `Moderately overweight’ 6.11 (0.15) 5.48 (0.15)* 6.36 (0.17) 5.44 (0.13)** `Very overweight’ 8.85 (0.21) 8.60 (0.25) 8.78 (0.24) 8.73 (0.22)

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

Table 6 Sociodemographic variables in¯uencing perception of body image using a proportional odds model after adjusting for BMI

Variable Estimate s.e. Odds ratio 95%CI P-value

D BMI since age 21 7 0.07 0.01 0.94 0.92 ± 0.95 < 0.0001 Male gender 1.55 0.07 4.70 4.14 ± 5.35 < 0.0001 Maori 0.54 0.11 1.70 1.38 ± 2.12 < 0.0001 Paci®c 2.20 0.10 8.99 7.30 ± 11.09 < 0.0001 Low education 0.23 0.06 1.27 1.13 ± 1.42 < 0.0001 Low EISES 0.34 0.06 1.41 1.25 ± 1.59 < 0.0001

Concordance 81.4%; discordance 15.5%; tied 3.1%. EISESˆElley ± Irving socioeconomic status index. CIˆ con®dence interval.

Ethnic differences in perceptions of body size PA Metcalf et al

597

International Journal of Obesity

Conclusion

This study has shown that Maori and Paci®c people perceive themselves as having a smaller body size compared to Europeans, even after controlling for BMI. It also shows that Maori and Paci®c Island people in the `Moderately’ and `Very overweight’ perception categories think that they keep their weight at the right level. This suggests that cultural factors in¯uence perceptions of the ideal body size. This was con®rmed by the ®ndings that at each perception of body size category Maori and Paci®c men had higher mean BMI compared with European men. Paci®c females were also found to have a higher mean BMI at each perception of body size category compared to Maori and European. A multivariate model adjusted for BMI showed that Polynesian ethnicity, low level of education and low socio-eco- nomic status were all associated with a higher odds of being in a lower perception of body size category, but that larger changes in weight since age 21 were associated with a higher perception of body size. These results suggest that the development of Nutri- tional Programmes in Maori and Paci®c people need, in addition to scienti®c nutritional principals, to address food and health in the context of their society, and to address education level, and socioeconomic status of the targeted groups. On the population level, obesity reduction programs may be more bene®cial if they are aimed at the maintenance of weight at age 21.

Acknowledgements

Major funding was provided by the New Zealand Medical Research Council and the Medical Research Council Diabetes Task Force. Supplementary funds were received from the National Kidney Foundation of New Zealand and Lotteries Medical Research. Dr P Metcalf and this study were funded by the Health Research Council of New Zealand.

References 1 Manson JL, Stampfer MJ, Hennekens CH, Willett WC. Body

weight and longevity: a reassessment. JAMA 1987; 257: 353 ± 358.

2 NHIS ®ndings. Nutrition knowledge and baseline data for the weight-loss objectives. Public Health Reports January ± Feb- ruary, 1982.

3 Rimm EB, Stampfer M, Giovannucci E, Ascherio A, Spiegel- man D, Colditz GA, Willet WC. Body size and fat distribution as predictors of coronary heart disease among middle-aged and older US men. Am J Epidemiol 1995; 141: 1117 ± 1127.

4 Larsson B. Obesity, fat distribution and cardiovascular dis- ease. Int J Obesity 1991; 15: 53 ± 57.

5 Despres JP, Moorjani S, Lupien PJ, Tremblay A, Nadeau A, Bouchard C. Regional distribution of body fat, plasma lipo- proteins, and cardiovascular disease. Arteriosclerosis 1990; 10: 497 ± 511.

6 Stamler R, Stamler J, Riedlinger WF, Algera G, Roberts RH. Weight and blood pressure: ®ndings in hypertension screening of one million Americans. JAMA 1978; 240: 1607 ± 1610.

7 Kannel WB, Brand N, Skinner JJJ, Dawber TR, McNamara PM. The relations of adiposity to blood pressure and development of hypertension. Ann Intern Med 1967; 67: 48 ± 59.

8 Chan JM, Rimm EB, Colditz GA, Stampfer MJ, Willett WC. Obesity, fat distribution, and weight gain as risk factors for clinical diabetes. Diabetes Care 1994; 17: 961 ± 969.

9 Zimmet P, Dowse G, Finch C, Sergeantson S, King H. The epidemiology and natural history of NIDDM Ð lessons from the South Paci®c. Diabetes Metab Rev 1990; 6: 91 ± 124.

10 Hartz AJ, Rupley DC Jr, Kalkhoff RD, Rimm AA. Relation- ship of obesity to diabetes: in¯uence of obesity and body fat distribution. Prev Med 1983; 12: 351 ± 357.

11 Nichter M, Nichter M. Hype and weight. Medical Anthropol 1991; 13: 249 ± 84.

12 Furnham A, Alibhai N. Cross-cultural differences in the perception of female body shapes. Psychol Med 1983; 13: 829 ± 837.

13 Barrow T. Women of Polynesia.Seven Seas: Wellington, 1967. 14 van Dijk N. The Hansel and Gretel syndrome: a critique of

Houghton’s cold adaptation hypothesis and an alternative model. NZ J Archaeol 1991; 13: 65 ± 89.

15 Pollock N. Cultural elaborations of obesity Ð fattening prac- tices in Paci®c Societies. Asian Paci®c J Clin Nutr 1995; 14: 357 ± 360.

16 Wing R, Adams-Campbell L, Marcus M, Janney C. Effect of ethnicity and geographic location on body weight, dietary restraint and abnormal eating attitudes. Obes Res 1993; 1: 35 ± 41.

17 Woo J, Leung S, Ho S, Sham A, Lam T, Janus E. In¯uence of educational level and marital status on dietary intake, obesity and other cardiovascular risk factors in a Hong Kong Chinese population. Eur J Clin Nutr 1999; 53: 461 ± 467.

18 Cairney J, Wade T. Correlates of body weight in the 1994 National Population Health Survey. In J Obes 1998; 22: 584 ± 591.

19 Sundquist J, Johansson S. The in¯uence of socioeconomic status, ethnicity and lifestyle on body mass index in a long- itudinal study. Int J Epidemiol 1998; 27: 57 ± 63.

20 Rahkonen O, Lundberg O, Lahelma E, Huuhka M. Body mass and social class: a comparison of Finland and Sweden in the 1990s. J Public Health Policy 1998; 19: 88 ± 105.

21 Department of Statistics. New Zealand standard industrial classi®cation of occupations. Department of Statistics; Well- ington, 1987.

22 Elley W, Irving J. The Elley ± Irving Socio-economic Index 1981 Census Revision. NZ Educ Stud 1985; 20: 115 ± 128.

23 SAS Institute Inc. SAS=STAT user’s guide, version 6, 4th ed. SAS Institute Inc.: Cary, NC, 1989.

24 Statistical Sciences. S-PLUS user’s manual, version 3.2. StatSci, a division of Mathsoft Inc: Seattle, WA, 1993.

25 Brewis AB, McGarvey ST, Jones J, Swinburn BA. Perceptions of body size in Paci®c Islanders. Int J Obes 1998; 22: 185 ± 189.

26 Craig PL, Swinburn BA, Matenga-Smith T, Matangi H, Vaughan G. Do Polynesians still believe that big is beautiful? Comparison of body size perceptions and preferences of Cook Islands Maori and Australians. NZ Med J 1996; 109: 200 ± 203.

27 Garner DM, Gar®nkel PE, Schwartz D, Thompson M. Cultural expectations of thinness in women. Psychol Rep 1980; 47: 483 ± 491.

28 Pomare E. Maori standards of health Ð a study of the 20 year period 1955 ± 75. Medical Research Council of New Zealand, December 1980.

29 Bathgate M, Alexander D, Mitikulena A, Borman B, Roberts A, Grigg M. The Health of Paci®c Islands People in New Zealand. Public Health Commission; Wellington, 1994; pp 107 ± 111.

30 Gifford I, Calder L. The Central Auckland Health status review, 1989.

Ethnic differences in perceptions of body size PA Metcalf et al

598

International Journal of Obesity

31 South Paci®c Commission. The 1986 National Nutrition Survey of the Kingdom of Tonga: South Paci®c Commission: Noumea, New Caledonia, 1992.

32 Swinburn BA, Craig PL, Prijatmoko D, Strauss B. Body mass index in Polynesians: are current de®nitions for obesity appro- priate? Int J Obes 1996; 20: 889 ± 894.

33 Rush E, Plank L, Robinson S. Resting metabolic rate in young Polynesian and Caucasian women. Int J Obes 1997; 21: 1071 ± 1075.

34 Rush E, Plank L, Coward W. Energy expenditure of young Polynesian and European women in New Zealand and rela- tions to body composition. Am J Clin Nutr 1999; 69: 43 ± 48.

35 Rush E, Plank L, Laulu M, Robinson S. Prediction of percen- tage body fat from anthropometric measurements: comparison of New Zealand European and Polynesian young women. Am J Clin Nutr 1997; 66: 2 ± 7.

Ethnic differences in perceptions of body size PA Metcalf et al

599

International Journal of Obesity