Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Structural, material and economic influences on the food‐purchasing choices of socioeconomic groups

Structural, material and economic influences on the food‐purchasing choices of socioeconomic groups Structural, material and economic influences on the foodpurchas'ing choices of socioeconomic groups Gavin Turrell School o Public Health, Queensland University of Technology, Brisbane f Abstract: Australian and overseas studies have found that respondents in low socioeconomic groups are least likely to purchase food that accords with recommendations in dietary guidelines. British and United States researchers have proposed that this consistently observed association is partly due to structural, material and economic factors that differentially affect socioeconomic groups. This study tested that proposition. Specifically, this study examined thr notion that socioeconomic variability in food-purchasing choices are in part a function of the availability,accessibility and affordability of food recommended by dietary guidelines. Data collected from socioeconomic groups in the general community, and information provided by welfare recipients living in lowincome areas of Brisbane and Logan city provided little support for this notion. Although significant differences were found between socioeconomic groups in terms of their food-purchasing choices, most respondents from all socioeconomic groups shopped at large supermarkets where recommended food was readily available, few reported difficulties obtaining access to these facilities, and the price difference between recommended and regular foods was, in most cases, small or nonexistent. This evidence leaves largely unanswered the question of why socioeconomic groups differ in their foodpurchasing behallours. (Aust iV ZJPublac Health 1996; 20: 611-17) URING, the last few decades, governments and health authorities, both in Australia and overseas, have developed dietary guidelines. The main purpose of these guidelines is to .L provide advice to the general population about healthy food choices, so that their usual diet contributes to a healthy litestylt: and is consistent with minimal risk for the development o t diet-relnted disenqes.' related behaviour.' That is, via the guidelines, the general public are encouraged to make 'healthy' food choices when shopping, to engage in 'nutritionally sound' preparation and cooking methods, and to alter their Eonsumption practices in accordance with current 'official' beliefs about what constitutes a healthy diet.? Australian researchers have found that low socioeconomic groups are least likely to comply with these dietary recommendations.?-' Although it is now well established that groups of differing socioeconomic Status in Australia differ in their compliance with recommended dietary practices, we know very little why these diffearences exist. This paper attempts to further our knowledge and understanding of this issue by focusing on the structural, VOL. The Australian dietary guidelines, as presented and discussed in government publications and health promotion campaigns, consist of numerous qualitative recommendations for a range of foodCorrespondence to Dr Gavin Turrell, Postdoctoral Research Fellow, School oaf Public Health, Queensland University of Technology, Locked Bag N o . 2, Red Hill, Qld 4059. Fax (07) 3864 3369, e-mail g.turrell&qut.edu.au. AUSTRALIAN AND NEW ZEAIAND JOURNAL O PUBLIC HEALTH 1996 F 20 NO. 6 TURRELL material and economic determinants of socioeconomic variability in food-purchasing choices. Researchers in the United Kingdom and the United States have suggested that persons of low socioeconomic status experience disproportionate difficulties in their attempts to procure healthy food.Ci- I ”, These difficulties relate to three issues: availability, accessibility and affordability. First, lowstatus groups often live in areas where there are few large supermarkets. As a consequence, these groups are reliant on smaller shops, which typically stock a limited range of foods; their prices are higher and the food is often of a lesser quality. Second, low-status groups experience difficulties in obtaining access to large (and often distant) shopping facilities because they lack private transport or live in areas where public transport is inadequate or nonexistent. Again, this situation sometimes results in low-status fanlilies buying a disproportionately large amount of their food from smaller local shops. Third, foods recommended by the dietary guidelines are considered too expensive for economically disadvantaged groups such as the elderly, the unemployed and the 1ow-paid. The latest policy document from the Public Health Association of Australia reviewed some of this overseas material.” It proposed that disadvantaged groups in Australia may also experience many of the same food-related problems as their counterparts in other countries. This study tests this proposition by examining three questions. First, are recommended foods equally available to all socioeconomic groups? Second, do groups with low socioeconomic status experience difficulties with access to shopping facilities that stock foods recommended by dietary guidelines? Third, can groups with low socioeconomic status afford the type of food that is promoted in the dietary guidelines? These questions are considered in relation to a non-Aboriginal sample, living in a densely populated urban area of southeast Queensland. Measures As part of the questionnaire, respondents were asked to indicate their degree of involvement in their household’s food shopping. Those who indicated that they were never involved in this activity ( n = 66) were not required to answer any of the remaining questions about food-purchasing and were deleted from the analysis. In addition, five respondents from the electoral roll sample missed a number of the initial filter questions pertaining to food purchasing, and these respondents were also excluded. These deletions reduced the number in the two samples to 402, 139 males (34.6 per cent) and 263 females (65.4 per cent). This gender imbalance was largely accounted for by the disproportionate number of males who were excluded because they never undertook the food shopping for their households (60 males out of 66 respondents). The final sample, therefore, comprised men and women who reported that they purchased food for their households at least some of the time. Food purchasing Food purchasing was examined by 11 questions, each of which had two or more response categories. For example, respondents were asked: ‘When you go food shopping what type of milk do you usually buy?’. The response options included: I do not buy milk, full-cream milk, skim (low-Fat) milk, Trim (reduced-fat) milk, Shape (calcium-modified) milk, plus others. Multiple responses were permitted for Table 1 : Classification of foods into recommended and regular categories Food Minced meat Bread Recommended low-fat wholemeal multigrain white hyfibre rye (dark and light) white ‘n’ bran low-fat (fruit and natural) no added sugar (in natural juice) polyunsaturated table margarine brown Regular full-fat white Methods The research design and method used has been described in detail elsewhere.’” Two samples were used. The first ( n= 500) was randomly selected from the Commonwealth electoral roll, which comprised all eligible persons in Brisbane city registered at the end of January 1993. The second sample consisted of 70 self-selected clients from three Salvation Army welfare centres located in low-income areas in Brisbane and Logan city. Respondents in each sample completed an identical structured questionnaire. A mailsurvey method developed by Dillman was used to collect data from the electoral roll sample.I6 Adherence to the established approaches and protocols recommended by Dillman yielded a final usable response rate of 80.6 per cent, which translated into a sample of 403 respondents. Data were collected from the welfare sample by the researcher, who personally delivered the questionnaire to each of the 70 respondents, who completed the instrument in the researchers’ presence. Yoghurt Tinned fruit Spreads Rice full-fat (fruit and natural) in syrup butter dairy blend white Milk skim (low-fat) full-cream Trim (reduced-fat) Shape (calcium-modified) low- or reduced-fat wholemeal no added sugar (unsweetened) vegetable oils futl-fat white spinach sweetened cooking margarine vegetable shortening solidified oil beef dripping lard Cheese Pasta Fruit iuice Fats and oils AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1996 vot. 20 NO 6 F FOOD CHOICES OF SOCIOECONOMIC GROUPS each question. The other 10 food questions were structured in an identical manner. The foods in each food type were subsequently classified into a ‘recommended’ and ‘regular’ category (Table 1). A preliminary examination of the data revealed that respondents exhibited three types of purchasing pattern for each of the 11 foods. They either purchased the recommended or regular option exclusively, or they purchased a variety of food including bofh the recommended and the regular option. Respondents were assigned the value 3 if they reported purchasing the recommended food exclusively, the value 2 if they purchased both the recommended and regular options, and the value 1 if they purchased only the regular option. Therefore, for all 11 food types, respondents were assigned to either a recommended (3), a mixed (2) or a regular ( 1 ) category. Each of the food items was summed to form a food choice index, which was rescored to range from 0 to 20. The mean and median were 9.2, standard deviation 4.3 and skewness 0.25. Those respondents scoring ‘LO on the index consistently chose the recommended food option and those scoring zero consistently chose the regular option. was possible that perceived availability rather than actual availability was the more important determinant of food choice. If the consumer was unaware of the availability of a recommended alternative, even if it was present where they usually shopped, then it followed that the food item was unlikely to be purchased. Accessibility ofjoods recornmmdcd in dielarq’ Lpiciplines The question of whether food recommended in dietary guidelines was accessible to respondents in the two samples was measured with two precoded structured items and one open-ended item. The first precoded question asked ‘What type of transport do you usually use when you go food shopping for your household?’. The second asked ‘Do you experience any transport difficulties getting to o r from your usual food shopping centre?’. If respondents indicated ‘ves’ to this second question they were then invited to identify (in their own words) the difficulties they encountered. Socioeconomic stutus A defining ;and homogenising feature of the Salvation Army sample was that its members were partly or totally dependent on welfare for their income needs. This, more than any other factor, distinguished them from the electoral roll sample. It was therefore decided to treat the welfare sample as a discrete socioeconomic group. It was also considered important that a comparable indicator should be used to measure socioeconomic inequality within the electoral roll sample. Therefore, total annual household income was used as the measure of socioeconomic status for this group. As part of the questionnaire, respondents were required to indicate their annual household income on a single item consisting of 12 narrow-ranged categories. This measure was subsequently collapsed into three categories: $0 to $13500 (low), $13501 to $43000 (medium) and $45 001 o r more (high). Availability ofjbod recommended in dietmy cpiddines The question of whether food recommended in dietary guidelines was equally available to respondents in the electoral roll and welfare samples was examined by three different approaches. The first consisted of asking respondents ‘Where is the bulk of your household’s food purchased from?’. The second approach involved contacting a major food retailer to determine whether the foods recommended by the dietary guidelines and included in the study were stocked in shops throughout the Brisbane metropolitan region and in Logan city, especially those low-income areas whence the welfare sample was recruited. The third approach asked respondents to indicate whether they agreed, disagreed or were unsure about the availability of the recornmended foods that were used to construct the purchasing index. This item was included because it AfJbrdubilitj of food recommendpd i n the dietnry guidelines T~vo different approaches were used to examine the relationship between economic factors and food purchasing. First, one of Australia’s largest food retailers was given a list of foods recommended by the dietary guidelines and their regular equivalents and asked to provide unit prices for both sets of food items. Again, these foods matched those used in the construction of the purchasing index. The food prices pertained to the period during which the questionnaire data were collected (April 1993). The food items within a particular food group (for example, white versus wholemeal bread) were identical in brand and product size. This had the effect of standardising the food items, ensuring that the price comparisons were valid. Clearly, the prices were likely to vary across brands. According to the food retailer, however, the brands chosen for coinpara tive purposes were commonly purchased and therefore likely to be sufficiently representative. Second, on the basis of a variety of data provided by respondents, three different estimates of household food expenditure were calculated: estimated percentage of total weekly household income spent on food; estimated mean amount ($) spent on food per household per week; and estimated mean amount ($) spent on food per household member per week. Analpis The general linear models procedure in SAS was used to examine the association between socioeconomic status and scores on the food-purchasing index.]’ This procedure (which is appropriate for unbalanced designs) was used in the first instance to compute main-effect (unadjusted) means for each of the groups. Given that age and sex have been shown to be significantly related to both socioeconomic status and food choice it was necessary to make statistical adjustments for these potentially confounding factors.’ This was undertaken by NO. AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1996 vot. 20 F TURRELL Table Table 20: Comparison of socioeconomic groups: mean food 4: Perceived availability of the recommended foods included in the questionnaire ~~ ~ purchasing scores (unadjusted and adiusted estimates) Food purchasing scores Income level High Medium tow Welfare Mean Perceived to be availablea Unadiusted score Adjusted Food F P item and Yes Unsure scorea socioeconomic groupb Vegetable oils high medium low welfare Fruit juice (no added sugar) high medium low welfare Wholemeal pasta high medium low welfare tow-fat cheese high medium low welfare Trim or skim milk high medium low welfare Brown rice high medium low welfare Tinned fruit (no added sugar) high medium low welfare Wholemeal or multigrain bread high medium low welfare Low-fat or diet minced meat high medium low welfare tow-fat yoghurt high medium low ~~ Note: (a) Score adiusted for age and sex Table 2b: Comparison of socioeconomic significance values paired-mean comparisons groups: (PI for adjusted Income level High Medium tow Welfare High Income level Medium Low Welfare 1 .o 0.348 0.028 0.0001 1.o 0.094 0.0001 1 .o 0.0001 1 .o Table 3: Shopping facilities used by respondents when pur- chasing the bulk of the household's food Income level High Shopping facility Medium low Welfare n = 7 7 n=166 n = 8 1 n = 6 8 % % % % Large supermarket exclusively 65 Local suburban shop or 7-day convenience store 1 Specialty shops (e.g. butcher grocer, baker) Large supermarket plus specialty shops including age and sex as covariates within the linear model. Least-squares (adjusted) means were then computed for each of the groups. The general linear models procedure was also used to examine the relationship between socioeconomic status and estimated food expenditure. The relationship between socioeconomic status and a number of the other explanatory variables was examined with the chisquare test. Results The sex, age and labour force profiles of the two samples have been examined in detail elsewhere.'j This earlier analysis showed that the electoral roll sample was not too dissimilar from the general p o p ulation in its sociodemographic characteristics. The welfore Notes: (a) The few respondentswho perceived some food items to be unavailable are not included. (b) Marginal frequencies for each group: high 76-77, medium 165-1 67, low 75-81, welfare 68 xz for differences between income groups: P > 0.05 AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1996 vot. 20 NO. 6 F FOOD CHOICES O SOCIOECONOMIC GROUPS F welfare sample however, was quite different in its composition and displayed many of the characteristics commonly associated with social and economic disadvantage. The groups were compared in terms of their mean scores on the food-purchasing index (Table 2). The analysis of variance test showed that the unadjusted and adjusted mean scores between at least two of the groups were significantly different. The adjusted paired-mean comparisons showed that the welfare group scored significantly lower on the purchasing index than each of the other three groups. These findings indicate that welfare respondents (independent of age and sex) were more consistent purchasers of regular foods, whereas respondents in the high socioeconomic group were more likely to purchase recommended foods consisten tly. Contact with a major food retailer revealed that all of the recommended foods included in the questionnaire were readily available at large food outlets in Brisbane and Logan city. As most respondents usually purchased food from these types of estal, lishments, it would appear that availability was not likely to be a factor influencing food choice. As was argued earlier, however, perceived availability rather than actual availability may have been more important in decision making. When the four groups were compared on 10 items measuring perceived availability, it was evident that most respondents from each group reported that all of the recommended foods were available where they usually shopped (Table 4). A series of chi-square tests showed that the groups did not differ significantly in terms of their perceptions about the availability of food recommended by the dietary guidelines. Availability ofJbod recommended in dietary guidelines The four groups were compared in terms of the types of shopping facility used when purchasing food (Table 3). Over 96 per cent of respondents in each of the groups reported that they usually purchased the bulk of their household’s food at a large supermarket (either exclusively or in conjunction with a number of specialty shops). There was no evidence, from these data at least, that respondents in the low-income group or welfare sample were buying most of their food at small suburban shops or convenience s8tores. Table 5: Type of transport used by respondents when purchasing the households food Income level Medium Low Welfare n = 168 n = 8 0 n = 6 8 % % % 92 2 1 2 2 61 6 23 69 6 12 9 4 Accessibility o food recommended by the dietary guideline f The socioeconomic groups were compared in terms of the types of transport they reported using when shopping for the household’s food (Table 5 ) . There was a significant association between these variables (f = 74.0, 12 df, P < 0.0001). Specifically,while most respondents from all four groups used their own private vehicles when shopping, respondents in the low-income and welfare groups were least likely to report that they used this form of transport. Respondents in these two groups were more likely to purchase the household’s food using public transport, or a friend’s or relative’s private vehicle, or they simply walked. Respondents were subsequently asked if they experienced any transport difficulties when shopping, and over 94 per cent of each group reported that they encountered no problems travelling to or from their usual shopping facility. Affordability ofjood recommended in dietary guidelines The 1988-89 household expenditure survey found that money spent on food and nonalcoholic beverages comprised the single largest component of average weekly expenditure on commodities and services, accounting for approximately 15.0 per cent of mean total weekly income.” It was estimated that households in the electoral roll sample spent approximately 17.0 per cent of their total weekly income on food, which is very similar to the national estimate. This gross estimate, however, masks a considerable degree of socioeconomic variability in food expenditures (Table 6). Respondents in the high-income group spent approximately 9 per cent of their total weekly household income on food, whereas the low-income and welfare groups spent about 30 per cent ( F =51.0, 3 df, P = 0.0001). Households in each of the four groups spent significantly different absolute amounts on food each week (F=3.4, 3 df, P = 0.0177; Table 6). It was possible that these differences were in part accounted for by variations in the size of the households in each group. For example, households in the high, medium, low and welfare groups averaged 3.2, 2.6, 1.8 and 3.6 members respectively. The estimated VOL. Transport Own private vehicle Friend’s or relative’s vehicle Bus, train or taxi Walk Combination of above High n=77 % Table 6: Estimated food expenditure patterns of households according to income level Income level High Medium Low Welfare n=77 n=166 n = 8 1 n = 6 8 Measure of estimated food expenditure Percentage of total weekly household income Mean amount spent per household per week 9% $102 15% $88 $39 31% $68 30% $79 $27 Mean amount spent per household member per week $38 $41 AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1996 F 20 NO 6 TURRELL Table 7: Retail prices of foods included in the food-choice section of the questionnaire (April 1993) Price per Food iype Minced meat Bread Unit Food items low-fat regular white wholemeal multigrain white hyfibre white’n‘bran low-fat (natural) full-fat (natural) low-fat (fruit) full-fat (fruit) no added sugar (natural juice) in syrup butter table margarine (polyunsaturated) dairy blend white brown full-cream skim (law-fat) Trim (reduced-fat) Shape (calcium-modified) full-fat low-fat white wholemeal no added sugar unit ($) 1 kg 7.99 5.99 1.45 1.45 1 .A5 1.52 1.52 0.60 0.60 0.60 0.60 1.69 1.69 1.55 1.52 1.90 1.31 1.27 2.02 2.02 2.10 2.14 3.65 3.77 1.29 1.29 3.79 3.79 680 g ‘foghurt 200 g Tinned fruit Spreads 825 g 500 g Rice 1 kg 2L Milk Cheese Pasta Fruit iuice 500 g 500 g 2L reaular lsweetenedl V . Source: Woolworths (Queenslond) Pty Ltd amount spent on food in each household per week was adjusted for the number of members compris- ing the household unit. After these adjustments, the amount spent on food per household member per week in the welfare group was significantly less than that spent on members of households in the other three income groups (F=3.2, 3 df, P = 0.0251). One factor that potentially contributes to the food-purchasing difficulties faced by low-income families is food prices. Table 7 presents the retail prices of 10 recommended foods and compares these with their regular equivalents. In most rases, the price difference behveen foods within the same food group was nil-to-small. For example, there was no difference in price between white and wholemeal bread, low-fat and full-fat yoghurt, white and wholemeal pasta, unsweetened and regular tinned fruit and fruit juice, and hill-cream milk and skim (lowfat) milk. There was a small price difference between white and brown rice, calcium-modified milk arid other milk, and full-fat and low-fat cheese. There was a much more substantial price difference however, between lowfat mince and regular mince. Discussion Previous Australian and overseas studies have shown that the food behaviours of groups of low socioeconomic status are least likely to accord with dietary recommendations. The results of this study corroborated these findings. Respondents in the welfare sample were significantly less likely than their more affluent counterparts to choose food recommended in the dietary guidelines when shopping. Studies conducted in the United States and the United Kingdom have found that the food-related behaviours of low-income groups are often constrained and circumscribed by structural, material and economic factors. Specifically, large supermarkets are less likely to be located in disadvantaged areas, groups of low socioeconomic status are sometimes unable to reach shopping facilities that stock food recommended in the dietary guidelines, and the difficult economic circumstances of low-income groups limit their capacity to afford these foods. This study examined these issues in an Australian context. Over 96 per cent of respondents from all four socioeconomic-status groups reported that they regularly purchased their food from a large supermarket and an equally large proportion reported that each of the recommended foods included in the questionnaire was available where they usually shopped. Information provided by a large food retailer confirmed that most supermarkets in the areas studied stocked food recommended by the guidelines. Further, although respondents in the low-income group and welfare sample were less likely to have had access to a private vehicle for shopping purposes, very few indicated that this was a problem. Possibly, for many low-income persons, a lack of private transport is a ubiquitous and accepted feature of day-to-day living. Based on the findings of this study, it appears that recommended foods were readily available and accessible to both affluent and disadvantaged alike. The question of whether low-income groups can afford to purchase food recommended by the guidelines is more complicated. If we take a broad view and ask ‘Do low income persons face difficulties in their attempts to procure food for themselves or their families?’, then the answer is unequivocally ‘yes’. This study and research conducted elsewhere have shown that low-income groups have to allocate a much greater proportion of their household income for food, yet they spend significantly less on food per household member. Moreover, because food is usually the only discretionary item in a lowincome family’s budget, spending on this necessary commodity is likely to be reduced to meet nondiscretionav costs such as rent and bills. An unfortunate consequence of this reduced spending power is that lowincome families are often forced to purchase food less frequently and in smaller quantities, the range and diversity of foods that can be purchased is limited, and the foods bought are very often of a lesser If we take a more narrow view however, and ask, ‘Can low-income groups afford to purchase the type VOL. AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1996 F 20 NO 6 FOOD CHOICES O SOCIOECONOMIC GROUPS F of food which is promulgated in the dietary guidelines?’, then the answer is less certain. The results of studies conducted in the United Kingdom suggest that foods recommended in the dietary guidelines are priced out of the reach of low-income famiBy lies.~,ll.l~ contrast, the few Australian studies examining this issue have presented evidence suggesting that recommended foods and diets are affordable by low-income families.’”“’ In this study, it was evident that, with only a few exceptions, the prices of recommended foods and their regular equivalents were identical or very similar. Of course, the difficulty that we face is determining what constitutes a price difference sufficiently large that it would adversely affect the purchasing decision of the low-income consumer. From our relatively advantaged position, a price difference of 12 cents between full-fat and low-fat cheese seems small, and all other things being equal, is not likely to be a determining factor in the purchase decision. However, when a large number of food items are being purch.ased, the seemingly small difference between the recommended and regular foods may take on a new significance. This point was clearly highlighted by Mooney, who found that a hypothetical diet consisting of 15 recommended food items cost S13.99 per week compared with S8.33 for an equivalent diet comprising regular foods.’ A similar situation can be constructed from the data in Table 7. If we sum the retail prices of the most expensive recommended food items, we find that this hypothetical shopping basket costs $25.58. An identical procedure for the cheapest regular items results in a basket costing $23.34. Only one food item accounts for this disparity in price between the recommended and regular shopping baskets: in 1993, low-fat minced meat cost $7.99/kg compared with $5.99/kg for regular minced meat. Based on this evidence, it seems reasonable to suggest that if the low-income consumer is selective and discerning in his or her purchasing of food, then the cost of a diet that conforms with dietary guidelines need not cost substantially more than one that is at variance with these guidelines. Conclusion structural, material and economic factors did not explain socioeconomic variability in food choice raises the possibility that social-psychological and subcultural factors (beliefs, attitudes and knowledge) account for the differences. This possibility will be explored in detail in a subsequent paper. Acknowledgments The author wishes to thank the officer-s and staff at the Salvation Army headquarters (Brisbane) who helped to recruit low-income families, and Ms Pam Lucas from Woolworths (Qld) Pty Ltd who provided the retail prices for the recommended and regular foods. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Australian and New Zealand Journal of Public Health Wiley

Structural, material and economic influences on the food‐purchasing choices of socioeconomic groups

Download PDF
7 pages

Loading next page...
 
/lp/wiley/structural-material-and-economic-influences-on-the-food-purchasing-xdJ1P1feeh

References (20)

Publisher
Wiley
Copyright
Copyright © 1996 Wiley Subscription Services, Inc., A Wiley Company
ISSN
1326-0200
eISSN
1753-6405
DOI
10.1111/j.1467-842X.1996.tb01075.x
Publisher site
See Article on Publisher Site

Abstract

Structural, material and economic influences on the foodpurchas'ing choices of socioeconomic groups Gavin Turrell School o Public Health, Queensland University of Technology, Brisbane f Abstract: Australian and overseas studies have found that respondents in low socioeconomic groups are least likely to purchase food that accords with recommendations in dietary guidelines. British and United States researchers have proposed that this consistently observed association is partly due to structural, material and economic factors that differentially affect socioeconomic groups. This study tested that proposition. Specifically, this study examined thr notion that socioeconomic variability in food-purchasing choices are in part a function of the availability,accessibility and affordability of food recommended by dietary guidelines. Data collected from socioeconomic groups in the general community, and information provided by welfare recipients living in lowincome areas of Brisbane and Logan city provided little support for this notion. Although significant differences were found between socioeconomic groups in terms of their food-purchasing choices, most respondents from all socioeconomic groups shopped at large supermarkets where recommended food was readily available, few reported difficulties obtaining access to these facilities, and the price difference between recommended and regular foods was, in most cases, small or nonexistent. This evidence leaves largely unanswered the question of why socioeconomic groups differ in their foodpurchasing behallours. (Aust iV ZJPublac Health 1996; 20: 611-17) URING, the last few decades, governments and health authorities, both in Australia and overseas, have developed dietary guidelines. The main purpose of these guidelines is to .L provide advice to the general population about healthy food choices, so that their usual diet contributes to a healthy litestylt: and is consistent with minimal risk for the development o t diet-relnted disenqes.' related behaviour.' That is, via the guidelines, the general public are encouraged to make 'healthy' food choices when shopping, to engage in 'nutritionally sound' preparation and cooking methods, and to alter their Eonsumption practices in accordance with current 'official' beliefs about what constitutes a healthy diet.? Australian researchers have found that low socioeconomic groups are least likely to comply with these dietary recommendations.?-' Although it is now well established that groups of differing socioeconomic Status in Australia differ in their compliance with recommended dietary practices, we know very little why these diffearences exist. This paper attempts to further our knowledge and understanding of this issue by focusing on the structural, VOL. The Australian dietary guidelines, as presented and discussed in government publications and health promotion campaigns, consist of numerous qualitative recommendations for a range of foodCorrespondence to Dr Gavin Turrell, Postdoctoral Research Fellow, School oaf Public Health, Queensland University of Technology, Locked Bag N o . 2, Red Hill, Qld 4059. Fax (07) 3864 3369, e-mail g.turrell&qut.edu.au. AUSTRALIAN AND NEW ZEAIAND JOURNAL O PUBLIC HEALTH 1996 F 20 NO. 6 TURRELL material and economic determinants of socioeconomic variability in food-purchasing choices. Researchers in the United Kingdom and the United States have suggested that persons of low socioeconomic status experience disproportionate difficulties in their attempts to procure healthy food.Ci- I ”, These difficulties relate to three issues: availability, accessibility and affordability. First, lowstatus groups often live in areas where there are few large supermarkets. As a consequence, these groups are reliant on smaller shops, which typically stock a limited range of foods; their prices are higher and the food is often of a lesser quality. Second, low-status groups experience difficulties in obtaining access to large (and often distant) shopping facilities because they lack private transport or live in areas where public transport is inadequate or nonexistent. Again, this situation sometimes results in low-status fanlilies buying a disproportionately large amount of their food from smaller local shops. Third, foods recommended by the dietary guidelines are considered too expensive for economically disadvantaged groups such as the elderly, the unemployed and the 1ow-paid. The latest policy document from the Public Health Association of Australia reviewed some of this overseas material.” It proposed that disadvantaged groups in Australia may also experience many of the same food-related problems as their counterparts in other countries. This study tests this proposition by examining three questions. First, are recommended foods equally available to all socioeconomic groups? Second, do groups with low socioeconomic status experience difficulties with access to shopping facilities that stock foods recommended by dietary guidelines? Third, can groups with low socioeconomic status afford the type of food that is promoted in the dietary guidelines? These questions are considered in relation to a non-Aboriginal sample, living in a densely populated urban area of southeast Queensland. Measures As part of the questionnaire, respondents were asked to indicate their degree of involvement in their household’s food shopping. Those who indicated that they were never involved in this activity ( n = 66) were not required to answer any of the remaining questions about food-purchasing and were deleted from the analysis. In addition, five respondents from the electoral roll sample missed a number of the initial filter questions pertaining to food purchasing, and these respondents were also excluded. These deletions reduced the number in the two samples to 402, 139 males (34.6 per cent) and 263 females (65.4 per cent). This gender imbalance was largely accounted for by the disproportionate number of males who were excluded because they never undertook the food shopping for their households (60 males out of 66 respondents). The final sample, therefore, comprised men and women who reported that they purchased food for their households at least some of the time. Food purchasing Food purchasing was examined by 11 questions, each of which had two or more response categories. For example, respondents were asked: ‘When you go food shopping what type of milk do you usually buy?’. The response options included: I do not buy milk, full-cream milk, skim (low-Fat) milk, Trim (reduced-fat) milk, Shape (calcium-modified) milk, plus others. Multiple responses were permitted for Table 1 : Classification of foods into recommended and regular categories Food Minced meat Bread Recommended low-fat wholemeal multigrain white hyfibre rye (dark and light) white ‘n’ bran low-fat (fruit and natural) no added sugar (in natural juice) polyunsaturated table margarine brown Regular full-fat white Methods The research design and method used has been described in detail elsewhere.’” Two samples were used. The first ( n= 500) was randomly selected from the Commonwealth electoral roll, which comprised all eligible persons in Brisbane city registered at the end of January 1993. The second sample consisted of 70 self-selected clients from three Salvation Army welfare centres located in low-income areas in Brisbane and Logan city. Respondents in each sample completed an identical structured questionnaire. A mailsurvey method developed by Dillman was used to collect data from the electoral roll sample.I6 Adherence to the established approaches and protocols recommended by Dillman yielded a final usable response rate of 80.6 per cent, which translated into a sample of 403 respondents. Data were collected from the welfare sample by the researcher, who personally delivered the questionnaire to each of the 70 respondents, who completed the instrument in the researchers’ presence. Yoghurt Tinned fruit Spreads Rice full-fat (fruit and natural) in syrup butter dairy blend white Milk skim (low-fat) full-cream Trim (reduced-fat) Shape (calcium-modified) low- or reduced-fat wholemeal no added sugar (unsweetened) vegetable oils futl-fat white spinach sweetened cooking margarine vegetable shortening solidified oil beef dripping lard Cheese Pasta Fruit iuice Fats and oils AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1996 vot. 20 NO 6 F FOOD CHOICES OF SOCIOECONOMIC GROUPS each question. The other 10 food questions were structured in an identical manner. The foods in each food type were subsequently classified into a ‘recommended’ and ‘regular’ category (Table 1). A preliminary examination of the data revealed that respondents exhibited three types of purchasing pattern for each of the 11 foods. They either purchased the recommended or regular option exclusively, or they purchased a variety of food including bofh the recommended and the regular option. Respondents were assigned the value 3 if they reported purchasing the recommended food exclusively, the value 2 if they purchased both the recommended and regular options, and the value 1 if they purchased only the regular option. Therefore, for all 11 food types, respondents were assigned to either a recommended (3), a mixed (2) or a regular ( 1 ) category. Each of the food items was summed to form a food choice index, which was rescored to range from 0 to 20. The mean and median were 9.2, standard deviation 4.3 and skewness 0.25. Those respondents scoring ‘LO on the index consistently chose the recommended food option and those scoring zero consistently chose the regular option. was possible that perceived availability rather than actual availability was the more important determinant of food choice. If the consumer was unaware of the availability of a recommended alternative, even if it was present where they usually shopped, then it followed that the food item was unlikely to be purchased. Accessibility ofjoods recornmmdcd in dielarq’ Lpiciplines The question of whether food recommended in dietary guidelines was accessible to respondents in the two samples was measured with two precoded structured items and one open-ended item. The first precoded question asked ‘What type of transport do you usually use when you go food shopping for your household?’. The second asked ‘Do you experience any transport difficulties getting to o r from your usual food shopping centre?’. If respondents indicated ‘ves’ to this second question they were then invited to identify (in their own words) the difficulties they encountered. Socioeconomic stutus A defining ;and homogenising feature of the Salvation Army sample was that its members were partly or totally dependent on welfare for their income needs. This, more than any other factor, distinguished them from the electoral roll sample. It was therefore decided to treat the welfare sample as a discrete socioeconomic group. It was also considered important that a comparable indicator should be used to measure socioeconomic inequality within the electoral roll sample. Therefore, total annual household income was used as the measure of socioeconomic status for this group. As part of the questionnaire, respondents were required to indicate their annual household income on a single item consisting of 12 narrow-ranged categories. This measure was subsequently collapsed into three categories: $0 to $13500 (low), $13501 to $43000 (medium) and $45 001 o r more (high). Availability ofjbod recommended in dietmy cpiddines The question of whether food recommended in dietary guidelines was equally available to respondents in the electoral roll and welfare samples was examined by three different approaches. The first consisted of asking respondents ‘Where is the bulk of your household’s food purchased from?’. The second approach involved contacting a major food retailer to determine whether the foods recommended by the dietary guidelines and included in the study were stocked in shops throughout the Brisbane metropolitan region and in Logan city, especially those low-income areas whence the welfare sample was recruited. The third approach asked respondents to indicate whether they agreed, disagreed or were unsure about the availability of the recornmended foods that were used to construct the purchasing index. This item was included because it AfJbrdubilitj of food recommendpd i n the dietnry guidelines T~vo different approaches were used to examine the relationship between economic factors and food purchasing. First, one of Australia’s largest food retailers was given a list of foods recommended by the dietary guidelines and their regular equivalents and asked to provide unit prices for both sets of food items. Again, these foods matched those used in the construction of the purchasing index. The food prices pertained to the period during which the questionnaire data were collected (April 1993). The food items within a particular food group (for example, white versus wholemeal bread) were identical in brand and product size. This had the effect of standardising the food items, ensuring that the price comparisons were valid. Clearly, the prices were likely to vary across brands. According to the food retailer, however, the brands chosen for coinpara tive purposes were commonly purchased and therefore likely to be sufficiently representative. Second, on the basis of a variety of data provided by respondents, three different estimates of household food expenditure were calculated: estimated percentage of total weekly household income spent on food; estimated mean amount ($) spent on food per household per week; and estimated mean amount ($) spent on food per household member per week. Analpis The general linear models procedure in SAS was used to examine the association between socioeconomic status and scores on the food-purchasing index.]’ This procedure (which is appropriate for unbalanced designs) was used in the first instance to compute main-effect (unadjusted) means for each of the groups. Given that age and sex have been shown to be significantly related to both socioeconomic status and food choice it was necessary to make statistical adjustments for these potentially confounding factors.’ This was undertaken by NO. AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1996 vot. 20 F TURRELL Table Table 20: Comparison of socioeconomic groups: mean food 4: Perceived availability of the recommended foods included in the questionnaire ~~ ~ purchasing scores (unadjusted and adiusted estimates) Food purchasing scores Income level High Medium tow Welfare Mean Perceived to be availablea Unadiusted score Adjusted Food F P item and Yes Unsure scorea socioeconomic groupb Vegetable oils high medium low welfare Fruit juice (no added sugar) high medium low welfare Wholemeal pasta high medium low welfare tow-fat cheese high medium low welfare Trim or skim milk high medium low welfare Brown rice high medium low welfare Tinned fruit (no added sugar) high medium low welfare Wholemeal or multigrain bread high medium low welfare Low-fat or diet minced meat high medium low welfare tow-fat yoghurt high medium low ~~ Note: (a) Score adiusted for age and sex Table 2b: Comparison of socioeconomic significance values paired-mean comparisons groups: (PI for adjusted Income level High Medium tow Welfare High Income level Medium Low Welfare 1 .o 0.348 0.028 0.0001 1.o 0.094 0.0001 1 .o 0.0001 1 .o Table 3: Shopping facilities used by respondents when pur- chasing the bulk of the household's food Income level High Shopping facility Medium low Welfare n = 7 7 n=166 n = 8 1 n = 6 8 % % % % Large supermarket exclusively 65 Local suburban shop or 7-day convenience store 1 Specialty shops (e.g. butcher grocer, baker) Large supermarket plus specialty shops including age and sex as covariates within the linear model. Least-squares (adjusted) means were then computed for each of the groups. The general linear models procedure was also used to examine the relationship between socioeconomic status and estimated food expenditure. The relationship between socioeconomic status and a number of the other explanatory variables was examined with the chisquare test. Results The sex, age and labour force profiles of the two samples have been examined in detail elsewhere.'j This earlier analysis showed that the electoral roll sample was not too dissimilar from the general p o p ulation in its sociodemographic characteristics. The welfore Notes: (a) The few respondentswho perceived some food items to be unavailable are not included. (b) Marginal frequencies for each group: high 76-77, medium 165-1 67, low 75-81, welfare 68 xz for differences between income groups: P > 0.05 AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1996 vot. 20 NO. 6 F FOOD CHOICES O SOCIOECONOMIC GROUPS F welfare sample however, was quite different in its composition and displayed many of the characteristics commonly associated with social and economic disadvantage. The groups were compared in terms of their mean scores on the food-purchasing index (Table 2). The analysis of variance test showed that the unadjusted and adjusted mean scores between at least two of the groups were significantly different. The adjusted paired-mean comparisons showed that the welfare group scored significantly lower on the purchasing index than each of the other three groups. These findings indicate that welfare respondents (independent of age and sex) were more consistent purchasers of regular foods, whereas respondents in the high socioeconomic group were more likely to purchase recommended foods consisten tly. Contact with a major food retailer revealed that all of the recommended foods included in the questionnaire were readily available at large food outlets in Brisbane and Logan city. As most respondents usually purchased food from these types of estal, lishments, it would appear that availability was not likely to be a factor influencing food choice. As was argued earlier, however, perceived availability rather than actual availability may have been more important in decision making. When the four groups were compared on 10 items measuring perceived availability, it was evident that most respondents from each group reported that all of the recommended foods were available where they usually shopped (Table 4). A series of chi-square tests showed that the groups did not differ significantly in terms of their perceptions about the availability of food recommended by the dietary guidelines. Availability ofJbod recommended in dietary guidelines The four groups were compared in terms of the types of shopping facility used when purchasing food (Table 3). Over 96 per cent of respondents in each of the groups reported that they usually purchased the bulk of their household’s food at a large supermarket (either exclusively or in conjunction with a number of specialty shops). There was no evidence, from these data at least, that respondents in the low-income group or welfare sample were buying most of their food at small suburban shops or convenience s8tores. Table 5: Type of transport used by respondents when purchasing the households food Income level Medium Low Welfare n = 168 n = 8 0 n = 6 8 % % % 92 2 1 2 2 61 6 23 69 6 12 9 4 Accessibility o food recommended by the dietary guideline f The socioeconomic groups were compared in terms of the types of transport they reported using when shopping for the household’s food (Table 5 ) . There was a significant association between these variables (f = 74.0, 12 df, P < 0.0001). Specifically,while most respondents from all four groups used their own private vehicles when shopping, respondents in the low-income and welfare groups were least likely to report that they used this form of transport. Respondents in these two groups were more likely to purchase the household’s food using public transport, or a friend’s or relative’s private vehicle, or they simply walked. Respondents were subsequently asked if they experienced any transport difficulties when shopping, and over 94 per cent of each group reported that they encountered no problems travelling to or from their usual shopping facility. Affordability ofjood recommended in dietary guidelines The 1988-89 household expenditure survey found that money spent on food and nonalcoholic beverages comprised the single largest component of average weekly expenditure on commodities and services, accounting for approximately 15.0 per cent of mean total weekly income.” It was estimated that households in the electoral roll sample spent approximately 17.0 per cent of their total weekly income on food, which is very similar to the national estimate. This gross estimate, however, masks a considerable degree of socioeconomic variability in food expenditures (Table 6). Respondents in the high-income group spent approximately 9 per cent of their total weekly household income on food, whereas the low-income and welfare groups spent about 30 per cent ( F =51.0, 3 df, P = 0.0001). Households in each of the four groups spent significantly different absolute amounts on food each week (F=3.4, 3 df, P = 0.0177; Table 6). It was possible that these differences were in part accounted for by variations in the size of the households in each group. For example, households in the high, medium, low and welfare groups averaged 3.2, 2.6, 1.8 and 3.6 members respectively. The estimated VOL. Transport Own private vehicle Friend’s or relative’s vehicle Bus, train or taxi Walk Combination of above High n=77 % Table 6: Estimated food expenditure patterns of households according to income level Income level High Medium Low Welfare n=77 n=166 n = 8 1 n = 6 8 Measure of estimated food expenditure Percentage of total weekly household income Mean amount spent per household per week 9% $102 15% $88 $39 31% $68 30% $79 $27 Mean amount spent per household member per week $38 $41 AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1996 F 20 NO 6 TURRELL Table 7: Retail prices of foods included in the food-choice section of the questionnaire (April 1993) Price per Food iype Minced meat Bread Unit Food items low-fat regular white wholemeal multigrain white hyfibre white’n‘bran low-fat (natural) full-fat (natural) low-fat (fruit) full-fat (fruit) no added sugar (natural juice) in syrup butter table margarine (polyunsaturated) dairy blend white brown full-cream skim (law-fat) Trim (reduced-fat) Shape (calcium-modified) full-fat low-fat white wholemeal no added sugar unit ($) 1 kg 7.99 5.99 1.45 1.45 1 .A5 1.52 1.52 0.60 0.60 0.60 0.60 1.69 1.69 1.55 1.52 1.90 1.31 1.27 2.02 2.02 2.10 2.14 3.65 3.77 1.29 1.29 3.79 3.79 680 g ‘foghurt 200 g Tinned fruit Spreads 825 g 500 g Rice 1 kg 2L Milk Cheese Pasta Fruit iuice 500 g 500 g 2L reaular lsweetenedl V . Source: Woolworths (Queenslond) Pty Ltd amount spent on food in each household per week was adjusted for the number of members compris- ing the household unit. After these adjustments, the amount spent on food per household member per week in the welfare group was significantly less than that spent on members of households in the other three income groups (F=3.2, 3 df, P = 0.0251). One factor that potentially contributes to the food-purchasing difficulties faced by low-income families is food prices. Table 7 presents the retail prices of 10 recommended foods and compares these with their regular equivalents. In most rases, the price difference behveen foods within the same food group was nil-to-small. For example, there was no difference in price between white and wholemeal bread, low-fat and full-fat yoghurt, white and wholemeal pasta, unsweetened and regular tinned fruit and fruit juice, and hill-cream milk and skim (lowfat) milk. There was a small price difference between white and brown rice, calcium-modified milk arid other milk, and full-fat and low-fat cheese. There was a much more substantial price difference however, between lowfat mince and regular mince. Discussion Previous Australian and overseas studies have shown that the food behaviours of groups of low socioeconomic status are least likely to accord with dietary recommendations. The results of this study corroborated these findings. Respondents in the welfare sample were significantly less likely than their more affluent counterparts to choose food recommended in the dietary guidelines when shopping. Studies conducted in the United States and the United Kingdom have found that the food-related behaviours of low-income groups are often constrained and circumscribed by structural, material and economic factors. Specifically, large supermarkets are less likely to be located in disadvantaged areas, groups of low socioeconomic status are sometimes unable to reach shopping facilities that stock food recommended in the dietary guidelines, and the difficult economic circumstances of low-income groups limit their capacity to afford these foods. This study examined these issues in an Australian context. Over 96 per cent of respondents from all four socioeconomic-status groups reported that they regularly purchased their food from a large supermarket and an equally large proportion reported that each of the recommended foods included in the questionnaire was available where they usually shopped. Information provided by a large food retailer confirmed that most supermarkets in the areas studied stocked food recommended by the guidelines. Further, although respondents in the low-income group and welfare sample were less likely to have had access to a private vehicle for shopping purposes, very few indicated that this was a problem. Possibly, for many low-income persons, a lack of private transport is a ubiquitous and accepted feature of day-to-day living. Based on the findings of this study, it appears that recommended foods were readily available and accessible to both affluent and disadvantaged alike. The question of whether low-income groups can afford to purchase food recommended by the guidelines is more complicated. If we take a broad view and ask ‘Do low income persons face difficulties in their attempts to procure food for themselves or their families?’, then the answer is unequivocally ‘yes’. This study and research conducted elsewhere have shown that low-income groups have to allocate a much greater proportion of their household income for food, yet they spend significantly less on food per household member. Moreover, because food is usually the only discretionary item in a lowincome family’s budget, spending on this necessary commodity is likely to be reduced to meet nondiscretionav costs such as rent and bills. An unfortunate consequence of this reduced spending power is that lowincome families are often forced to purchase food less frequently and in smaller quantities, the range and diversity of foods that can be purchased is limited, and the foods bought are very often of a lesser If we take a more narrow view however, and ask, ‘Can low-income groups afford to purchase the type VOL. AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1996 F 20 NO 6 FOOD CHOICES O SOCIOECONOMIC GROUPS F of food which is promulgated in the dietary guidelines?’, then the answer is less certain. The results of studies conducted in the United Kingdom suggest that foods recommended in the dietary guidelines are priced out of the reach of low-income famiBy lies.~,ll.l~ contrast, the few Australian studies examining this issue have presented evidence suggesting that recommended foods and diets are affordable by low-income families.’”“’ In this study, it was evident that, with only a few exceptions, the prices of recommended foods and their regular equivalents were identical or very similar. Of course, the difficulty that we face is determining what constitutes a price difference sufficiently large that it would adversely affect the purchasing decision of the low-income consumer. From our relatively advantaged position, a price difference of 12 cents between full-fat and low-fat cheese seems small, and all other things being equal, is not likely to be a determining factor in the purchase decision. However, when a large number of food items are being purch.ased, the seemingly small difference between the recommended and regular foods may take on a new significance. This point was clearly highlighted by Mooney, who found that a hypothetical diet consisting of 15 recommended food items cost S13.99 per week compared with S8.33 for an equivalent diet comprising regular foods.’ A similar situation can be constructed from the data in Table 7. If we sum the retail prices of the most expensive recommended food items, we find that this hypothetical shopping basket costs $25.58. An identical procedure for the cheapest regular items results in a basket costing $23.34. Only one food item accounts for this disparity in price between the recommended and regular shopping baskets: in 1993, low-fat minced meat cost $7.99/kg compared with $5.99/kg for regular minced meat. Based on this evidence, it seems reasonable to suggest that if the low-income consumer is selective and discerning in his or her purchasing of food, then the cost of a diet that conforms with dietary guidelines need not cost substantially more than one that is at variance with these guidelines. Conclusion structural, material and economic factors did not explain socioeconomic variability in food choice raises the possibility that social-psychological and subcultural factors (beliefs, attitudes and knowledge) account for the differences. This possibility will be explored in detail in a subsequent paper. Acknowledgments The author wishes to thank the officer-s and staff at the Salvation Army headquarters (Brisbane) who helped to recruit low-income families, and Ms Pam Lucas from Woolworths (Qld) Pty Ltd who provided the retail prices for the recommended and regular foods.

Journal

Australian and New Zealand Journal of Public Health – Wiley

Published: Dec 1, 1996

There are no references for this article.