• Open Access

A comparison of Australian families’ expenditure on active and screen-based recreation using the ABS Household Expenditure Survey 2003/04

Authors


Correspondence to:
Ms Lesley King, NSW Centre for Overweight and Obesity, Level 2, K25 – Medical Foundation Building, University of Sydney, NSW 2006. Fax: (02) 9036 3184; e-mail: lking@health.usyd.edu.au

Abstract

Objective: This study aimed to investigate how much households with dependent children spend on active recreation (physical activity) compared with screen-based (sedentary) recreation, according to their household socioeconomic and demographic characteristics.

Methods: The study analysed data from the 2003–04 Australian Bureau of Statistics Household Expenditure Survey, which collected information on household expenditure from a representational cross-section of private dwellings across Australia.

Results: In 2003–04, Australian households with dependent children spent an average of 1.5% and 3.3% of their weekly disposable income on active and screen recreation respectively, and 24.9% of their total active and screen recreation expenditure on active recreation. There was significant variation across household characteristics, with higher income and socioeconomic status households, and families with more than one dependent child more likely to spend a larger portion of their recreation budget on active recreation instead of screen recreation.

Conclusions: Overall, Australian families spend more money on screen recreation items than they do on active recreation, although there are strong economic and cultural gradients in their patterns of expenditure on both active and screen recreation. This suggests that while the costs of active recreation may be a barrier to participation for some families, there are also social and cultural values influencing recreational choices.

Implications: For the first time, specific information on Australian families’ expenditure on active and screen recreation is available. These results contribute to identifying cultural and economic barriers influencing families’ health-related behaviours and their participation in organised physical activity.

There are many health benefits for children and young people from leading an active lifestyle and regularly participating in physical activity. Australia's Physical Activity Recommendations for Children and Young People recommend young people aged from 5 to 18 years spend at least 60 minutes per day in moderate to vigorous physical activity and less than two hours a day in small screen recreation, which includes watching TV, playing computer games and other electronic media for entertainment.1,2 Data on New South Wales (NSW) school students shows that while 73% of boys and 64% of girls aged 12- 16 years meet the activity guidelines, the majority exceed the recommended time for small screen recreation, with 75% of boys and 66% of girls engaging in more than two hours per day of small screen recreation.3 No information is available on the levels of physical activity and small screen recreation of younger children in NSW or Australia.

While the majority of children aged 12–15 years report being involved in organised physical activity, there are many factors which influence the extent and levels of participation. Systematic differences are most influenced by gender and age, although family structure and socio-economic status also appear to play a role, with children in single parent families and low SES families participating less.4 Other studies report similar results, with having a family car and the attitudes of parents found to be relevant factors for participation in sport and recreation, as well as other extra-curricular activities.5,6 Children's participation in organised sport is influenced by family type and parents’ employment status, as well as socioeconomic status.7 The 2003 Survey of Children's Participation in Culture and Leisure Activities shows that 52% of children from families in the lowest Socioeconomic Status (SES) quintile participated in some form of sport or dancing, compared to 82% for children from the highest quintile.7 Participation in sport by children in one-parent families where the parent was not employed was 39.7%, compared with 71.5% for children in couple families with both parents employed.

Other studies have identified costs as a barrier to participation. A study of families with children in specific junior sports in Victoria and Queensland in 1997 concluded that family income and structure were key factors in determining the likelihood of a child's involvement in junior sport.8 Similar findings have been reported from the UK.9

The cost of organised physical activity has been frequently identified as a barrier in qualitative research with parents.10 Similarly, professionals and service providers, such as GPs and teachers, perceive that cost is a barrier for parents.11,12 Cost, as well as transport and time, is perceived by young people as a factor influencing their participation in activity.4

While actual and perceived costs appear to be barriers that limit participation in active recreation for some young people and families in Australia, there is no comprehensive picture of what the costs actually are, how they correspond to different income levels or how they compare to expenditure on other forms of recreation, such as screen recreation.

The Household Expenditure Survey (HES) conducted by the Australian Bureau of Statistics (ABS) provides a reliable source of information on actual expenditure on a range of recreation items that has not been fully examined in this way. Using the 2003/04 HES data for households with dependent children, this study examined:

  • Expenditure on active and screen recreation relative to income levels.
  • Expenditure on active recreation compared with screen-based recreation.
  • How these measures varied according to household socioeconomic and demographic characteristics.

Methods

Data source

The HES is a cross-sectional survey conducted by the ABS every five years, most recently in 2003–04. It used a stratified, multistage cluster design to ensure a representative sample of private dwellings from urban and rural areas of Australia. Information was collected through personal interviews with a usual resident, distributed randomly over a twelve month period to capture yearly income and expenditure patterns. Participating households were required to maintain a diary recording their expenditure over a two week period and recall any major purchases during the last three months. There were 6,957 households included in the HES, representing a response rate of 71%.13 De-identified HES responses were accessed through Confidentialised Unit Record Files (CURFs) provided on CD-ROM, and via secure online Remote Access Data Laboratory (RADL).

This study examined HES data for the sub-sample of couple or sole parent households with at least one dependent child, and a positive disposable income. Dependent children were defined as all persons aged less than 15 years and full-time students aged 15–24 years with a parent and no partner or child of their own in the household.13

Data description

To represent the economic resources available to meet household requirements we used disposable household income, defined as gross household income less personal income tax and the Medicare levy (which is a compulsory income-based contribution made by residents to help fund Australia's universal health care system). Weekly expenditure information, available for 625 separate expenditure items in the HES, was used to estimate aggregated average weekly expenditure by households on active and screen recreation. Sampling weights were applied to ensure that household estimates of income and expenditure accurately reflect the distribution of the Australian population residing in private dwellings.

In order to identify expenditure associated with active and screen recreation, we distinguished five groups of household expenditure items involving, or useable for physical activity or sedentary screen-based recreation. Core active recreation included expenditure on items such as equipment, specialised clothing, or services associated with organised sport and activities or non-organised activities, including swimming, games, or walking. Sports footwear and pets were identified as two supplementary expenditure groups that are potentially, but not necessarily, associated with active recreation. Core screen recreation included expenditure on equipment or services associated with activities such as watching TV and films, and playing video games. Given that computers can serve educational, as well as entertainment purposes, expenditure relating to computer hardware, software or services, and internet usage charges was grouped separately.

For the purposes of this study we focused on core active recreation and total screen recreation (core plus computers), and investigated separately the impact of including sports footwear and pets in active recreation, and excluding computers from screen recreation.

From the set of household socioeconomic and demographic characteristics available as standard HES data items, we identified several that were potentially associated with household active and screen recreation expenditure:

  • family structure (couple; or sole parent),
  • number of dependents (one; two; three; or four or more),
  • SES (ABS Socio-economic indexes for Areas (SEIFA) 2001 index of relative socioeconomic disadvantage (IRSD) quintile),14
  • remoteness area (major city; inner regional; or outer regional and remote Accessibility/Remoteness Index of Australia (ARIA) score),15
  • parents’ average hours worked per week (not employed; 5–19; 20–34; 35–49; or 50 or more),
  • dwelling structure (separate house; semi/row/terrace house; or flat/unit/apartment),
  • parents’ highest education (tertiary; vocational; school; or school not completed),
  • parents’ geographic region of birth (Australia and New Zealand; Europe; Asia; Africa and the Middle East; or other, including households where parents were born in disparate geographical regions).

Household disposable income quintiles calculated from the study sub-sample, were also derived.

Data analysis

Weekly household expenditure was summarised by expenditure item and aggregated for each defined variable related to active and screen recreation. For each household we calculated expenditure on both active and screen recreation as a proportion of disposable income. In households with any expenditure on recreation, we calculated expenditure on active recreation as a proportion of combined expenditure on active and screen recreation. Summary statistics were calculated for all households in the study sample, and used to define dichotomous outcomes that indicate whether a household spent relatively more or less than average on active and screen recreation given their financial circumstances. Households were characterised into three (non-exclusive) categories:

  • those that spent a greater than average proportion of their weekly disposable income on active recreation were described as ‘promoting active recreation’,
  • those that spent a greater than average proportion of their weekly disposable income on screen recreation were described as ‘promoting screen recreation’, and
  • those that spent a greater than average proportion of their combined active and screen recreation expenditure on active recreation were described as ‘promoting active over screen recreation’.

Logistic regression analysis was used to investigate associations between outcomes and household socioeconomic and demographic characteristics with crude and adjusted odds ratios (OR). Parsimonious adjusted models were obtained for each outcome by using the likelihood-ratio test to exclude characteristics that did not significantly improve model fit (p>0.05). Overall, model fit was assessed by the Hosmer-Lemeshow goodness-of-fit test16, with variance inflation factors (VIF) calculated to inspect for strong correlation between household characteristics in adjusted models, otherwise known as multicollinearity.17

To examine the impact of different definitions on the associations of outcome variables and household characteristics, models were re-specified with sports footwear and pets included in active recreation, and computers excluded from screen recreation. To similarly investigate whether associations differ for households with younger dependent children, we re-specified models to the sub-sample of households with all dependent children aged less than 15 years.

SAS version 9.1 was used for all statistical analyses.18

Results

There were 2,398 households with dependent children identified from the 2003/04 HES. Households in the sample had median weekly disposable income of $1,028.86, and an average of 1.9 dependent children (1.5 aged less than 15 years; 0.4 aged 15 to 24 years). See Table 1 for further sample descriptive characteristics.

Table 1.  Descriptive statistics of the HES sample of households with dependent children, Australia, 2003-04.
Household characteristicN = 2,398
 n (%)
Family structure 
 Couple1,928 (804)
 Sole parent470 (19.6)
Number of dependents 
 1911 (38.0)
 2971 (40.5)
 3382 (15.9)
 4 or more133 (5.6)
Disposable income per week 
 <$635479 (20.0)
 $635-$912479 (20.0)
 $913-$1,164482 (20.1)
 $1,165-$1,513479 (20.0)
 >$1,513478 (19.9)
SES quintile 
 Lower 20%405 (16.9)
 2nd480 (20.0)
 3rd531 (22.1)
 4th518 (21.6)
 Upper 20%464 (19.4)
Hours worked per week 
 Not employed336 (14.0)
 1-19265 (11.0)
 20-34932 (38.9)
 35-49732 (30.5)
 50+134 (5.6)
Remoteness area 
 Major City1,587 (66.2)
 Inner Regional542 (22.6)
 Outer regional and remote269 (11.2)
Dwelling type 
 Separate House2,171 (90.6)
 Semi, Row or Terrace House107 (4.4)
 Flat, Unit, or Apartment120 (5.0)
Highest education 
 Tertiary723 (30.1)
 Vocational1,027 (42.8)
 School239 (10.0)
 Did not complete school410 (17.1)
Geographic region of birth 
 Australia & New Zealand1,226 (51.1)
 Europe96 (4.0)
 Asia161 (6.7)
 Africa & Middle East48 (2.0)
 Other867 (36.2)

Table 2 shows that during 2003/04, Australian households with dependent children spent an average of $14.58 per week on active recreation and $31.69 per week on screen recreation. The majority of active recreation expenditure was on sports fees and charges (62.6%), primarily sports lessons, sporting club subscriptions, and health and fitness studio charges. Most screen recreation expenditure was on home computer equipment (25.6%), followed by televisions (15.1%).

Table 2.  Average weekly expenditure by expenditure group and item, Australia, 2003/04.
 Average weekly expenditure
Expenditure groups and items$% group% totala
  1. Notes:

  2. (a) % total =% of total active or screen recreation expenditure

Active recreation24.36100.0100.0
 Core14.58100.059.9
  Sports equipment4.1628.517.1
  Hire & repair of sports equipment0.261.81.1
  Sports fees & charges9.1362.637.5
  Other recreational services1.037.14.2
 Sports footwear1.38100.05.7
 Pets8.40100.034.5
Screen recreation31.69100.0100.0
 Core19.06100.060.1
  Televisions4.7825.115.1
  Video equipment2.3412.37.4
  Pre-recorded media3.9120.512.3
  Hire of televisions & video media & equipment1.759.25.5
  Other recreational services3.0115.89.5
  Pay tv fees3.2717.210.3
 Computers12.63100.039.9
  Home computer equipment8.1164.225.6
  Pre-recorded media0.574.51.8
  Internet charges3.9531.312.5
Total56.05100.0100.0

During 2003–04, households in the study sample spent an average of 1.5% and 3.3% of their weekly disposable income on active and screen recreation respectively, and 24.9% of their combined active and screen recreation expenditure on active recreation.

Modelling results

There was significant variation in the crude and adjusted odds of promoting active recreation across all household characteristics. Couple households, households with more dependents, higher disposable income and higher SES, and households with parents working more hours, all had greater crude odds of spending a greater than average proportion of disposable income on active recreation. Families living in more remote areas, smaller dwellings, with less formal education, or with parents not born in Australia, New Zealand (NZ) or Europe, had lower odds of promoting active recreation in this way. The adjusted model, which fits the data well (Hosmer-Lemeshow χ2= 4.3, p= 0.8270), indicated that once we accounted for other characteristics, households with two or three dependents, in the 3rd to 4th disposable income quintile, and 4th SES quintile continued to have greater odds of promoting active recreation, and those with less educated parents or parents born in Asia had lower odds of doing so (Table 3).

Table 3.  Odds ratios for promoting active recreation by household characteristic, Australia, 2003-04.
 Promote active recreationa
Household characteristicn (%)Unadjusted OR (95% CI)bAdjustedc OR (95% CI)b
  1. Notes:

  2. (a) Household expenditure on active recreation > 1.5% of disposable income

  3. (b) All 95% CI that exclude one indicate a statistically signifcant OR and are in bold

  4. (c) Adjusted for all variables included in the fnal model

Family structure   
 Couple401 (20.8)1.00 
 Sole parent64 (13.5)0.60 (0.45-0.80) 
Number of dependents   
 1131 (14.4)1.001.00
 2197 (20.3)1.51 (1.19-1.51)1.48 (1.16-1.90)
 3109 (28.5)2.37 (1.78-2.37)2.11 (1.57-2.85)
 4 or more28 (20.7)1.55 (0.98-1.55)1.49 (0.93-2.40)
Disposable income per week   
 < $63551 (10.6)1.001.00
 $635-$91288 (18.4)1.91 (1.31-1.91)1.40 (0.91-2.16)
 $913-$1,164109 (22.6)2.47 (1.72-2.47)1.63 (1.04-2.54)
 $1,165-$1,513114 (23.9)2.65 (1.85-2.65)1.72 (1.08-2.72)
 > $1,513102 (21.4)2.30 (1.60-2.30)1.33 (0.82-2.15)
SES quintile   
 Lower 20%57 (14.0)1.001.00
 2nd77 (16.0)1.17 (0.81-1.17)1.01 (0.69-1.48)
 3rd97 (18.3)1.38 (0.97-1.38)1.22 (0.84-1.77)
 4th126 (24.4)1.98 (1.40-1.98)1.58 (1.10-2.29)
 Upper 20%107 (23.1)1.85 (1.30-1.85)1.41 (0.95-2.09)
Hours worked per week   
 Not employed35 (10.5)1.001.00
 1-1938 (14.2)1.41 (0.87-1.41)0.94 (0.55-1.62)
 20-34233 (25.0)2.85 (1.95-2.85)1.42 (0.87-2.30)
 35-49131 (17.9)1.87 (1.25-1.87)0.95 (0.56-1.60)
 50+27 (20.4)2.18 (1.26-2.18)1.13 (0.59-2.17)
Remoteness area   
 Major City320 (20.2)1.00 
 Inner Regional107 (19.7)0.97 (0.76-0.97) 
 Outer regional and remote37 (13.9)0.64 (0.44-0.64) 
Dwelling type   
 Separate House431 (19.9)1.00 
 Semi, Row or Terrace House20 (19.0)0.95 (0.58-0.95) 
 Flat, Unit or Apartment13 (10.6)0.48 (0.27-0.48) 
Highest education   
 Tertiary174 (24.1)1.001.00
 Vocational195 (19.0)0.74 (0.59-0.74)0.75 (0.58-0.97)
 School47 (19.5)0.76 (0.53-0.76)0.90 (0.61-1.33)
 Did not complete school48 (11.7)0.42 (0.30-0.42)0.57 (0.39-0.84)
Geographic region of birth   
 Australia & New Zealand278 (22.7)1.001.00
 Europe24 (25.0)1.14 (0.70-1.14)1.24 (0.75-2.04)
 Asia21 (13.4)0.53 (0.33-0.53)0.50 (0.31-0.82)
 Africa & Middle East9 (19.7)0.84 (0.41-0.84)0.89 (0.42-1.89)
 Other131 (15.1)0.61 (0.48-0.61)0.80 (0.62-1.02)

There was less variation amongst households in promoting screen recreation. Sole parent households, and households with two dependent children, had greater crude odds of spending a greater than average proportion of their disposable income on screen recreation; while households with higher disposable income, the highest SES, parents working greater hours, and households living in remote areas or in a flat, unit, or apartment, had lower crude odds of promoting screen recreation. After adjusting for other characteristics (Hosmer-Lemeshow χ2= 7.6, p= 0.4763), similar associations remained for number of dependents, disposable income, remoteness area, and dwelling type (Table 4).

Table 4.  Odds ratios for promoting screen recreation by household characteristic, Australia, 2003-04.
 Promote screen recreationa
Household characteristicn (%)Unadjusted OR (95% CI)bAdjustedc OR (95% CI)b
  1. Notes:

  2. (a) Household expenditure on screen recreation > 3.3% of disposable income

  3. (b) All 95% CI that exclude one indicate a statistically signifcant OR and are in bold

  4. (c) Adjusted for all variables included in the fnal model

Family structure   
 Couple444 (23.0)1.00 
 Sole parent133 (284)1.32 (1.06-1.66) 
Number of dependents   
 1199 (21.8)1.001.00
 2257 (26.5)1.30 (1.05-1.60)1.28 (1.03-1.59)
 392 (24.0)1.14 (0.86-1.51)1.18 (0.88-1.57)
 4 or more30 (22.5)1.04 (0.67-1.61)1.08 (0.69-1.68)
Disposable income per week   
 < $635159 (33.2)1.001.00
 $635-$912121 (25.3)0.68 (0.52-0.90)0.64 (0.48-0.85)
 $913-$1,164101 (21.0)0.53 (0.40-0.71)0.50 (0.37-0.67)
 $1,165-$1,513111 (23.1)0.60 (0.45-0.80)0.56 (0.42-0.75)
 > $1,51385 (17.8)0.44 (0.32-0.59)0.40 (0.29-0.54)
SES quintile   
 Lower 20%107 (26.3)1.00 
 2nd118 (24.5)0.91 (0.67-1.23) 
 3rd125 (23.5)0.86 (0.64-1.16) 
 4th134 (25.9)0.98 (0.73-1.32) 
 Upper 20%95 (20.5)0.72 (0.53-0.99) 
Hours worked per week   
 Not employed104 (31.1)1.00 
 1-1965 (24.5)0.72 (0.50-1.03) 
 20-34228 (24.5)0.72 (0.55-0.95) 
 35-49160 (21.9)0.62 (0.46-0.83) 
 50+20 (15.2)0.40 (0.24-0.67) 
Remoteness area   
 Major City385 (24.3)1.001.00
 Inner Regional143 (26.4)1.12 (0.90-1.40)1.02 (0.81-1.28)
 Outer regional and remote50 (18.5)0.71 (0.51-0.98)0.63 (0.45-0.88)
Dwelling type   
 Separate House533 (24.6)1.001.00
 Semi, Row or Terrace House26 (24.2.0)0.98 (0.62-1.54)0.85 (0.53-1.35)
 Flat, Unit or Apartment19 (15.5)0.57 (0.34-0.93)0.48 (0.29-0.81)
Highest education   
 Tertiary161 (22.3)1.00 
 Vocational249 (24.2)1.11 (0.89-1.39) 
 School67 (28.3)1.37 (0.98-1.91) 
 Did not complete school100 (24.5)1.13 (0.85-1.50) 
Geographic region of birth   
 Australia & New Zealand276 (22.5)1.00 
 Europe27 (27.7)1.32 (0.83-2.11) 
 Asia40 (25.0)1.15 (0.79-1.68) 
 Africa & Middle East14 (28.1)1.35 (0.71-2.56) 
 Other222 (25.6)1.19 (0.97-1.45) 

For the promotion of active over screen recreation, there was significant variation for all household characteristics, except remoteness. The crude odds of spending a greater than average proportion of combined (active and screen) recreation expenditure on active recreation increased for households with more dependents, higher disposable income and SES, and parents who work more hours. Sole parent households and those living in smaller dwellings, with less educated parents, or parents not born in Australia, NZ or Europe, had lower odds of promoting active of screen recreation in this way. In the adjusted model (Hosmer-Lemeshow χ2= 6.2, p= 0.6237), households with more dependents, higher disposable income, or in the 4th SES quintile continued to have increased odds of promoting active over screen recreation; and households with parents born in Asia, Africa and the Middle East, and other geographic regions, had reduced odds of doing so, after adjusting for other characteristics (Table 5).

Table 5.  Odds ratios for promoting active over screen recreation by household characteristic, Australia, 2003-04.
 Promote active over screen recreationa
Household characteristicn (%)Unadjusted OR (95% CI)bAdjustedc OR (95% CI)b
  1. Notes:

  2. (a) Household expenditure on active recreation > 24.9% of total household expenditure on active and screen recreation

  3. (b) All 95% CI that exclude one indicate a statistically signifcant OR and are in bold

  4. (c) Adjusted for all variables included in the fnal model

Family structure   
 Couple640 (36.9)1.00 
 Sole parent95 (25.5)0.59 (0.46-0.76) 
Number of dependents   
 1212 (27.6)1.001.00
 2316 (36.8)1.53 (1.24-1.53)1.53 (1.23-1.90)
 3153 (43.0)1.98 (1.52-1.98)1.83 (1.39-2.41)
 4 or more54 (43.5)2.02 (1.37-2.02)1.84 (1.22-2.77)
Disposable income per week   
 <$63572 (20.3)1.001.00
 $635-$912132 (31.9)1.83 (1.31-1.83)1.59 (1.08-2.35)
 $913-$1,164178 (40.7)2.68 (1.94-2.68)2.15 (1.43-3.21)
 $1,165-$1,513169 (38.0)2.40 (1.74-2.40)1.91 (1.26-2.89)
 >$1,513185 (40.2)2.63 (1.91-2.63)2.17 (1.42-3.32)
SES quintile   
 Lower 20%93 (28.1)1.001.00
 2nd135 (33.0)1.26 (0.92-1.26)1.18 (0.85-1.64)
 3rd157 (33.4)1.28 (0.94-1.28)1.23 (0.89-1.70)
 4th187 (40.4)1.74 (1.28-1.74)1.52 (1.10-2.10)
 Upper 20%163 (37.6)1.54 (1.13-1.54)1.32 (0.94-1.85)
Hours worked per week   
 Not employed60 (23.6)1.001.00
 1-1955 (24.7)1.06 (0.70-1.06)0.70 (0.44-1.13)
 20-34348 (41 4)2.29 (1.66-2.29)1.14 (0.75-1.74)
 35-49230 (34.5)1.71 (1.22-1.71)0.83 (0.53-1.29)
 50+43 (33.9)1.66 (1.04-1.66)0.81 (0.46-1.41)
Remoteness area   
 Major City494 (34.6)1.00 
 Inner Regional166 (34.8)1.01 (0.81-1.01) 
 Outer regional and remote75 (36.9)1.11 (0.82-1.11) 
Dwelling type   
 Separate House689 (35.7)1.00 
 Semi, Row or Terrace House26 (27.0)0.67 (0.42-0.67) 
 Flat, Unit or Apartment20 (24.3)0.58 (0.35-0.58) 
Highest education   
 Tertiary264 (39.3)1.00 
 Vocational312 (34.1)0.80 (0.65-0.80) 
 School60 (30.0)0.66 (0.47-0.66) 
 Did not complete school98 (30.8)0.69 (0.52-0.69) 
Geographic region of birth   
 Australia & New Zealand443 (40.0)1.001.00
 Europe38 (44.0)1.18 (0.76-1.18)1.24 (0.78-1.96)
 Asia35 (24.5)0.49 (0.33-0.49)0.53 (0.35-0.80)
 Africa & Middle East8 (18.2)0.33 (0.16-0.33)0.36 (0.16-0.77)
 Other212 (29.1)0.61 (0.50-0.61)0.76 (0.61-0.95)

Multicollinearity was not an issue in the adjusted models, as VIF for all household characteristics were low (< 4).

The effect of including sports footwear and pets in active recreation

When sports footwear and pets were included in active recreation, average weekly expenditure increased to $24.36, and the average proportion of disposable income and combined recreation expenditure spent on active recreation increased to 2.6% and 40.7% respectively. Logistic regression modelling of active recreation outcomes revealed that associations with most household characteristics were considerably weaker than for core active recreation (family type, disposable income, SES, parents hours worked, and parents education all p>0.05).

Effect of excluding computers from screen recreation

When computers were excluded, average weekly expenditure on screen recreation fell to $19.06, and the average household spent 2.0% of their disposable income and 34.3% of their combined recreation expenditure on active recreation. Logistic regression modelling of screen recreation outcomes indicated that excluding computers had little effect on associations with household characteristics, as similar ORs were observed across most characteristics.

Effect of excluding households with dependents aged more than 15 years

There were 1,688 households included in the HES with all dependents aged less than 15 years. There was little difference in expenditure patterns by this subgroup, with an average of $12.77 and $24.96 per week spent on active and screen recreation respectively. The average proportion of disposable income (1.6%), and combined recreation expenditure (25.8%) spent on active recreation increased slightly, while the average proportion of disposable income spent on screen recreation decreased slightly to 3.2%. Consequently, logistic regression modeling of associations between outcomes and household characteristics produced similar results to the full sample of households with dependent children.

Discussion

In the context of low to moderate levels of physical activity and increased prevalence of childhood obesity, socio-economic variations in patterns of expenditure on active and screen recreation are of public health relevance. For the first time, comparative information is available on patterns of expenditure on active and screen recreation. In 2003–4 Australian households with dependent children spent about $14.58 per week on active recreation, which was about half of average expenditure on screen recreation.

Patterns of expenditure on active and screen recreation varied substantially by socio-economic factors. Increased household income was associated with increased odds of promoting active recreation and active recreation over screen recreation, and decreased odds of promoting screen recreation. This effect was independent of the effects of socio-economic status. Most simply, this suggests that the cost of active recreation is a limiting factor for some families in supporting active recreation. In terms of other household characteristics, households with more dependent children had significantly greater odds of promoting active recreation over screen recreation. This finding reflects increased expenditure on active recreation with fairly consistent expenditure on screen recreation, as the number of children increases.

The independent effects of social and cultural factors indicate that costs are not the only influences on active recreation, and that expenditure is also influenced by activity preferences and social norms. Education level influenced expenditure patterns, with households with a tertiary educated adult having greater odds of promoting active recreation, and active recreation over screen recreation, in most cases. Cultural differences according to region of birth were evident, with households with parents from Asia, Africa and the Middle-East, or other geographical regions, having significantly lower odds of promoting active recreation over screen recreation, and in some cases active recreation, independent of their income or SES. The characteristics associated with lower expenditure on active recreation that have been identified in this study can provide guidance in determining target groups for community-based sport and recreation programs.

While the expenditure patterns observed are associated with social, economic and cultural factors, there are also some inherent differences in the nature of the expenditure items related to active and screen recreation. Firstly, screen recreation expenditure predominately involves the purchase of durable goods, which are available for use continuously rather than in a single episode, and are primarily home-based and available for use at a household, as opposed to individual, level. By contrast, active recreation expenditure items are often non-durable, or durable goods purchased for individual use. Compared to screen recreation, active recreation is low-tech, making the items less expensive, less novel and in many cases, less marketed. Active recreation is also more likely to occur away from home, and this can involve associated transport costs, which are not included here as they could not be specifically identified from the HES. Despite the cost of new, high-tech screen equipment, there is an apparent efficiency and economy for screen recreation at a household level, in terms of time spent and number of users.

In general, expenditure patterns reflect contemporary social and leisure patterns, which are known to be characterized by the increasing adoption of home communication products, driven in turn by new technologies and major marketing campaigns.19 While expenditure on screen recreation contributes to economic growth, it may not be optimal for preventing childhood sedentarism and obesity. Given the nature and patterns of expenditure, any increase in active recreation and reduction in screen recreation may be accompanied, at least initially, by slight negative economic growth. At the same time, the health benefits of physical activity have medium and long-term economic benefits, related to reduced morbidity.20

As home computer equipment was the largest single item contributing to the screen recreation expenditure variable, and as computers can be used for educational as well as recreational purposes, screen recreation could be considered to be unduly weighted. However, excluding computers from screen recreation did not substantially alter the patterns of expenditure on active or screen recreation according to income, education level or cultural background.

In interpreting the findings, we recognise that expenditure is not a direct reflection of actual recreation behaviour. Expenditure on active recreation items may primarily derive from participation in organised activity, and is unlikely to reflect cost-free or incidental physical activities. Similarly, screen recreation is usually, but not necessarily, sedentary.

This study is also limited by the specific expenditure items collected by the HES, which is designed as a generic survey and not specifically to investigate patterns of recreation. The HES is a comprehensive and rigorous dataset, however, and provides data that gives a new, economic angle to the understanding of health-related behaviours.

These results contribute to identifying cultural and economic barriers influencing families’ health-related behaviours and their participation in organised physical activity. This was designed as an exploratory study, and does not provide sufficient information to identify appropriate public health responses. It does provide insights into factors influencing recreational choices and barriers for physical activity. This study suggests that more specific investigation of links between families’ expenditure, time use and behaviour patterns, as well as studies on families’ values, expectations and perceptions about the relative costs of different types of recreation, would add to our understanding and potential policy responses.

Overall, the findings of this study are consistent with descriptive accounts of how contemporary society has increasingly adopted sedentary, screen-based leisure activities.

Acknowledgements

Robert Aitken was funded through the NSW Biostatistical Officer Training Program, NSW Department of Health.

Ancillary