• Open Access

Changing associations of Australian parents' physical activity with their children's sport participation: 1985 to 2004


Correspondence to:
Dr James Dollman, University of South Australia, GPO Box 2471, Adelaide, South Australia 5000. Fax: (08) 8302 6558; email: james.dollman@unisa.edu.au


Objective: The socio-ecological milieu of children's physical activity is changing, perhaps causing perturbations within the causal ‘web’ that explains physical activity behaviours. It is unclear if the relative importance of parental role modelling is changing. Accordingly, this study examined associations of child-perceived parent physical activity and children's sport participation in 1985 and 2004.

Methods: In 1985 (179 girls, 211 boys) and 2004 (210 girls, 218 boys), Australian schoolchildren (9–15 years) in the same eight schools were surveyed on sport participation and perceptions of parents’ physical activity.

Results: In the 1985 sample, girls with active fathers played more sport. In 2004, boys and girls with active fathers or active mothers reported higher sport participation. In 1985, there were no differences in sport participation between those with both, either or neither parent active. In 2004, sport participation was highest among boys and girls with both parents active.

Conclusions: These results underscore the current role of parents as socialising agents for physical activity.

Implications: Intervention design should be founded on the most recent evidence of children's physical activity correlates.

The importance of physical activity among children for immediate and longer term health benefits is well established.1 Accordingly, close attention has been directed to the intra-personal and environmental influences on children's physical activity. Current social ecological models such as the Youth Physical Activity Promotion Model2 postulate a complex web of direct and indirect agencies, representing the child's psychological disposition as well as inter-connected features of the social and physical environments. The relative importance of these influences is yet to be confirmed. Recent reviews present quite a confused picture,3,4 with few variables consistently linked with physical activity across studies that vary by demographic characteristics of the samples. This suggests that the associations of physical activity and its predictors are moderated by other influences within the ‘causal web’. Moreover, predictors of children's physical activity vary according to the context and setting in which the physical activity takes place, such as physical education, organised sport and free play at home or in the school playground.5

There is evidence for the supportive roles of parents in shaping children's attitudes to physical activity and providing access to opportunities.6 Parents influence their children's physical activity in a range of ways, including role modelling through their own participation.7 Studies from various countries have reported significant relationships between the physical activity patterns of parents and their children.4 Overall, father's activity influences children's physical activity patterns regardless of sex, whereas mothers’ activity appears to be more consistently related to their daughters’ rather than sons’ physical activity.4

The socio-ecological milieu in relation to youth physical activity is changing, characterised by, for example, ‘time-poor’ parents, denser traffic, more ubiquitous electronic entertainments and less safe neighbourhoods.8 While parental role modelling has gained acceptance as a socialising agent for young people's physical activity, it is unclear how its relative importance has been modified by recent micro- and macro-environmental changes. Published studies vary in sample characteristics, measurement and analytical techniques, as well as the context of physical activity under study, for instance organised sport or total energy expenditure. In order to address this question more clearly, there is a need to have data collected in demographically matched samples with the same methodologies, and interpreted using the same analytical techniques.

The aim of this study was to examine associations of child-perceived parent physical activity and children's sport in 1985 and 2004, using the same sampling strategy and survey questionnaire.


The Australian Schools Health and Fitness Survey (ASHFS), 1985

In 1985, a nationally representative sample of 8,484 children (7–15 years) participated in the Australian Schools Health and Fitness Survey (ASHFS), providing a broad range of data on health, fitness and behaviours. Participants aged 9–15 years (n=6,659) completed a questionnaire on health behaviours and influences on these behaviours. The sampling design has been described in detail elsewhere.9 Briefly, a self-weighted sample was drawn by selecting schools with probability proportional to enrolment numbers, and then sampling 10 children per age/sex category, with replacement, from each school. One hundred and nine schools agreed to participate (90% of those approached), and 78% of students selected in the initial sample received parental consent to participate.

The South Australian Physical Activity Survey, 2004

In 1985, 10 South Australian schools were included in the ASHFS sample, with eight of these agreeing to participate again in 2004. One of the 10 schools from 1985 no longer exists, while in 2004 a single sex non-government school chose not to participate. Within participating schools, random selection of participants replicated the 1985 sampling strategy. In 2004, 82% of parents provided consent for their children to complete the survey. The decision to re-visit the South Australian schools from the ASHFS, rather than employ the 1985 school sampling strategy in 2004, was based on the relatively small number of schools and the likelihood that 10 randomly selected schools in 2004 could result in a demographic profile substantially different to the 1985 South Australian sample. As the demographic characteristics of the schools had not changed between 1985 and 2004, the proportions of urban:rural (6:2 schools), government:non-government (6:2 schools) and primary:secondary (4:4 schools) were retained. Records of school-card register percentages (SCR; the percentage of children in a school receiving means-tested government support to attend school) for the participating South Australian schools suggest there had been no substantial shift for each school from the earliest (1997) to the most recent (2004) records. Between 1997 and 2004, the coefficient of variation of SCR was 0.2–0.6% for the eight schools taking part, indicating that the socioeconomic characteristics of the schools’ feeder neighbourhoods have been relatively stable in recent times. Data collection in both surveys occurred during the winter months, minimising the impact of seasonal variation on comparisons between surveys.

The questionnaire

The full ASHFS questionnaire has been published previously.9 Items relevant to the current study are: school and club sport (up to a maximum of six); and perceived physical activity levels of parents.Table 1 displays the survey items analysed in the study. In both surveys, trained data collectors read instructions to groups of four to six respondents and supervised completion of the questionnaire.

Table 1.  Items from the 1985 AHFS Survey relevant to this study. Thumbnail image of

Organised sport participation

The number of organised sports played in the previous 12 months was summed to form the variable sport number. While test-retest reliability of this item was not assessed in 1985, a sub-sample of 68 children in 2004 responded to this question twice on one day, separated by the usual lunch break. The test-retest reliability was high for the number of sports reported (ICC=0.85). Evidence for construct validity of the item has recently been reported, based on the 1985 dataset.7 Time to complete the 1.6 km run/walk field test of aerobic performance was inversely related to the number of sports played in the previous 12 months (p<0.001 for boys and girls).

Parent physical activity

All responses of ‘don't know’ to the questions on parent physical activity were removed from analyses. Mother's (Mother PA) and father's (Father PA) physical activity were dichotomised as: ‘active’ if ‘yes’; and ‘inactive’ if ‘no’ (see Table 1). A reliability study of responses related to parent activity has been conducted among Australian children in this age.10 Kappa coefficients (and percent agreement) were 0.57 (64%) and 0.59 (68%) for Father physical activity and Mother physical activity, respectively.

Data analysis

Exploratory ANCOVA models compared respondents with active and inactive parents on Sport number, separately for Mother PA and Father PA, including the interaction term for survey year and adjusting for age. Because several interaction terms were significant, comparisons were made separately in each survey year. To explore dose-response relationships, Mother PA and Father PA categories were combined to form parent PA, as follows: two parents active; only one active parent; neither parent active. ANCOVA models compared these three categories on Sport number, separately by sex and survey year. Statistical significance was inferred at p<0.05 for all comparisons. All analyses were conducted using STATA version 9.

Approval for the 1985 survey was granted through each state's Director General of Education, while the 2004 survey was approved by the South Australian Department of Education and Children's Services and the University of South Australia Human Research Ethics Committee.


Comparisons of Sport number, Mother physical activity and Father physical activity between survey years are presented in Table 2. There were no differences in Sport number between 1985 and 2004 samples, among both boys and girls. Between surveys there was a significant increase in the proportion of boys and girls who perceived that their mothers and fathers were active (Table 2).

Table 2.  Descriptive statistics and comparisons between each survey year. Data are presented as means (SD) for the variables Age and Sport number, and percent for the variables Mother physical activity and Father physical activity.
  1. Notes:

  2. PA = physical activity

  3. Sport number = number of organised sports in previous 12 months

  4. ns not significant for sex-specific comparisons between surveys

  5. a)  p<0.0001

Sport number2.272.10ns1.821.99ns
Mother PA
% yes35.552.8a40.257.3a
% no50.225.1a50.322.7a
% don't know14.222.1a9.520.0a
Father PA
% yes40.749.8a32.450.3a
% no44.521.9a50.918.1a
% don't know14.828.3a16.831.5a

Among boys in 1985, ANCOVA models indicated that there were no differences in Sport number between those with active and inactive parents (Table 3). Among girls in 1985, there were no differences in Sport number between those with active and inactive mothers, while those with active fathers played more sport (Table 3). Among the 2004 sample, boys and girls with either active fathers or active mothers reported higher Sport number (Table 3).

Table 3.  Comparisons of sport participation among children with active and inactive mothers and fathers in each survey year. Data are presented as means (SD) for the variable Sport number.
 Mother physical activityFather physical activity
  1. Notes:

  2. PA = physical activity        Sport = number of organised sports in previous 12 months

  3. Significance for comparisons between those with active and inactive mothers and fathers: ns not significant; a) p0.05; b) p0.01; c) p0.0001

 1985 2004 1985 2004 
 active motherinactive motheractive motherinactive motheractive fatherinactive fatheractive fatherinactive father
Sport2.37 (1.65)2.21ns (1.29)2.35 (1.52)1.88a (1.46)2.38 (1.50)2.31ns (1.46)2.41 (1.54)1.98a (1.46)
 active motherinactive motheractive motherinactive motheractive fatherinactive fatheractive fatherinactive father
Sport1.76 (1.34)1.71ns (1.52)2.39 (1.69)1.46c (1.18)2.16 (1.76)1.49a (1.26)2.30 (1.65)1.57b (1.47)

When Mother physical activity and Father physical activity were combined, and transformed into a trichotomous independent variable for Parent PA (‘both active’, ‘one active’, ‘neither active’), there were no differences in Sport number in the 1985 sample (Table 4). In 2004, boys and girls with both parents active reported higher Sport number than those with one or neither parent active (Table 4).

Table 4.  Comparison of Sport number between children with neither, one or both parents active, separately by sex and survey year. Data are presented as means (SD).
  1. Notes:

  2. a) Significantly different to ‘both active’

  3. b) Significantly different to ‘one active’

  4. c) ignificantly different to ‘neither active’

males2.,c2.2 a1.7a


This study has identified more consistent associations among children's perceptions of parent physical activity and child sport participation in 2004 compared with 1985, suggesting that role modelling as a socialising agent for organised sport is more important now than twenty years ago. In 2004, there were substantial differences in sport participation among boys and girls with and without active parents. These findings are meaningful from the public health perspective as engagement in organised sport is associated with many benefits, including its contribution to overall energy expenditure. Recent studies, one using self-reported physical activity11 and another using accelerometry as an objective measure,12 have reported that organised sport contributes ∼20% to total daily physical activity. Engagement in organised sport has been associated with higher physical fitness13 and lower levels of body fat14 among young people.

Reviews of the literature have highlighted confusion around the familial aggregation of physical activity.3,4,15 A meta-analysis examined the methodological issues that might explain the disparate research findings, focusing on sampling procedures, sample age and method of measuring parental physical activity.16 Effect sizes for associations among parent and child physical activity were: larger if convenience sampling methods were used compared with ‘principled’ (random or clustered) sampling; lower if physical activity was self-reported compared with objectively measured; larger for younger children (<9.75 years) and older adolescents (>12.75 years), compared with early adolescents (9.75 – 12.75 years); and larger if parents reported their own physical activity levels.16 A strength of the current study is that these confounding influences were minimised by retaining the same demographic characteristics and methodologies in the two surveys. This encourages confidence that the observations represent a true secular trend in parental role modelling as a socialising agent for sport participation.

There is evidence that parents play a more significant ‘gate-keeper’ role for organised sport participation compared with the less structured and independent activities in which children engage during their leisure time.17 The associations of child-perceived parent physical activity patterns and child sport in the 2004 survey can be attributed to several plausible mechanisms, both direct and indirect. Children may be influenced through direct observation of their parents’ active lifestyles, and consequently adopt these values and behaviours.7 It is feasible that a shared genetic disposition to physical activity exists as heritability coefficients of 0.35–0.83 for sports participation have been reported.18 It is also likely that instrumental (e.g. provision of transport and fees for participation) and affective (e.g. encouragement) support mechanisms co-exist with actual physical activity behaviours in families,19 such that active parents are more likely to offer these forms of support to their children.7 Notably, recent literature reviews indicate that the association of these support mechanisms with children's physical activity is stronger than the association of parent and child physical activity,4 raising the possibility that role modelling may be acting as a statistical surrogate of parental support mechanisms. A recent study has identified that the strength of parent-child bonding mediates the association of parent and child physical activity, such that children most strongly bonded to a parent are more likely to adopt that parent's physical activity behaviour.20 Unfortunately, the potential mediators of the associations of parent and child physical activity were not measured in the 1985 ASHFS, and therefore reasons for the changing strength of these associations could not be explored in this study.

The results of this study did not support the specific associations among mothers’ and daughters’ physical activity patterns previously reported in the literature. Earlier studies have identified associations among girls’ physical activity levels and those of their mothers, but not fathers, while fathers’ physical activity predicted the physical activity of both sons and daughters.4 In the more recent survey from the current study, boys’ and girls’ sport participation was associated with both fathers’ and mothers’ physical activity.

In accord with previous research, this study provides evidence of an ‘additive’ effect of parents’ physical activity on children's sport participation. Wagner and colleagues21 reported that French children were more likely to engage in structured physical activity outside of school when both compared with neither parents played sport. In a US study of 4–7 year olds where physical activity of children and adults was measured by accelerometry,22 those with active mothers were twice as likely to be active than children of inactive mothers, while those with active fathers were 3.5 more likely to be active. When both parents were active, children were 5.8 times as likely to be active as children of two inactive parents. These observations and those of the current study are consistent with various causal mechanisms, including more persistent and intensive role modelling and gestures of encouragement, shared activities among family members and a stronger ‘dose’ of genetic predisposition to physical activity.22

The study provides some evidence that sport participation among Australian children has been somewhat stable over recent decades. This finding is in accord with other studies that have reported declines in some specific physical activity contexts, such as Physical Education participation and active transport between home and school,23 but not ‘across the board’.8

There are several limitations to this study that should be acknowledged. The independent variable, parent activity, was reported by children, and the test-retest reliability of the items used is low to moderate.10 As highlighted by Pugliese and Tinsley,16 the source of parent activity data (parent report, child report, or objective measurement) seems to be important, in that associations of parent and child physical activity are more likely to be observed if parent physical activity is self-reported or objectively assessed. However, a Norwegian study that related parent self-reports to their 13-year-old children's estimates of parent physical activity demonstrated reasonable agreement (r=0.41–0.54).24 In any case, if parent-child bonding predicts whether a child will adopt the behaviours of a parent,20 the child's perceptions of those parent behaviours assumes importance.

Organised sport participation was an outcome variable in this study, represented by the number of sports played. No data were collected on frequency and duration of training sessions and games, and it follows that the number of sports reported may not truly reflect participation for some children. For instance, participation would be seriously underestimated for individuals who participate in one sport only, for which they train or play on most days of the week. It is feasible that the definition of sport has changed between surveys. Children in the latter survey would be surrounded by a broader ‘menu’ of opportunities, from the traditional, competition-based sports that predominated in 1985, to the proliferating modified versions of traditional sports that have been specifically designed by sporting organisations to affordably increase participation among children. As these activities are structured and supervised by adults and played in teams, children would be likely to list these in their response to the question on organised sport. The definition of ‘regular’ physical activity by parents (exercise two or more times a week) is based on outmoded recommendations for adults. Current guidelines are based on 30 minutes of accumulated physical activity on most days of the week,25 in contexts that are not confined to episodes of structured exercise. Accordingly, children in the latter survey may have underestimated their parents’ physical activity if only considering exercise as the criterion.

While the sampling design closely matched the area-level socioeconomic status of the samples in 1985 and 2004 by visiting the same schools, it is possible that relative affordability of leisure time activities such as organised sport has changed between surveys. According to data from the Australian Bureau of Statistics,26 between 1984 and 2003/04 price inflation averaged across all consumer goods and services purchased by Australian households increased at a much lower rate than both household expenditure and household income. This suggests that households have experienced improvements in aspects of material wellbeing in recent decades.


Of the approximately 30 published studies on parental influences on youth physical activity, the strength and direction of associations vary according to study design,16 leaving a confused picture. The current study used consistent methodologies at two time points to identify that family-centred interventions are as important now as ever before. Interpreted more broadly, this study suggests that interventions to promote physical activity in young people need to be founded on the most recent available evidence.