Physical inactivity has become a public health problem in many industrialized countries [1–3], with few people meeting the recommendations for physical activity . Moreover, population-based studies from the USA  and Canada [6, 7] as well as from European countries [3, 7–10] show that lower-educated people report lower levels of physical activity. The low-educated people therefore also miss the beneficial health effects of regular physical activity [4, 11]. A strong need exists to assess the determinants of leisure-time physical activity (LTPA) to promote physical activity in low-educated individuals.
Physical activity has been shown to decrease during the transition from youth to adulthood [12, 13]. Tracking of physical activity in adolescence and from youth to adulthood have varied from low to moderate. The tracking correlations are higher when the time points of the two measurements are close . Several longitudinal studies [14–17], however, suggest that physically active individuals tend to stay physically active from adolescence to adulthood. During adolescence, LTPA is more regular among those who participate in several different types of sports after school hours compared to those who participate in only one sport . This diverse adolescent participation in sports and exercise--for example running, cross-country skiing, and endurance sports in men, and running, track and field, and orienteering in women--also associates to higher levels of physical activity in adulthood [12, 15, 19].
Prospective research shows that not only LTPA  but also physical education (PE)  explain adulthood LTPA. A nationally representative prospective study in the USA  showed that high participation in school physical activities, including team and individual sports, academic clubs, and PE, was associated with adulthood LTPA. In a recent prospective study , the average minutes of PE did not predict adulthood total physical activity or fitness. It has been shown that also high grade in PE  explains adulthood LTPA. In addition, physically active adolescents might progress better in their educational career than their physically inactive counterparts [14, 22]. Sizable amount of evidence exists [23, 24] that enjoyment of exercise, self-efficacy and value of expected outcomes have positive effects on physical activity. However, there is, to our knowledge, no information how enjoyment, pleasantness and usefulness of PE in childhood at population level could motivate to participate in sports and exercise in adolescence as well as in LTPA in adulthood. One might hypothesize that those opinions on PE in childhood could directly affect to physical activity and exercise in youth and which could indirectly affect to adulthood LTPA.
The roots of adulthood unhealthy behaviours, as well as for physical inactivity, may lay in childhood socioeconomic conditions , and in childhood and adolescence health behaviours . Physically strenuous work during the lifespan may as well affect on willingness and ability to participate in LTPA . Some qualitative studies [28, 29] suggest that, among adults, determinants of physical activity might depend on the socioeconomic position. Recent studies [30, 31] showed that there might be socioeconomic variation in parental support and perceived outcomes among adolescents which determine their physical activity. No studies, to our knowledge, have examined how the socioeconomic variation in childhood and adolescence physical activity might affect on adulthood physical activity. Thus, our aim was to examine how retrospective information on competitive sports in youth, exercise in late adolescence, and opinions on PE in childhood determine adulthood LTPA in low- and high-educated groups. We assumed that there might be unobserved constructs, such as socioeconomic cultural team spirit, behind participation in competitive sports in youth and opinions on PE in childhood that could vary in low- and high educated groups.