Study sample and data collection procedures
The Eurobarometer is a multi-national survey of public opinion and social trends in the European Union conducted since the 1970s on behalf of the European Commission (http://www.gesis.org/eurobarometer). In this study, we accessed data from Eurobarometer 58.2, which was the most recent survey that collected concurrent data on physical activity and happiness (October 28 - December 8, 2002) [24]. The data is publicly available and its use for non-commercial research is authorized.
Eurobarometer 58.2 was conducted among 15,334 respondents from 15 European Union countries: Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal, Spain, Sweden, United Kingdom. All European Union citizens residing in those countries (nationals and non-nationals) who were at least 15 years old were eligible to participate. Sample sizes within countries ranged from n = 602 (Luxembourg) to n = 2,042 (Germany) and the overall response rate was 54.6%.
Participants in the Eurobarometer 58.2 survey were sampled via a multi-stage design. First, primary sampling units (PSU) were selected from each European statistical administrative region in participating countries, with the sampling probability proportional to population size and stratified by urbanisation. Second, addresses were randomly selected within each PSU, and then one respondent was randomly selected per household. All necessary informed consent was obtained for each respondent and they were interviewed face-to-face in their homes in their national language. The data were collected by the European Opinion Research Group, which is a consortium of marketing and public opinion research agencies. Further details of Eurobarometer methods are available online (http://ec.europa.eu/public_opinion/description_en.htm).
Measures
The Eurobarometer 58.2 survey categories for socio-demographic data about the participants were used for: sex (male, female); age (15–24, 25–39, 40–54, 55 years and older); age when finished full-time education (≤15 years, 16–19 years, ≥20 years, still studying); self-reported general state of health (very good, good, fair, bad/very bad); employment status (self-employed, employed, not working). We recoded the relationship status responses into a dichotomized variable based on current cohabitation with a partner (currently cohabitating: “married”, “remarried”, “unmarried, currently living with partner”; not currently cohabitating: “unmarried, having never lived with a partner”, “unmarried, having previously lived with a partner but on my own”, “divorced”, “separated” and “widowed”).
The measurement of happiness was adopted from a single question from the SF-36 survey: “In the past month, have you felt happy?” [25]. There were six response options that we recoded into a dichotomized variable to differentiate between those who provided a positive vs. negative response (YES: “all the time”, “very often”, “often”; NO: “rarely”, “very rarely”, “never”).
Physical activity was assessed by adapting questions from the short form of the International Physical Activity Questionnaire (IPAQ-short) into a multiple-choice format as presented in Table 1 [26]. The IPAQ-short had test-retest reliability (Spearman’s rho = 0.76) and criterion validity (Spearman’s rho = 0.30) that was comparable to other self-report measures [27]. Participants were asked to report the number of days in the past week and the total time per day (hours/minutes) of walking, moderate-intensity and vigorous-intensity physical activity in bouts of at least 10 minutes. We calculated physical activity volume as the sum of the three activities with vigorous physical activity weighted by two to account for higher energy expenditure. Participants were then classified based on the global physical activity recommendations as “inactive” (0–9 minutes of physical activity), “insufficiently active” (10–149 min), “sufficiently active” (150–299 min), and “very active” (300+ min). We also calculated the independent amounts of walking, moderate-intensity and vigorous-intensity physical activity for subsequent analysis of intensity specific associations with happiness.
Table 1
Adaptation of the of the IPAQ-short questions used in Eurobarmoeter 58.2
The Eurobarometer 58.2 survey included four additional questions to assess physical activity in various domains (i.e. vocation, transport, domestic and leisure). The participants were asked: “In the last seven days, how much physical activity did you get: 1) at work; 2) when moving from place to place; 3) doing work in and around your house (including housework, gardening, general maintenance, or caring for your family); 4) for recreation, sport and leisure-time activities?” There were three response options: “a lot”, “some” or “little or none”. These data were used to assess domain specificity of the association between physical activity and happiness.
Statistical analysis and reporting
Data analyses were performed using IBM SPSS19.0. Participants who did not have complete physical activity data (n = 3,607) and an additional 90 individuals who reported they were “severely restricted to physical activity” were excluded from the final analytical sample. Reported happiness was presented according to socio-demographic characteristics and bivariate associations were examined using a Pearson’s chi-square test. We ran logistic regressions to examine the crude and adjusted associations of happiness with: 1) physical activity volume; 2) physical activity intensity; 3) physical activity domain. The model for physical activity volume was adjusted for sex, age group, country, general state of health, relationship status, employment status and age when finished full-time education. The model for physical activity intensity included time spent engaging in walking, moderate-intensity and vigorous-intensity physical activity as independent variables, adjusted for all of the covariates described above. Since previous studies have identified sex differences for the association between physical activity intensity and mental health, we tested these interactions and conducted stratified analysis for any significant findings [21]. The associations between domains of physical activity and happiness were tested in separate models adjusted for physical activity volume in addition to all covariates. The model for vocational physical activity only included those currently employed (n = 5,816). Statistical significance was assessed at the levels of p < 0.05, p < 0.01 and p < 0.001.