Data
The database for our analyses is the ‘Sozialer Survey Österreich’ (Austrian Social Survey) 2018, which consists of a random sample of 1200 Austrians representative for the national population aged 18 and above. The survey was conducted with computer-assisted personal interviews (CAPI) during the second quarter of 2018. The response rate reached 51%. The dataset, as well as the method report with an elaborate description of the sample, can be found at AUSSDA—The Austrian Social Survey Data Archive (Hadler et al. 2019). To use the best representation of the Austrian population, we use a post-stratification plus design-weight for descriptive statistics.
Variables
In the following, we will describe the relevant variables for this study. Besides the original scale, we rescaled every single item from 0 to 100 for descriptive results such that higher scores represent higher amounts of social capital and well-being. We will present Cronbach’s α estimations for reflexive scales.
Structural (informal) social capital
-
1.
Contacts with high- and low educated workers
To measure contacts with various occupations as part of informal social capital, participants were asked if they … “know a woman or a man who is …? a.) a bus/lorry driver, b.) a senior executive of a large company, c.) a home or office cleaner, d.) a hairdresser/barber, e.) a human resource manager/personnel manager, f.) a lawyer, g.) a car mechanic, h.) a nurse, i.) a police officer, j.) a school teacher”; with the possible answers “Family or relative”, “Close friend”, “Someone else I know”, and “No one”.
Since jobs in this list differ in terms of required education, we distinguished two kinds of professional contacts, separating those jobs which require tertiary education from those jobs that do not. While having contacts whose occupations require less education include knowing a (wo)man who is a.) a bus/lorry driver, c.) a home or office cleaner, d.) a hairdresser/barber, g.) a car mechanic or i.) a police officer, having contacts whose occupations require higher education includes knowing a (wo)man who is b.) a senior executive of a large company, e.) a human resource manager/personnel manager, f.) a lawyer, h.) a nurse, or j.) a school teacher. We summed all answers wherein “1” indicates that participants know a woman or man in the respective profession.
- 2.
Contact frequency with family and friends
Several items asked, “Think about the person (…) you have contact with most frequently: How often do you have contact with this person, either face-to-face, by phone, internet or any other communication device?”. Participants were asked this question about “a parent”, “a brother or sister”, “an adult child”, “another family member besides their spouse or partner, parents, siblings or adult children” and “a close friend”.
Due to the non-linear scale, we first computed dummy variables that indicate if someone has contact with the respective friend or family member on a “Daily” or “Several times a week” basis, or if the respective friend or family member lives in the same household. All remaining answers (“Once a week” or fewer, as well as “I don’t have … [a sibling] etc.”, “[my parents] etc. are not alive anymore”) are transformed into “0”. The proportion of respondents who have contact with a given person at least several times a week ranges from 27% for “other relatives” to 40% for “adult child”. The answers for the four questions on contact with relatives were combined into an index for family contact, ranging from “0” (indicating no frequent contact with family members) to “4” (indicating frequent contact with all mentioned family members). Besides this indicator for family contact, the dummy variable “contact with a close friend” remains a dummy variable as explained above. Since variable calculations are arbitrary acts, we created alternative variables for “family contact” and “contact with a close friend” which additionally included contact “once a week”. We present findings with those alternative variables in the result section as well.
- 3.
Social participation
The survey includes one item asking, “How often do you go out to eat or drink with three or more friends or acquaintances who are not family members?” on a scale from “1” (daily) to “8” (never) to measure “social participation”. We transformed this item such that “8” indicates daily while “1” indicates never.
Structural (formal) social capital
To measure formal social capital, the survey asks … “In the past 12 months, how often, if at all, have you taken part in activities …? a.) of groups or associations for leisure, sports or culture? b.) of political parties, political groups or political associations? c.) of charitable or religious organizations that do voluntary work?”. We form one index consisting of those three group activities in which “1” indicates group participation of “once in the past year” or more often, resulting in an index ranging from “0” to “3”.
Cognitive social capital
-
1.
Perceived social support
To measure perceived social support, the Austrian social survey 2018 contains two sets of questions.
One question asks regarding personal problems/situations, namely “Who would you turn to first to … a.) help you with a household or a garden job that you can’t do yourself? b.) help you around your home if you were sick and had to stay in bed for a few days? c.) be there for you if you felt a bit down or depressed and wanted to talk about it? d.) give you advice about family problems? e.) enjoy a pleasant social occasion with?” Participants can answer these questions with “a close family member”, “a more distant family member”, “a close friend”, “a neighbor”, “someone I work with”, “someone else”, and “no one”. The second question reads “Who or where would you turn to first to … a.) help you if you needed to borrow a large sum of money? b.) help you if you needed to find a job? c.) help you with administrative problems or official paperwork? d.) help you if you needed to find a place to live? e.) look after you if you were seriously ill?”. Possible answers are “family members or close friends”, “other persons”, “private companies”, “public services”, “non-profit or religious organizations”, “other organizations”, “no person or organization”. From these ten items, we derived an index which ranges from 0 to 10 points (for each item 1 point was assigned if the respondent indicated a person or institution s/he would turn to). Thus, as long as the respective respondent does not answer “no one”, we count 1 point.Footnote 2
- 2.
Social and Institutional trust
The survey includes an item list to measure institutional trust asking, “How much confidence do you have in … a.) Parliament, b.) Business and industry, c.) Churches and religious organizations, d.) Courts and the legal system, e.) Schools and the educational system”. The scale ranges from “1” (complete confidence) to “5” (no confidence at all). From these items we computed a mean index (Cronbach’s α = 0.73).
The last measured aspect of social capital, social trust, was measured with a single item, namely “Generally speaking, would you say that people can be trusted or that you can’t be too careful in dealing with people?”, ranging from “1” (“People can almost always be trusted”) to “5” (“You almost always can’t be too careful in dealing with people”). For both aspects of trust, we rescaled all answers such that higher values represent higher trust.
Outcome variable
The survey includes three items to measure subjective well-being, namely life satisfaction (1 “completely satisfied” to 7 “completely unsatisfied”), happiness (1 “very happy” to 4 “not happy at all” and momentary well-being (“how do you feel at the moment?”; 1 “very good” to 5 “very bad”). Again, we rescaled every item such that higher values represent higher subjective well-being. We additionally calculated a mean index out of these three single items (Cronbach’s α = 0.82).
Urbanization variable
At the end of the interview, interviewers noted whether the participant lives …. a.) in a big city (over 100,000 residents), b.) on the edge or in a suburb of a big city, c.) in a bigger city (40,000 to 100,000 residents), d.) in a small or medium town (5000 to under 40,000 residents), e.) in a village or f.) in a single house or farm in the countryside.
Due to low case numbers and in accordance with Sørensen (2016), we recoded this question into three categories, namely “urban” (living in a big city or on the outskirts or in a suburb of a big city), “suburban” (living in a town/city with 5000 to 100,000 residents) and “rural” (living in a village or in a single house or farm in the country). Preliminary exploratory analyses showed that results for suburban areas typically lie between the results for urban and rural areas. Therefore, we will only present the results for urban and rural areas.
Control variables
In accordance with previous research in this field (i.e. Helliwell and Putnam 2004; Portela et al. 2013), we control for sex, age, education, marriage, personal income, place of residence, unemployment, and church attendance. We use dummy variables to measure sex (1 = female), education (primary education or below, secondary education [“Matura”] and tertiary education), marriage (1 = married), place of residence (urban, suburban, rural) and unemployment (1 = unemployed). We also use dummy variables to measure age to consider non-linear effects (categories: ages 18–29, ages 30–44, ages 45–59, ages 60–74 and ages 75–95). Personal income is measured in quintiles (1–5) while church attendance is measured on a scale from “1” (Never) to “8” (several times a week).
Analytic strategy
First, we present descriptive results for social capital variables as well as for subjective well-being variables for the entire sample as well as for urban and rural areas. By using variance analysis, we analyze how urban and rural areas differ in terms of social capital and subjective well-being. For the multivariate analysis, we first used multilevel modelling which allows us to control for potential interviewer effects on the second level since a total of 75 interviewers conducted this survey (see Beullens and Loosveldt 2016). The analysis showed that 13.4% of the variation of subjective well-being is due to interviewer effects. However, a simple multiple regression analysis shows similar effects of social capital variables on subjective well-being and allows us to present standardized β coefficients (see Hox et al. 2017). Thus, to give a clear overview of the multivariate results, we decided to report results from multiple regression analyses while results from the multilevel models are depicted in Table 5. Our preliminary analysis indicates a non-linear effect of “family contact” on subjective well-being. We thus included these predictors as dummy variables into our models.