Scientific Report for the Level of Living Survey 2010 (LNU 2010)

Jan O. Jonsson, Institutet för social forskning - SOFI, Stockholms universitet

2010-2011

The Level of Living Survey is a comprehensive survey of a nationally representative sample of individuals living in Sweden, mapping their resources, conditions and activities in different arenas - such as working life, family life, and the civil society. Individual life trajectories can be followed prospectively over time due to the panel structure, and retrospectively via biographic information on family, education, and work. This combined longitudinal design is crucial not only for descriptive analyses but, more importantly, for the possibility to uncover the causal mechanisms in social life.

The first LNU survey was carried out already in 1968, with new waves in 1974, 1981, 1991, and 2000, containing much the same questions. LNU is a repeated cross-sectional study, each time representative of the adult Swedish population, 18-75 years of age (15-75 years in 1968, 1974, and 1981), making it possible to draw conclusions about change over time.

LNU also has a multidimensional design that makes for a comprehensive content: the level of living of an individual is defined by her/his command over resources in the form of money, possessions, knowledge, mental and physical energy, social relations, security etc. through which s/he can control and consciously direct her/his living conditions (Johansson 1970:25; cf. Sen 1999: 70). Respondents are accordingly asked questions about their conditions within a number of 'welfare components' (i.e. childhood conditions, housing, family, social network, education, employment, health, household work, economic resources, security, leisure activities, political participation, and general assessments) via face-to-face interviews. Moreover, information has been added from official registers, primarily on annual incomes; LNU is a prime example of how an integrated data base improves on each kind of data source.

The longitudinal design means that LNU to a large extent contains the same individuals from wave to wave (though, to ensure representativeness of each cross-sectional survey, older cohorts have been replaced by younger, and immigrants have been added). Thus, individuals are followed prospectively which enables researchers to register important changes in e.g. individuals' economic conditions, health or working conditions.
Another important feature of the LNU survey is that it contains retrospective biographical information on important life events (viz. economic activity, educational careers and family histories), which together with yearly register data on incomes (between waves) make it possible to study not only that changes have occurred in individuals' lives, but when (to the level of months). This chronological ordering of events allows the study of life-course processes and greatly facilitates causal interpretation of the relation between, say, job or family change and income change (though a temporal order is still not necessarily a causal one). The biographical data were collected for the first time in 1991 and cover cohorts born 1925 and onwards, meaning that we can analyze family and work histories for successive cohorts from WWII and onwards (Jonsson and Mills 2001).

Finally, during the years, via new extensions of the data, LNU has added not only temporal but also essential contextual information on individuals' family and work. In 1991 and 2000 we collected data from the respondents' work places (The Establishment Survey, see le Grand, Szulkin, and Tåhlin 1996). Hence, to the employees' information about their working conditions etc., information was added on contextual characteristics of the firm, such as its competitiveness, personnel policies, organizational structure, and work force composition. In LNU 2000, the possibility of analyzing the family as a context for individuals' scope of action was further improved. First, we collected data (via an audio-questionnaire) directly from children, 10-18 years of age, who lived in the respondents' households: we focus also here on their resources, in the level-of-living framework, with somewhat more attention paid to social relations and psychological well-being (Child-LNU, see Jonsson and Östberg 2001, 2010). This extension has further increased our understanding of processes of intergenerational transmission of advantages and disadvantages. Second, we enlarged the section on household division of labour and family relations in the main questionnaire, including a number of questions on each child in the household and also on those who moved out. Third, a (postal) questionnaire was given to the partners of our respondents, including a fraction of the questions asked to the respondents (Spouse-LNU). This has meant richer information on the household, the possibility of studying how household members' living conditions are connected to each other, and it aids in analyses of children by giving information on both parents.

All in all, LNU is a mature and wide-ranging research infrastructure, the usefulness of which ranges from its historically relevant description of social change to its possibilities to offer data for advanced modeling of micro-level dynamics. LNU 1968, 1974, 1981, 1991, and 2000 are all easily accessible through SND (The Swedish National Data Service), and extensively used by researchers in a number of disciplines, and coming from all Swedish universities, and from many countries, including the major universities in the USA.

Organization

The Level of Living Survey 2010 has been managed by a steering committee (board), formed by the Principal Investigator, professor Jan O. Jonsson, professor Michael Tåhlin, professor emeritus Robert Erikson, docent Michael Gähler, docent Marie Evertsson, and docent Magnus Nermo, all at SOFI. An executive committee, with practical responsibility for project management, has been formed by the three latter, with Michael Gähler as director and Marie Evertsson and Magnus Nermo as deputy directors. During 2009-2011 two research assistants were attached to the project. In parts of the process, all researchers, doctoral candidates, and research assistants, approximately 20 people, participated.

A reference group was formed early on in the project to aid the project group. It consists of highly qualified researchers from different disciplines and research interests, and from different departments and institutes, but all with an extensive experience of collecting and/or empirically using survey data in general, and LNU in particular. Members are: associate professor Carin Lennartsson and professor Mats Thorslund, Ageing Research Center (ARC), Karolinska Institutet and Stockholm University; professor Johan Fritzell, professor Olle Lundberg, associate professor Bitte Modin, and associate professor Viveca Östberg, Centre for Health Equity Studies (CHESS) (Modin and Östberg were partly responsible for Child-LNU 2010); professor Gunnar Andersson, associate professor Ann-Zofie Duvander, and professor Elizabeth Thomson, Stockholm University Demography Unit (SUDA) at the Department of Sociology; professor Carl le Grand, professor Ryszard Szulkin, and professor Eskil Wadensjö, Stockholm University Linnaeus Center for Integration Studies (SULCIS) (le Grand and Szulkin are affiliated with the Department of Sociology, Stockholm University, and Wadensjö is affiliated with SOFI); professor Anders Björklund, Labor Economics, SOFI, and professor Sten-Åke Stenberg, Social Policy, SOFI.

Preparation

The fieldwork was preceded by more than one year of careful preparation. As noted, in October 2008 the LNU 2010 project was awarded a planning grant from VR. These means were used mainly for three purposes: First, in cooperation with the Stockholm University Linnaeus Center for Integration Studies (SULCIS, www.su.se/sulcis/), we prepared for the extension of the sample frame by including an additional sample of foreign-born and their children (7 350 individuals). Due to its size this project, entitled LNU-UFB (Levnadsnivåundersökningen för utrikes födda och deras barn), has been led, organized, and funded separately from LNU, but the cooperation between the projects has been close. Discussions led to the conclusion that the LNU-UFB sample should mainly be asked the same questions as the main LNU sample. For this purpose, some questions in LNU were harmonized with LNU-UFB. In LNU-UFB some questions were also replaced by special questions on attachment to and time in the country of origin, and the move to Sweden. Fieldworks for LNU-UFB and LNU have been essentially parallel, although with LNU2010 somewhat ahead in time.

The LNU sample was further extended by the inclusion of the Previous Child-LNU respondents as a special stratum in the main sample. These children, who were 10-18 in the LNU 2000 survey, were consequently 20-28 years old in 2010. Adding this group to the main sample provides information on adolescents' transition into adulthood and on how childhood conditions relate to young adults' lives. Thanks to the availability of rich data about the parents and the childhood households, this addition facilitates studies of a crucial formative period in people's lives. The combination of interviews with (a) parents (the main LNU respondents in 2000), (b) their children when they were children (in 2000), and (c) the children as adults (in 2010) forms a detailed source of information on many aspects of intergenerational resource transmission, something that has not been present before in Sweden.

Second, much time was devoted to carefully review other renowned international surveys that are comparable to LNU (i.e. the British Household Panel Survey (BHPS), Generation and Gender Survey (GGS), German Socio-Economic Panel (GSOEP), The Household, Income and Labour Dynamics in Australia Survey (HILDA), Panel Study of Income Dynamics (PSID), Survey of Health, Ageing and Retirement in Europe (SHARE), Swiss Household Panel (SHP), and Canadian Survey of Labour and Income Dynamics (SLID)). The general conclusion from this review was that LNU is already highly comparable to international surveys but to increase comparability even further some questions were added to the LNU questionnaire. We were reluctant to change the wording of already existing questions, however, as our priority is to guarantee comparability between LNU waves.

Third, large efforts were made to evaluate a possible method change from face-to-face to telephone interviews, the change being prompted by an urge to limit fieldwork costs. Based on a literature review and own empirical analyses of Undersökningen av levnadsförhållanden (ULF) 2006 (Statistics Sweden), which was based on a mixed-methods approach, we decided, however, to stay with face-to-face interviews. Our conclusion was that a method change would jeopardize response rates, in particular for some sub-categories (immigrants and low-income earners) and for Child-LNU. Moreover, response patterns seem to differ slightly by method. Nevertheless, one method change was carried out as paper-aided personal interviews (PAPI), used in the first five LNU waves, were replaced by computer-aided personal interviews (CAPI). This change has resulted in several advantages: lowered costs as special personnel for data registration is no longer required, shortened interview time, increased data quality as interviewers are automatically guided to the correct questions, and, finally, the possibility to adopt "dependent interviewing", i.e. that respondents are guided to the correct questions according to part-taking and answers in previous waves of LNU. The only drawback of this method change was that extensive work was required to build a digital questionnaire. Due to the volume of the questionnaire, and the complex data structure, where many questions should just be answered by sub-strata, several versions of the questionnaire had to be programmed, tested and scrutinized. Considerable personnel resources were spent on this work during the fall of 2009 and spring of 2010.

Parallel to these tasks, the main assignment before the start of the fieldwork was to decide which questions were to be included in the three questionnaires, i.e. Main-LNU, Spouse-LNU, and Child-LNU. As noted, a guiding principle for LNU management is to give priority to maintaining comparability between waves and preserving uninterrupted time series. At the same time, however, LNU needs to be updated to represent living conditions in contemporary society, allowing for some renewal. The practical work was organized in such a manner that the entire LNU group was divided into small groups (4-6 individuals) to carefully review each part of the fifth wave questionnaires (i.e. LNU 2000) to form questionnaires for LNU 2010. One goal was to reduce the main questionnaire length in order to shorten the average interview time (this endeavour was successful as the average interview time decreased from 75 minutes in 2000 to 64 minutes in 2010). After months of work and continuous meetings, including the LNU 2010 reference group, a decision on final questionnaires was reached. For the Main-LNU questionnaire the result was not only a subtraction of questions but also an innovative renewal of some parts, for example: i) a biography on moving where respondents were asked to report any living outside Sweden and the time period for these stays (on a monthly basis), ii) a package of questions on capital and wealth (necessary because register data are no longer available due to the removal of taxation of capital), iii) the adding of a questionnaire on personality traits. The personality questionnaire contains 31 items, carefully selected on the basis of psychological theory, previous empirical research, and performance in other surveys, and was handed to the respondent to fill out after the ordinary main interview was completed.

Before new questions could enter the questionnaire they were carefully tested, and qualitatively reviewed, by Statistics Sweden in a cognitive test involving seven respondents (see the technical report). The Spouse-LNU questionnaire was concentrated on a sub-set of vital questions from the main questionnaire in order to shorten the questionnaire and maximize response willingness whereas a conservative approach, i.e. few changes, was adopted for the Child-LNU questionnaire.

During spring 2010, a close to final version of the main questionnaire was used for a pilot study with 50 respondents. The three purposes with this test were to estimate interview time, review new questions, and carry out a sharp test of the digital questionnaire.
Before the fieldwork could commence different survey documents were produced, such as cover letters and information leaflets for mailing out to respondents. Moreover, all (approximately) 120 interviewers at Statistics Sweden took part in survey-specific face-to-face lectures, held around the country by LNU 2010 management. Moreover, all interviewers were handed instruction-DVDs and instruction booklets, and interviewer time was set aside for self-study; all efforts with the purpose of attaining highest possible data quality.

Fieldwork

The fieldwork for LNU 2010 started in April 2010. The fieldwork was divided into three sub-waves, planned to take place during i) April-July 2010 with follow-up during August-September 2010, ii) August-October 2010 (November 2010-January 2011), and iii) November 2010-January 2011 (February-March 2011). During follow-ups interviewers took renewed contact with individuals that had previously not been able or willing to take part in the survey. Moreover, interviewers tried to reach respondents that had not previously been reached (see the technical report for details).

In order to secure data of highest possible quality, a number of measures were taken during the fieldwork, and they have permeated the entire data collection process. First, when interviewers had carried out their first interviews, these were controlled by staff at SOFI. Each and every one of the 120 interviewers was contacted via telephone and given feedback on their interviews. Any errors were corrected and interviewers were asked about their experiences of the survey, questionnaires, and interviews, and given a chance to give feedback to researchers.
Second, the LNU project management and research assistants have been in continuous contact with the project management at Statistics Sweden. Regular meetings were held, reports on fieldwork progress were delivered on a weekly basis, and SOFI personnel spent a great amount of time at Statistics Sweden in Örebro to monitor, guide, and assist in the data collection process.

Third, during the process Statistics Sweden has continuously sent part delivery data files. The first delivery contained approximately 100 interviews, and subsequent deliveries were sent at the end of each sub-wave. Two research assistants at SOFI spent large parts of 2011 to carefully scrutinize these data to find any errors.

Fourth, interviewers were encouraged to report any uncertainty or obscurity in answers via digital notes, recorded in the data file. All these notes (around 8 200 in total) were also checked by research assistants at SOFI. Most notes were trivial and did not cause any action. In a non-negligible number of cases, however, data were re-coded and/or interviewers were asked to get back to the respondent for complementary information. Statistics Sweden still needs to complete some missing data and the delivery of the final data file is scheduled for February 2012.

To conclude, then, extensive measures have been taken to guarantee data of the very best quality. Nevertheless, despite measures to increase the response rate, e.g. appealing letters to non-respondents and continuous demands on Statistics Sweden to put in more effort, the response rate after follow-ups, 61.5 percent, was clearly lower than expected. This seems to be part of a general trend as other surveys have also experienced strongly receding response rates during recent years. It is still an open question, however, whether this trend has hit Statistics Sweden harder than other survey agencies. In a last try to reach non-respondents we decided to produce a short questionnaire, only demanding 5-10 minutes of respondents' time. This questionnaire (answered via telephone or as a postal questionnaire) only contained the most central questions in LNU. Moreover, by taking part respondents also gave us their permission to collect register data. Although it prolonged the fieldwork period by approximately two months, the short questionnaire must be described as a success, with the total response rate increasing from 61.5 to 72.0 percent. In total, then, of 8 889 individuals in the sample, 6 259 individuals took part in Main-LNU (of which 915 responded to the short questionnaire). Moreover, 3 347 spouses and partners of main respondents, or 76.8 percent, filled in postal questionnaires for Spouse-LNU and 1 282 children (aged 10-18), or 72.0 percent, responded to questions via an audio questionnaire for Child-LNU.

Remaining tasks

Since fieldwork started, data has continuously been managed to guarantee highest possible quality, and to facilitate prompt data access for the research community. Still, five labor-intensive tasks remain: 1) to finalize data controls and the production of approximately 1 200 variables, 2) to code open-ended answers, in particular 40 000 occupations, into approximately 160 categorized variables and into social class variables, 3) to produce biography files, based on retrospective information on monthly level, for family events, occupation, and education, 4) to gather, control, and adjust register data to LNU data, and 5) to document data files and variables with frequency distributions and descriptive statistics in codebooks (see application for further information).

Finally, for obvious reasons there are no scientific results based on the project to be presented yet. Neither have there been any publications based on LNU 2010. As soon as data are ready to use for the research community, however, we are confident that LNU 2010 data will generate a large quantity of publications. In fact, a first volume, edited by Marie Evertsson and Charlotta Magnusson, with contributions of most LNU faculty, and based on data from all LNU waves, including LNU 2010, has been initiated.

 
References

• le Grand, C., R Szulkin, M Tåhlin (eds.) 1996. Sveriges arbetsplatser: organisation, personalutveckling, styrning (2nd ed.). Stockholm: SNS Förlag.
• Johansson, S. 1970. Om levnadsnivåundersökningen, Stockholm: Allmänna förlaget.
• Jonsson, J.O. and C. Mills. 2001. From Cradle to Grave. Life-course Change in Modern Sweden. London: Routledge.
• Jonsson, J. O. and V. Östberg. 2001. In collaboration with M. Evertsson and S. Brolin Låftman. Barns och ungdomars välfärd. SOU 2001:55. Stockholm: Fritzes.
• Jonsson, J.O. and V. Östberg. 2010. "Studying Young People's Level of Living: The Swedish Child-LNU", DOI 10.1007/s12187-009-9060-8, Child Indicators Research 3:47-64.
• Sen, A. (1999). Development as Freedom. New York: Alfred A. Knopf.