The Baltie Countries Revisited:
Living Conditions and Comparative Challenges
The present report is a compilation of articles written by Nordic and Baltic scholars based on presentations from the Fafo seminar. The articles
examine various aspects of comparative living conditions research, and offer severai examples of comparative analyses based on the NORBALT and other complementary data sets. Moreover, the report sketches a strategy for further comparative studies.
The Ba/tie Countries Revisited is the latest in a series of publications deri
ving from the NORBALT project. This series consists of country reports from each of the countries involved in the study, as well as reports comparing the findings from the surveyed areas.
Supported by the Nordic Council of Ministers and the Joint Committee for Nordic Social Science Research Councils (NOS-S)
� � ��Fafo
... Fafo Institute for Applied Social Science
\ p.a. Box 2947 Tøyen,
N-0608 Oslo, Norway
Fafo-report 230 ISBN 82-7422-196-6
AAONEAASLANO, KNuo KNUOSEN, DAGMAR KUTSARANO ILZE TRAPENCIERE (EOS.)
THE BALTIC COUNTRIES REVISITED:
LIVING CONDITIONS AND COMPARATlVE CHALLENGES
THE NORBALT LIVING CONDITIONS PROJECT
Compilation of papers presented at the seminar
BALTIC-NORDIC LIVING CONDITIONS RESEARCH Oslo 5-7 December 1997
Supported by the Nordic Council of Ministers and the Joint Committee for Nordic Social Science Research Councils (NOS-S)
FAFO REPORT 230
© Forskningsstiftelsen Fafo 1997 ISBN 82-7422-196-6
Cover page: Premraj Sivasamy
Printed in Norway by: Falch Hurtigtrykk
CONTENTS
Preface ... 7
Knud Knudsen Introduction: Living Conditions and Comparative Research ........... 11
Stein Ringen Notes on the Definition and Measurement of Living Conditions ... 17
Definition ......................................................................... 17
Measurement ... 18
Vida Cesnuitytti Introduction to the NORBALT Data Sets and Guidelines for T heir Use . . . ...... .. .............. . . ... 21
Introduetion .............. . ............................................................. 21
The strueture of the data files ................................. ....................... 22
Potential teehnieal problems in data analysis ................... 27
Coneluding remarks ............. ............................ ...................... 33
AnuNarusk Living Conditions Surveys: Improving Tools for Measurement and the Search for Explanatory Devices .. . ............. 35
Living eonditions surveys eondueted in Estonia: dimensions and tools of measurement .......... ............................. 36
Subjeetive indieators of well-being ............................................. 39
Combining quantitative and qualitative methods of data eolleetion .... .42
Conelusion .............................................................................. 46
Mart Einasto Measuring Poverty and Deprivation in Countries in Transition: Behind the Estonian Indicators ... ; ... 49
Introduetion ........................................... . .................................... 49
The eoneept of poverty ........................................................ 50
Data-eolleetion methods ......................................................... 53
The operationalisation of poverty coneepts .................................. 53
Relations between different indieators ...... ............................ 60
Subjeetive and soeial assistanee-related poverty measures .................... 65
Conelusion ................................................................................. 68
Dagmar Kutsar
Multiple Welfare Losses and Risk of Sodal Exclusion in the
Baltie States during Societal Transition ... 79
From aetive involvement to the risks of soeial exclusion and alienation ... 80
Multiple losses of soeial welfare resourees ... ... 81
Three dimensions of welfare ... . ... 82
Soeial exelusion . . . . ... ... 84
The study ... 85
Deficits in «Loving» dimension of welfare: «isolation» ... ... 88
Deficits in «Being» dimension of welfare: «normalessness» ... ... 90
Deprivation, isolation and normlessness: Evaluation of welfare defieits ... ... ... 92
Aeeumulation of welfare defieits as risk of soeial exclusion ... 94
Instead of a conclusion ... 102
Aadne Aasland Ethnic Groups and Living Conditions: A Study of Unemployment in the Baltie Countries ... 105
Introduetion ... ... ... . ... .... 105
Comparing living conditions of ethnie groups in the Baltie states ... 106
Unemployment and welfare implieations ... . ... 109
Comparing unemployment rates ... ... . ... 111
Does eitizenship protect? ... ... . ... . ... ... 115
Interaction effeets ... . ... .... 116
Conclusions ... 118
Kjetil Du vold Comparing Attitudes: Political Trust among the Titular and Non-titular Groups of Latvia and Estonia ... 121
Introduetion ... ... 121
Trust in public authorities ... ... 123
Conclusions ... ... . . ... 127
Jorunn Wold Attitudes Towards Privatisation in Estonia, Latvia and Lithuania ... 129
Introduetion ... ... ... ... 129
Baekground ... . ... ... 129
Attitudes towards privatisation in the NORBALT survey ... 131
Faetors influeneing attitudes towards privatisation on an aggregate level ... ... 134
Soeial baekground indieators ... 140
Coneluding remarks ... 143
Jelena Helemae Winners of Transition ? Young Adults in the Baltie Countries according to Data of Longitudinal and NORBALT Surveys ... 147
Introduetion ... 147
Longitudinal studies on the life eourse of generations in the Baltie eountries ... 148
Paths of a Generation as a eomparative projeet ... 150
Winners of transition and the old elite in the eontext of ehanging living conditions ... 151
NORBALT and PG-ELS on edueation as a resouree of winners ... . . ... . ... ... 153
Monetary returns on edueation ... 154
Edueation and positional advantages ... 156
Soeial reproduetion and edueational attainment ... 159
Coneluding remarks ... 162
Jussi Simpura Comparing Incomparables? Living Conditions behind the Statisties in the Baltie Countries .... 165
Objeetive living conditions and their eolleetive and subjeetive interpretations ... 165
Teehniealities versus interpretative frameworks ... 166
Objeetive living eonditions, ehoiee and action in everyday life ... 169
Four examples of different routines and living conditions ... 170
Concluding remarks and proposals for eomparative settings ... . . 172
PREFACE
A research semmar entitled «Comparative Baltic-Nordic Living Conditions Research» was held at the Institute of Applied Social Science (Fafo) in Oslo, 5- 7 December 1 996. The idea of such a seminar had developed gradually, and the first concrete initiatives were made already in the winter of 1 995, in the midst of writing the country reports from the NORBALT living conditions project.
NORBALT, funded by the Norwegian ministries of Foreign Mfairs and Defence, comprises living conditions surveys conducted in the autumn of 1 994 in the three Baltic countries, Estonia, Latvia and Lithuania, as weU as in two regions ofRussia, St. Petersburg and Kaliningrad.
The NORBALT surveys were designed according to a standardised format, making it possible to compare results from all the survey areas. Many of the modules in the questionnaire were deliberately taken from living conditions surveys conducted in the Nordic countries, enabling also Baltic-Nordic comparisons.
Thus, the comparative perspective has been present from the early design phase of the project.
From the beginning, representatives of the authorities in the Baltic countries expressed a wish that NORBALT be us ed for comparative studies. It was thought that such comparisons could be an important step towards doser integration with neighbouring countries, not only in terms of research networks, but als o politically and symbolicaUy. In addition, it was thought that comparative studies would provide an interesting picture of regional developments and thereby contribute to a better understanding of important tendencies in the Baltic-Nordic region. The NORBALT data base serves as a unique baseline for this type of social reporting from the countries on the eastern Baltic rim, and it will undoubtedly be used as a benchmark for further developments in the Baltic-Nordic region.
Comparative living conditions research has been in focus for many years in the Nordic countries and elsewhere. In September 1990 a Nordic seminar «On living conditions and welfare statistics» was held in Helsingor, Sweden. The seminar dearly revealed that such research involves a number of methodological challenges.
While participants were then able to concentrate on countries that were relatively stable and quite similar in terms of level of welfare, the present seminar introduced a new dimension. Attention was drawn to the complications arising when the task is to compare countries undergoing rapid economic and social transition. Ihis point is especially crucial when com paring the Nordic and Baltic countries, which have developed in very different directions during the 20th century. The economic
and social characteristics are naturally quite diverse. How should specific differences be accounted for, given structural and cultural variations among these nations?
The challenge is not only to develop a data base which contains comparable variables, although this may be difficult enough. An important requirement of comparative research of this kind is to construet concepts and operationalise social phenomena as weU as living conditions outcomes so that such comparison becomes meaningful. How can one establish reliable and relevant indicators which make it possible to carry out comprehensive and meaningful analyses of living conditions across countries? Moreover, how can one ensure that cultural and institutional particularities of individual countries are interpreted in a sensible manner?
Against this background, Fafo and its cooperating partners in the Baltie states saw a clear need for a Baltic-Nordic seminar on comparative living conditions studies. Thanks to funding obtained from the Nordie Council of Ministers and the Joint Committee for Nordie Social Science Research Council (NOS-S), the plans for such a seminar materialised. Prominent researchers from the Baltic and the Nordie countries, and even participants from international organisations, gathered in Oslo.
The seminar had three main aims: first, to discuss the conceptual and methodological issues that need to be taken into consideration when carrying out comparative research of this kind; second, to present examples of comparative analyses based on the NORBALT and complementary data sets; and third, to sketch a strategy for further comparative studies.
This book, based on the papers presented at the seminar, reflect these three aims. Some of the chapters in the book deal with important conceptual and methodological issues concerning living conditions studies, with a special focus on their comparative potential. The concept of living conditions and the place of living conditions surveys in social reporting are diseussed. The usefulness of various social indicators is examined. Conceptual discussions focus on poverty and social exclusion, which now pose a threat to sustainable human development in the Baltic countries.
Other chapters deal with practical applications of comparative research based on the NORBALT data. The Baltie countries, although different historically and culturally, have shared a similar fate during the last fifty years. The comparisons made reflect their recent developments as restored independent states, on the way to market economies and civil societies. These articles, dealing with issues such as ethnicity, education and personal attitudes, reveal an obvious potential for many new in-depth analyses based on NORBALT.
The final article sums up the NORBALT experience and presents a potential framework for further comparative research. The analysis, which is based on different theoretical models, provides a valuable contribution to further research.
The contribudons are diverse, and many topics are covered. The aim of the book is to create a mosaie of different insights and traditions, which all can enrich the overall comparative perspective. Within this common framework, the authors
have had considerable freedom, both in the selection of topics and in their approach to the theme. The authors are responsible for the contents of each contribution and the views and opinions expressed.
It is our hope that (his book can attract a variety of readers. First of all, it could offer new insights for researchers and students in the field of living conditions studies as well as related areas. Also, we hope to find an audience among policy-makers, officiais, representatives of non-governmental organisations, journalists and others to whom social developments in the Baltic states are of immediate interest. Internationally, the book offers information about the resources and risks of social development in collntries in transition.
This book is a collaborative effort. The editorial board wOllld like to express their gratitude to all the seminar participants, without whom this book would never have become a reality. We are indebted to those who funded the NORBALT project as well as to those who carried out the surveys. The allthorities in all three Baltic countries have given great support during all stages of the project. Thanks also to Fafo's staff for providing guidance, facilities and an excellent research atmosphere. Glenn Martin edited the language and style of all the contributions, and deserves warm thanks. We are very grateful to Helen Hakami Sandal for several valuable contributions in producing the report and for communicating with the authors. Thanks also go to Premraj Sivasamy, who expediently transformed the various manuscripts into one readable form. Last, but not least, we very much appreciate the cooperation with all the contributors to this book. We sincerely hope that the seminar was the first step of a process towards new, improved and integrated social reporting in the Baltic-Nordic region.
A note regarding the lists of references: For the benefit of non-Nordic readers, names beginning with the lerters ø and A have been alphabetised as if spelled with O and A.
The editors
CHAPTER 1
INTRODUCTION: LIVING CONDITIONS AND COMPARATlVE REsEARCH
Knud Knudsen
It is interesting to note the strong position that living conditions studies - as a scientific approach and political tool - today enjoy in Norway, presurnably a fairly representative Nordie and Scandinavian country. As more and more surveys have been made on the national level - and developed into panel designs - these studies have given a framework for painting a picture, from a bird's-eye view, of social developments and trends in the nation. However, at least in Norway, living conditions studies have become important also at lower regional levels. In a typical case, a Fafo study ofliving conditions in Oslo - Norway's capital - told about the divided city, and the large differences between the haves and the have-nots. Later this study was followed by a survey of the standard of living of immigrants and refugees in the Oslo area. Also typically, when Oslo has its living conditions studies, Bergen - the next largest city, and a very ambitious one - must have its own. The Bergen study involved an initial survey of a sample of 1 0,000 Bergen residents.
Not only cities, but also regions and even various occupational groups, now have their own studies. In 1 995 there was a survey of living conditions of artists, and in 1 995 to 1 996 a comprehensive study of farmers was done.
In Scandinavia this movement started 30 years ago, introduced by Finnish social scientists and later adopted by Swedes. These studies were influenced by what was called the social indicator movement, which at that time was developing in many industrialised countries in the Western world. Basically, the use of standard economic indicators as the only guide for political decision-making was questioned. Planners seerned to lack adequate information to evaluate whether any progress had been made in important aspects of life.
Politically, there seems to have developed a need for a broader project, beyond that of steering the national economy. In the scientific community the time was ready for the new social sciences, in addition to the more traditional Keynes
influenced economics.
Conceptually the Scandinavian studies were influenced by British social scientist Richard Titmuss's ( I 958) definition of standard of living as:
The individual's command over resources in money possessions, knowledge, physic and physical energy, social relations, security, etc., by the help of
which the individual can control and consciously direct her/his conditions of life.
In short,. the chosen Scandinavian approach implied the following:
• More weight was to be placed on objective resources than subjective feelings and opinions, although the latter also had their place. In the beginning, there was scepticism about subjective measures because, as Sten Johansson ( 1 972), the sociologist behind the first Swedish study, argued, this would reflect only the political authorities' ability to manipulate feelings and opinions of ordinary citizens.
• The
bad life
should be used as a reference point of measurement more than the good life. The idea was a very pragmatie one. To agree on the ideal society is indeed a difficult question, involving philosophical and moral questions that could probably be diseussed for ever. However, it is easier to agree on what charaeterises bad and difficult living conditions. So let us then start with this as a reference point, it was suggested.• Furthermore, the Scandinavian approach meant focusing on certain
welfare or standard of living components
such as economic resources, health, education, housing, work and family. Thus the idea was that if one wanted a more comprehensive and precise picture of peoples' lives, one had to paint it with different colours, with a multidimensional approach.• The Scandinavian studies, and I think especially the Norwegian ones, where influenced also by the work of American scientist James S. Coleman. He suggested an idea parallel to what is known as input-output analysis in economics. He emphasised that social proeesses are taking place within specific institutional settings, such as the family, occupational life, the school, the local community and so on. So he suggested focusing on understanding the
conversion
ortransformation
of resources within social institutions - or what was calledarenas of action
(Coleman 1 97 1 ). Thus, the conception and development of models, at least in the Norwegian setting, have involved ideas about resources, actors, arenas of action, and conversion or transformation of resources.• Those involved have been concemed with fin ding an optimal balanee between academic and political aims of such research. This means designing a study that can add to our scientific understanding of the proeess of social change as weU as provide operational tools for improved policy planning.
• The Scandinavian research tradition has had an underlying assumption of a fairly harmonious, less conflict-ridden and more stable society. Thus, these studies have been carried out in the context and a strategy of
piecemeal social engineering,
the ide a that social science and politics can be brought together to make society a little better, step by step.Modem political theory emphasises that political decisions require answers to three questions: 1 ) What are the actual conditions? 2) What goals do we have? and 3) What means should be used? The answer to the third question should involve the best available expert knowledge. Given the goals, what is the best way to achieve them? The second question is mainly a normative one and can be answered only
by way of political processes and discussions. The first question is of a different nature. It cannot be answered through discussion alone. Whether living conditions are improving or becoming worse in specific areas involves questions abour which people of ten have different views and opinions. However, reliable answers to such questions cannot be found on the basis of individual experiences or discussions in the mass media. They can be found only if people, under different, relevant and representative conditions, are observed or counted with the help of modem scientific methods. It is such an approach - applying representative surveys and standardised questionnaires - that has provided the main method for Fafo's NORBALT studies.
The main research challenges for the future are linked partly to living conditions studies in general and pardy to applying sueh data as a tool in comparative research.
Generally speaking, there is a risk for surveys of standard of living to become static, isolated and oriented towards the individual. Such data give a good bird's
eye overview of the terrain, at the given time, bur less is leamed abour processes of change, social movements and dynamic forces. This points pardy towards conceptual questions, partly towards the choice of main research strategy. First, while acknowledging the importance of starting from below, with the individual (at the lowest leve\, so to speak), there is a need for supplementing traditional concepts with ideas of collective action, interest groups, and processes of exclusion and marginalisation. TI1US, there is a need for concepts of a more dynamic character which also emphasise forces of social change and causes and consequences of political actions. This is partly a question of long-term research planning and putting past experiences into use in new studies. Research gro ups should pay more attention to other studies when developing their own, so that results will give meaning in the widest possible sense. Much could also be learned, if, as a rourine for each survey, a special gro up of interest or a novellserious problem area were singled our for more thorough questioning. Furthermore, traditional survey methods ne ed to be supplemented by panel studies and analytic methods developed for such data.
Moreover, including ideas of subjective components could make data more relevant. This is an insight that has gradually dawned upon researchers, and little by little such variables have been included in living conditions surveys. On the one hand, this concems bridging two research traditions that for toa long have lived their lives apart - the living conditions traditions and the quality of life tradition.
On the other hand, it is abour being at the forefront in grasping new social changes and developments in a changing society.
I also think researchers in this area should be willing to experiment a bit more.
One may say that the strong research tradition of the Scandinavian approach has played a positive role in institutionalising research groups and resources. However, such a strong academic tradition may lead to too much conformity. Now and then I feel there is not enough creativity or new ideas coming through. To take one
example: Although there are good reasons for sticking to a multidimensional approach, it could be valuable and interesting to discuss which variable(s) should be used, or if we should decide on a simple, unidimensional approach - along the lines applied in economics.
One could add that there is also a need for systematic model building, especially with regard to how resources are created and converted. For too long, researchers in the area have only paid lip-service to Coleman's ideas about resources, actors, arenas of actions and conversion of resources. Basically, the time has come to either integrate such building blocks into a more coherent theoretical framework or discard them and formulate alternative foundations.
Finally, some comments about comparative approaches as such. A basic rule in comparative research is: If you want to compare, you need to do so along the same yardstick. In other words, if you want to detect real differences, you need to measure the variable of interest according to the same rule. This is in fact the argument behind choosing (as far as possible) identical questionnaires, set-ups and so on in comparative research. This fundamental rule for comparative research based on national surveys should still make some warning lights flash. First, seemingly identical variables may have different meaning within different national contexts. This point is not an argument against using identical yardsticks.
However, it implies that certain attention should be paid to understanding specific historical and cultural phenomena. It means that when doing comparative research, one should be especially careful to understand the given context, and take extra care in describing variations in settings, institutions and structural forces of the different countries involved. 5econd, one should be aware of the limitations that comparative strategi es may put on the relevance of the research agenda. If comparative survey research can be meaningfully done only for problem areas where a common yardstick can be found, one may risk that the space for such a common denominator will become quite narrow. This is especially true as more countries are added and when countries with very different historical experiences are brought into the analysis. Thus, one should also look for functional equivalents of concepts, variables and alternative approaches.
50 a double strategy is needed: to take into account dangers of seemingly similar variables having different meaning and, at the same time, to allow for the inclusion of specific and unique observations in the analysis. When standing alone, survey research clearly has its limitations as a comparative strategy. Therefore, researchers from a varied range of disciplines should be invited to become involved in such activities. In such a setting, I think comparative projects like the NORBALT project can fulfil their potential.
References
Coleman, James (1971), Resources for Social Change. New York: Wiley
Johansson, Sten (1972), Frame of Reference of the 1968 Survey ofLevels o/Living in Sweden. Uppsala: Department of Sociology, Uppsala University
Titmuss, Richard A. (1958), Essays on the Welfare State. London: Allen and Unwin
CHAPTER2
NOTES ON THE DEFINITION AND MEASUREMENT OF LIVING CONDITIONS
Stein Ringen
DEFINITION
Two arguments are pursued here. The core of the concept of living conditions is freedom. The core of the concept of freedom is choice.
I am free if I can choose to live in the way I wish to live. How I wish to live, no one knows except me, and even I do not know much about this since tomorrow I may wish to live in a different way from the way I wish to live today.
Freedom of choice cannot be absolute. Each person must accept limits to his or her free choice out of concern for the freedom of others. Furthermore, there are unacceptable preferences, either because they are irrational (darnaging to ourselves) or because they are wicked (darnaging to others). But within these limits, freedom is freedom of choice.
You can have a good life if you can choose how to live your life. We others must assurne that you are sensible enough to make the best of your freedom and accept that the way you choose to live your life is best for you (unless for good reasons you should be in custody) . Others can only judge whether the conditions are present which would enable you to plan your existence in the way you wish.
These arguments are preferred to an alternative position which says that I am free if what I can choose is what I wish to choose, so that I achieve satisfaction.
According to this position, the co re of the concept of living conditions is happiness. There are two problems with this utilitarian position which make it, to me, unacceptable. First, in a free life, I must have the possibility of changing my mind about how I wish to live. It is therefore not enough that I can choose what I wish to choose today, I must also be able to choose what I possibly might wish to choose tomorrow. Second, if the sole criterion of freedom were happiness, the state could convince citizens that what they can choose, or what the state believes they should choose, is what they wish to choose. It would be appropriate, for example, for the state to use the strongest possible propaganda, if that were effective in persuading citizens, because the process of choice itself would have no intrinsic value to citizens, aside from the resulting happiness.
What, then, is freedom of choice? I define it as having the capability to exercise choice. This, I believe, rests on four conditions: ( 1 ) constitutional freedom (i.e.
all citizens are secured basic human and civil rights), (2) personal resources (e.g.
income, competence, health); (3) opportunities for choice (i.e. my society offers me reasonable options to choose between) ; and (4) community standards, by which I understand soeial guidanee for wisdom in choice. The individual needs resources and options, but also a s()cial foothold. Alone, we are helpless, not free.
There is, therefore, in this concept of freedom an ambiguity: in order to have the capability to exereise choice, we must accept some social diseipline and relinquish absolute freedom in the libertarian sense. This is a different (and, I believe, richer and more positive) understanding of free choice than the libertarian theory, which says that people are free when there are no limitations on the free individual who
runs around choosing all the time.
MEASUREMENT
Measurement approaehes can be classified by the type and amount of information used. Starting from the concept of choice, living conditions can be measured either with information which describes the outcome of the choices people have made, or with information which describes the choices they can make. I call these direct measurement andindirect measurement, respectively.
In both direct and indirect measurement, we may find that a limited amount of information is sufficient or that a broader range of information is required.
Measures of well-being can thus be classified as more or less narrow or broad.
Choice is always exercised within limitations and constraints. These are of two main kinds: the resources people command and the options they are offered in their socio-economic environment for making productive use of resources.
Resources are things which can be used to produce or acquire things of value.
In a market economy, money is the basic resource. Other resources are skills, competence, physical strength, endurance, charm, «connections» and so on (and negative resources such as handicaps). The greater a person's resources, the greater his or her range of choices. Of two persons, the one with more money can make more choices. Of two persons with the same amount of money, the one who is more knowledgeable about the market can get more out of his or her means.
The socio-economic environment can be described as a set of arenas in which people work, live, buy or are active in other ways, such as the family, the consumer market, the labour market, the neighbourhood and the political arena. The more arenas open to YOll, the more options these arenas offer; the less risk they impose (including, if relevant, the risk of too much choice) , the greater your range of choices. Having money will do no good if there are no commodities to be bought or if stores are closed to you because of racial prejudice. Even the most skilled worker can get a good job only if there are openings in the labour market for his
or her particular skills. Family formation becomes difficult in communities in which HIV is endemic.
People make choices by us ing resources in arenas, for example money in the consurner market, skilIs in the labour market, shrewdness in the political arena and so on. This produces outeornes of various kinds, for example consumption, a job, a political position or other things people may seek. The outeornes a person achieves depend on his or her resources, the arenas which are open to that person and the options these offer, and the productivity of his or her particular resources in these particular arenas.
If we use information on outeornes, we measure living conditions directly. If we use information on resources and/or arenas, we measure living conditions indirectly.
The issue of broadness is simply a question of how much information we require before we are willing to trust or accept what we are offered as a valid measure of living conditions. Measures are at their narrowest when they are based on a single indicator. While the directlindirect distinetion is a dichotomy, narrowness/broadness is a continuum.
By far the most used measure is (disposable) income, which is a single-indicator indirect measure. This I eaU the income approach.
In order to take income as a valid measure of living conditions - that is, to accept that two persons with the same income have the same range of choices, we need to accept two assumptions: that income is the only resource which matters for what choices people can make, and that all people operate in idemical arenas (or markets). If these assumptions are considered too strong, we can broaden the measure by including more information.
The first possibility is to include information on resources other than income.
This would eliminate the problems caused by the first of the above assumptions.
This is the solution suggested by Titmuss in Income Distribution and Social Change ( 1 962) . He demonstrated a number of biases in the income approach and suggested instead an approach based on «command over resources-in-time». This I call the resource approach.
The second possibility is to include information on arenas in addition to resources. This would eliminate als o the problems caused by the second assumption. This solution has been suggested by Sen in the concept of
«entitlements» in Poverty and Famines ( 1 9 8 1 ) and in the related concept of
«capabilities» in later works (e.g. Inequality Re-examined, 1 994). These concepts can be interpreted as expressing a need to take imo consideration not only resources but also what use can be made of resources. Money is of no lise if there is nothing to buy with it; nor is the availability of things to buy useful if I am denied the right to buy them. Only if there are things to buy and I have the resources required to buy them do I have an emitlement. Capabilities express a person's freedom to achieve valuable functioning. In order to function, we must have both resources to make use of the opportunities our environment offers and opportunities in our
environment for making productive use of the resources we have. This I caU
the capabilities approach.
The simplest direct measure of living conditions is consumption expenditure, which is a single-indicator direct measure. This I caU
the expenditure approach.
Like the income approach, the expenditure approach rests on two strong assumptions, namely, that what we buy is a valid expression of what we consume and that what we consume is a valid expression of living conditions. Again, these assumptions may easily be considered toa strong, and again the problems caused by imposing such assumptions can be eliminated by introducing more information.First, we can include information on non-market consumption, such as consumption produced within the household, the consumption of «free» services transferred from outside the household (notably from government) and even the consumption of leisure (following Becker) . This would eliminate the problems caused by the first assumption above. This I caU
the consumption approach.
Second, we can broaden the measure beyond consumption, as we usually understand that term, by including more information on how people actuaUy live, for example on health, leisure, social relations and the like, thus eliminating also the problems caused by the second assumption. This is the procedure usually associated with the concept of social indicators. This I caU
the way oflife approach.
We now have a typology of measurement approaches with three types of direct and three types of indirect approaches.
A
M/h,nlm711roaehes to the measurement
Narrow
Broad
CHAPTER3
INTRODUCTION TO THE NORBALT DATA SETS AND
GUIDELINES FOR THEIR USE
Vida Chesnuityte
INTRODUCTION
The NORBALT living conditions project represents a great step forward in the formation of data bases on social conditions in the Baltic countries. Why is this so? The official statistical offices in the three countries have collected statistical data. Research institutes have conducted a num ber of sociological surveys on different aspects of people's lives. There have even been surveys which have included one or more living conditions indicators or components. Some of the questions asked have been similar or identical to questions asked in living conditions surveys in the Scandinavian countries. Moreover, during Soviet times statistical data were not always available to the research community or officially discussed, and the reliability of such statistical data was of ten suspect. Estonia has a long tradition of conducting living conditions surveys, dating from Soviet times (see Anu Narusk's article in this volurne) . In Lithuania a separate living conditions survey, coordinated by Fafo, was conducted in 1 990. But the NORBALT living conditions project represents something new, because it
includes three parallel full living conditions surveys
in Estonia, Latvia and Lithuania. IThe NORBALT surveys in the Baltic countries are design ed in a way that makes it possible not only to analyse living conditions in each country, but also to compare living conditions in Estonia, Latvia and Lithuania. The design of the survey allows for comparison due to the following elements:
• The
same methodological instruments
are us ed in all the NORBALT surveys, based on the Scandinavian model of living conditions surveys. In the Scandinavian countries, living conditions surveys have been carried out since the 1970s, so these countries have built up considerable experience in this activity. Applying this experience and tradition, we can assurne that manyl In the NORBALT projeet two additional living eonditions surveys were eonducted - in St.
Petersburg and Kaliningrad. In this article, however, the foeus will be only on the living conditions surveys eonducted in Estonia, Latvia and Lithuania.
possible mistakes and shortcomings have been avoided during the different stages of the surveys.
• The surveys have
one coordinator,
which ensures that the aspects of comparability are preserved. The field work was carried out by local institutions, bur the Institute for Applied Social Science (Fafo) in Oslo was coordinator for all the surveys.• All three surveys were conducted
at the same time
-in September 1 994.• The Baltic countries have shared a common history during the past 50 years.
Estonia, Latvia and Lithuania were all made into Soviet republics and integrated into the centralised Soviet system. Although these states are independent today, many elements from the Soviet past continue to make an
impact on the daily lives of the Baltic populations.
Against this background there are strong reasons to assurne that the survey data are of high qualiry and can be used for comparisons between the Baltic countries.
Moreover, many of the data from NORBALT are comparable to data from the living conditions surveys conducted in the Nordic countries, so to a certain extent they can be us ed fot comparison between the Baltic and the Nordie countties.
However, the NORBALT data should not be used unctitically for comparative purposes. In addition to the conceptual problems of comparing countries with important demographic, historical and institutional differences, there are a number of technical challenges. A great amount of caution must therefore be used by all researchers who want to carry out comparative studies based on NORBALT.
Certain problems are related to the analysis of individual countries. An intimate understanding of the structure of the different files and units of analysis is needed for a proper· analysis on the national leve!. Other problems are related to comparisons between the survey countries internally, and between the survey countries and other countries which have data of a different format.
In this article we shall give some guidelines and information that can be useful for potential users of the NORBALT data sets. After a general discussion of the NORBALT data files and theit structure, we shall move on to look at the technical problems that need to be taken into account when using the files for compatative analysis.
THE STRUCTURE OF THE DATA FILES
The questionnaire
The intention behind the NORBALT living conditions project was to investigate the welfare of
individuals
andhouseholds
in the Baltic countties in transition. The units of analysis were therefore both individuals and households. The questionnaire was designed so as to make it possible to include analysis both on the household and on the individual leve!. The structure of the questionnaire is illustrated in Scherne l.Scheme 1: The structure of the standard questionnaire
Part I Part III Part Il
Households: Individuals in the Random Seleded
Households: Individuals:
- Housi ng Conditions - Demographie Data - Employment Status - Economic Resou rces - Working Cond itions - Health
- Crime - Migration - Social Contacts
- Personal Attitudes - Life (yde
The first part of the questionnaire had the Household (HH) as the unit of analysis. In this part we find different questions about housing conditions and residential environment. These questions would normally be answered by the Randomly Selected Individuals (RSI) , bur in some cases where the individual was not able to provide the necessary details, a responsible adult answered this part of the questionnaire.
The unit of analysis in the second part of the questionnaire was Individuals in the Household (ID). This part included questions of three types: 1) demographie background characteristics: age, gender, ethnicity, education and citizenship; 2) employment and employment status; and 3) economic resources: income sources and activity in the informal economic sector. The questions were asked with regard to all the individuals in the household, and if the RSI did not have sufficient information about the various persons in the household, a responsible adult also answered this part of the questionnaire.
In the third part of the questionnaire, the Randomly Selected Individual (RSI) was the unit of analysis. This part included questions on work-place conditions for those who are working, health, crime and personal security, migration, social contacts, participation in social organisations, and a number of questions on personal attitudes.
One standard questionnaire was prepared for the NORBALT living conditions survey. However, as survey priorities varied somewhat from one country to another, it was possible for each project team to adapt the questionnaire to local conditions. Sometimes, additional questions were included in the different modules of the questionnaire; in other cases, even completely new modules were included. For example, in Lithuania a module on consurner rights was added to the questionnaire. In Estonia, where the preparations for a living conditions survey had already started before Fafo introduced the NORBALT project, the questionnaire included a module abour material conditions and needs as weU as a number of additional questions in the various other modules.
Sample design
As already mentioned, there are
three types
of units
of analysis
in the survey:Households, Individuals in the Households and Randomly Selected Individuals.
This required a sample design which took this complex structure of the survey
into account. At the same time, the sample design depended on the fact that the only available lists of the population in each country were lists of individuals aged 1 8 and over.
The sample was stratified according to the type (urban or rural) and size of towns and communities. In large towns the sample was carried out as a one-stage stratified simple random sample: the sample was characterised by having very few, usually only one, Primary Sampling Unit (PSU), with fairly large takes within each.
In rural areas there were usually a num ber of PSUs, with one or two selected, depending on the total population size (i.e. a two-stage stratified sample was provided) . The units of selection at the first stage were municipalities. At the second stage, individuals were drawn from the population registers.
RSIs were selected in each municipality. During the interviews, data were recorded abour households and individuals in the households in which the respondents lived. In other words, households were included through the selection of individuals. Thus, sample sizes and the probabilities for selection differed for households, individuals in the households and RSIs because of the selection strategy chosen for the final stage of the sample (Table 1 ) .
Table 1: Sample sizes for analysis units
Estonia Latvia Li1huania Total
Number of HH 4,455 3, 1 32 2,41 2 9,999
Number oflD 1 3, 247 9,446 7,741 30,434
N umber of RS I 4,455 3, 1 32 2,4 1 2 9,999
The sample design had severai constraints in all three countries. The most important was that only registers of individuals aged 1 8 and over were available at the time of the survey. There are no indicators that the registers were particularly incomplete for specific gro ups, in terms of age, gender or ethnicity. There was little information on the population distriburion in smaUer areas, such as parts of towns or municipalities. Population estimates from different sources were not entirely consistent, and some were not easily available. Frequently, people were not living live at the addresses fixed in the register; therefore, it was decided to regard the sample as a sample of addresses, rather than of individuals.
Since the sample design is complex, common statistical tests for significance or confidence intervals which ass urne simple random sampling are not appropriate.
Estimations could be provided by means of the programme CENVAR (US Bureau of the Census, 1 993). It indicates how variance differs between variables, as well as the generalIevel of the variance.2 Statistical tests for significance or estimation of the confidence intervals for the data bases consisting of all three countries' data is more complicated and could be resolved by appropriate software only, for example SUDAAN.
2 More about the sample design of the NORBALT living conditions surveys can be found in Grøgaard (1996), Aasland (1996) and Knudsen (1996).
The structure of the data files
In each country two files were created: one for Households (HH) and Randomly Selected Individuals (RSI) and a second one for Individuals in the Households (ID) (Scherne 2, 1 st stage) .
Scheme 2: Stages in the development of the NORBALT living conditions project data base
5tage 5 5tage 1 5tage 2 5tage 3 5tage 4
Estonia
I nd ivid uals i n the Households ( 1 3242)
Latvia
t
Sepa rate files for a l l Ba ltie countreis
(9999):
_Ag g r.
L-_____ --"
Variables
Lithuania Households, Random selected i n d ividuals (24 1 2) I n d ivid uals i n the H ouseholds
(7741 ) _ Aggr.
Va riables
--- Housing
- Work - Working
con d itions - Health - (rime - M i g ration - Social contacts - Pol itica l views
In the data file for HH and RSI, each case includes variables related to questions from the first and third parts of the standard questionnaire (Seheme 1 ) , i.e. they are related to two different types of analysis units. These files hold 4,45 5 eases in Estonia, 3, 1 32 eases in Latvia and 2,4 1 2 cases in Lithuania.
Each case in the data files for ID includes variables related to the questions from the seeond part of the questionnaire. These files indude 1 3,247 cases in the Estonian data file, 9,446 eases in the Latvian data file and 7,74 1 cases in the Lithuanian data file.
After the data entering and cleaning was completed, further transformations of the original files were made and new variables based on information from the
different questions were created. For example, in the ID file, two labour force category variables were created (based on the variables about employment status, using ILO definitions of participation in the labour force), along with household composition (created from a variable that indicates relationships between members in the household, us ing Aggregate and Merge commands of SPSSPC) and the household's dependency ratio.
In the 2nd stage (Scherne 2), variables on employment status and economic resources from the ID data files were aggregated to the household level and merged with the first data file (i.e. for HH and RSI) . Background variables (age, gender, nationality, etc.) and new variables related to the Randomly Selected Individuals from the ID data file were directly merged with the first data file. In this way, one data file was received for each country with the variables described above for all three types of units of analysis. They hold 4,455 cases in the Estonian version, 3, 1 32 cases in the Latvian version and 2,4 1 2 cases in the Lithuanian version.
During the reconstruction of the NORBALT data files in the 3rd stage (Scheme 2) , one common data base for all three countries was created. This file was constructed by merging the three files from the 2nd stage and thereby forming a new one. This new data file contains 9,999 cases. It indudes all variables from the previous stage (i.e. related to the questionnaire questions), recorded variables and newly created variables. The structure of this file is linked to the standard questionnaire, so before the merge, variables not identical for all three countries were unified into one standard form: names and scales were changed where possible, and in some places two variables were recoded into one (mostly in the Estonian data file) . One additional variable was included for country identification.
For data analysis from this common NORBALT data base, eight separate files (Scheme 2, 4th stage) described above were made: Housing, Work, Working Conditions, Health, Crime, Migration, Social Contacts and Political Views. A data file for Life Cyde was not created because of great differences among the three countries con cern ing this module. Eight new data files incorporate appropriate data from the HH and RSI data files, background variables and newly created variables for the Randomly Selected Individual and the country identification variable. For example, the data file called «Housing» indudes variables (related to the questions from the first part of the questionnaire) from the HH and RSI data files, aggregated variables from the ID data file and the country identification variable. Each of these eight data files indudes 9,999 cases.
One more file from the ID data files was constructed for all Baltic countries (Scheme 2, 5th stage) . This file includes variables for individuals in the households and is related to the questions from the second part of the questionnaire, newly created variables (on labour force category, economic resources and household composition) and the country identification variable. This file holds 30,434 cases.
During each stage of the transformation from the original form to the last stages, the NORBALT data bases contain four additional variables for the data
weighting. The weights are calculated as the inverse of the selection probabilities.
There are two types of weights in each data file: expansion and relative weights.
In practice, the expansion weights expand the sample to represent the total population. Relative weights are us ed in the analysis. They are the expansion weights divided by the mean of the expansion weights. The weights were adjusted for non-response by multiplying the weights with the inverse of the response rate, calculated for each PSU or stratum (for the strata without two-stage sampling) . Both expansion and relative weights have two types: household level and Randomly Selected Individual level. This requires some explanation: The nrst type of weight must be applied when using the ID data file to analyse the individuals in the households. This weight should als o be applied when using the HH and RSI data file for analysis of households. The second type of weight is applicable only to the HH and RSI data files for analysis of the RSIs (related to the questions from the third part of the questionnaire) .3
Table 2: Diffirent types of weights used in the NORBALT data files
Expansion weights Relative weights HH level RSl level HH l evel (a_hrelw*) RSl level
(a_hexpw*) (a_rexpw*) (a_rrelw*)
estimate the estimate the
ID data file actual size of - sample size of -
population population
estimate the estimate the estimate the estimate the H H/RSI data actual actual number of sample size of sample size of
file number of people aged 1 8 popu lation aged
households and over households 1 8 and over
* Name of the welghts variablcs in the data file.
For calculations for the Baltics as a whole, expansion weights could be used. But in this case any regression analysis, signincance test, etc. will show signincant results. However, due to the complex sample design, SPSS-type tests of signincance are not valid.
POTENTIAL TECHNICAL PROBLEMS IN DATA ANALYSIS
The NORBALT living conditions project surveys have provided data which give oppoftunities for a wide range of living conditions studies. First, it is possible to analyse both households' and individuals' living conditions in each country. Second, one can also compare living conditions between different countries. Thus, one could say that there are two main directions of investigation. We shall concentrate on potential problem areas where there is a need for extra caution using the
l More about the weights used in the NORBALT living conditions project survcys can be found in Grøgaard ( 1 996), Aasland ( 1 996) and Knudsen ( 1 996) .
NORBALT data bases, relating to 1 ) data analysis in a separate country and 2) international comparison.
Analysis of living conditions in a separate country
Theory on living conditions research has shown that living conditions of individuals are explained by demographic and socio-economic background variables, environmental conditions, attitudes and economic behaviour.
Background variables include age, gender, nationality/ethnicity, education and household composition. Environmental variables, such as a country's unemployment rate, level of public school expenditures and degree of urbanisation of respondents' residence, indicate access to resources. Attitudes include indexes of aspiration-ambition, trust-hostility, and personal efficacy. Measures of economic behaviour relate to planning ahead, risk avoidance, and connectedness to sources of help and information (Kiecolt and Nathan 1 98 5) .
The NORBALT data hold all required variables for a full living conditions study. However, some peculiarities of the questionnaire and of the sample design give reason for caution. The three types of units of analysis discussed above (Households, Individuals in the Households and Randomly Selected Individuals), with different sample sizes and appropriate variables, require us to keep in mind the relations between the units of analysis and variables during analysis. According to Coleman's sociological theories classification, there are three groups forrned according to their multilevel content:
In the first group, variation in a dependent variable is explained through independent variables that apply to the same social level (e.g. country, organisation, local community, individual) . In the second group, variation in a dependent variable at one level is explained by processes that operate at a higher level. In a third type, variations in outcomes at one level is explained by variation in outcomes at a lower level. (DiPrete and Forristal 1 994:332)
This means that we need to make clear what variables are dependent and what variables are independent, and for what analysis units (or levels) they are applicable.
In our case, during RSI analysis the independent variables (or «explanants») are age, gen der, ethnicity, education, living place and household composition. Other variables serve as dependent (or «explanators») . The latter variables relate to the above-mentioned three types of analysis units from the different sample sizes.
Variables related to the questions from the third part of the questionnaire have the same sample size as the independent variables for the RSIs. They can be directly applied to the calculations. This version corresponds to the first type of the sociological theories: both dependent and independent variables are from the same social level. From this relationship is it possible to investigate the individual living conditions through personal attitudes and feelings. At the same time, these variables can serve as explanators for the household level and show relationships of variables corresponding to the second type of the sociological theory by the