Jump to ContentJump to Main Navigation
The Quality of Life$

Martha Nussbaum and Amartya Sen

Print publication date: 1993

Print ISBN-13: 9780198287971

Published to Oxford Scholarship Online: November 2003

DOI: 10.1093/0198287976.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use. date: 05 August 2021

Descriptions of Inequality: The Swedish Approach to Welfare Research

Descriptions of Inequality: The Swedish Approach to Welfare Research

(p.67) Descriptions of Inequality: The Swedish Approach to Welfare Research
The Quality of Life

Robert Erikson

Oxford University Press

Abstract and Keywords

By the 1950s it had already become clear that, in spite of its widespread use, per capita GNP is an insufficient measure of the well‐being of citizens. Thus, in 1954, an expert group within the United Nations suggested that we should not rely on monetary measures alone: the measurement of well‐being should be based on several different components, together making up the level of living....

By the 1950s it had already become clear that, in spite of its widespread use, per capita GNP is an insufficient measure of the well‐being of citizens. Thus, in 1954, an expert group within the United Nations suggested that we should not rely on monetary measures alone: the measurement of well‐being should be based on several different components, together making up the level of living.1 Partly influenced by the UN expert group, Johansson made level of living, seen as a set of components, the basic concept in the first Swedish Level of Living Survey conducted in 1968.2 This survey has been followed by a number of similar studies, both in Sweden and in the other Nordic countries. To exemplify the Swedish approach to welfare research, I will here use the first survey from 1968 and its direct followers, conducted by the Swedish Institute for Social Research in 1974 and 1981. Apart from minor details, however, what I say also applies to what has been done by the Swedish Central Statistical Office as well as by other Scandinavian research organizations.3

The measurement and description of welfare implies a response to a series of questions. One concerns the basis of welfare measurement: should it be related to the needs or the resources of individuals? Another question is whether she herself or an outside observer should judge the individual's welfare. Furthermore, we must decide which types of indicator to use and how to use them, how relevant descriptions should be provided, and how we can give an overall picture of the individual's welfare. I will return to these questions later and discuss how they have been answered in the Swedish level of living surveys. Before doing so I will present some results from these surveys in the hope of making the issues clearer and more concrete.

1 The Level of Living Surveys

In 1965 the Swedish government set up a commission with the task of describing the conditions and problems of low income earners. The commission planned its task in three steps: (1) a study of the distribution of factor income, (2) a study of the distribution of disposable income, and (3) a study of the (p.68) distribution of welfare in non‐monetary terms. This third study was carried out by a group of sociologists who communicated their results to the commission in a series of reports.

For the purpose of the third study, around 6,000 people in the 15–75 age range living in Sweden were interviewed in 1968. In 1974 and 1981 surviving people under 76 years of age still living in Sweden were interviewed again. At both these later interviews young people and recent immigrants were added to the sample in order to make it representative for the Swedish adult population. The interviewees were asked about their living conditions in nine different areas or components of life. A large number of indicators were used for most of the components. The components together with some typical indicators are shown in Table 1.

Table 1 Components and Some Typical Indicators in the Swedish Level of Living Surveys*



1. Health and access to health care

Ability to walk 100 metres, various symptoms of illness, contacts with doctors and nurses

2. Employment and working conditions

Unemployment experiences, physical demands of work, possibilities to leave the place of work during work hours

3. Economic resources

Income and wealth, property, ability to cover unforeseen expenses of up to $1,000 within a week

4. Education and skills

Years of education, level of education reached

5. Family and social integration

Marital status, contacts with friends and relatives

6. Housing

Number of persons per room, amenities

7. Security of life and property

Exposure to violence and thefts

8. Recreation and culture

Leisure‐time pursuits, vacation trips

9. Political resources

Voting in elections, membership of unions and political parties, ability to file complaints

(*) In the first survey in 1968 no questions were asked about security of life and property, whereas questions were included about diet and nutrition.

(p.69) In a report on the three surveys which was published in 19844 the overriding aim was to answer three questions:

  1. 1. Had there been an average change in the level of living between 1968 and 1981?

  2. 2. Were there any differences in level of living between different population groups, specifically between men and women, social classes, age groups, or regions?

  3. 3. Had there been any changes between 1968 and 1981 in differences in level of living between groups?

Our attempts to answer these questions followed several routes. Some examples will illustrate how the results were presented.

Three questions were asked about physical mobility: whether the respondent could walk 100 metres briskly without problems, whether he or she could go up and down stairs without difficulty, and whether he or she could run 100 metres without difficulty. In Figure 1 we show results regarding the proportion saying that they had problems in at least two of these three respects, which in nearly all cases included those who said that they had problems in running and climbing stairs. The figure is a diagrammatic representation of the outcome of a logarithmic regression analysis.5

Descriptions of Inequality: The Swedish Approach to Welfare Research

Figure 1 Regression Diagram Of Proportions Of Disabled People

The diagram is an attempt to present a complicated reality in a simple form, though on the basis of a fairly sophisticated statistical technique. Technically, the horizontal lines within each subfield represent regression coefficients and their possible slopes give information on the interaction between the factor in question (sex, age, community, and class) and year of investigation. The vertical lines indicate the approximate length of 85 per cent confidence intervals. Our hope, of course, is that the statistically untrained reader, through such diagrams, will be able to get a grasp of the variations and changes in level of living. The interpretation of Figure 1 would be as follows: the signs in the leftmost field all appear on about the same level, thereby indicating that there has been no overall change from 1968 to 1981 in the proportion of disabled people. The next field to the right shows that women tend to be disabled more often than men, when other factors are accounted for. That the horizontal lines converge slightly (read from left to right) suggests that this difference has decreased slightly during the period. The field thereafter indicates what we could expect: that older people are disabled more often than younger, but also that this difference has diminished between 1968 and 1981. The following field shows that there are no clear differences in this respect between cities, towns, and the countryside, and the rightmost field shows that members of the working class (III) are disabled more often (having accounted for age, etc.) (p.70) than members of the upper middle class (I), with the lower middle class (II) in between.

In Figure 2 we show the corresponding results for political resources, in the sense of having taken part in opinion‐forming activities.6

Descriptions of Inequality: The Swedish Approach to Welfare Research

Figure 2 Regression Diagram Of Proportions Of People Participating In Opinion‐Forming Activities

Figure 2 indicates that the proportion participating in opinion‐forming activities has increased between 1968 and 1981, that men are more active than women but that this difference has diminished, that age differences have disappeared, that there are no differences between different types of communities, and that there are great and non‐changing differences between the social classes.

Table 2 shows, in a more conventional fashion, the inequality in income from employment between different classes and occupational groups. The overall income inequality decreased over the period 1967–80. This decrease was partly the result of diminishing differences between occupational groups, but partly also of lessening inequality within classes. Wages in occupations mainly employing women increased considerably during the period.

Table 2 Earned Incomes for Full‐Time Year‐Round Workers in Same Occupational Groups 1967, 1973, and 1980 (In SEK 1,000s Computed to 1980 Money Values)

Earned incomes

Percentage of average wage

Coefficients of variation










All employees










Professionals executives in private employment










Professionals in public employment




















Private technical and clerical










Public salaried










Metal workers










Other manufacturing workers










Construction workers










Workers in local government










Workers in state government










In an attempt to get a more complete picture of variation and change in welfare problems at the individual level we counted the number of components, out of five, for which we had recorded problematic conditions for the individual respondent. The components were health, economic resources, political resources, social relations, and housing.7 Figure 3 shows the variation and (p.71) (p.72) change in the proportion showing three or more of these problematic states.

Descriptions of Inequality: The Swedish Approach to Welfare Research

Figure 3 Regression Model For Proportions With Three Or More Problems In the Population As a Whole

Figure 3 shows that the proportion with many problems has decreased slightly over time, that women are more often exposed to many problems than men, that many problems are relatively common among old people, that there are small and unsystematic differences between different types of community, and that problems are more common in the working class than in other classes. Generally, the relative differences between groups seem to be stable over time.

Overall, a slight improvement in the average level of living seems to have taken place in the years 1968–81, especially in the areas of housing and education. Women's position relative to men improved considerably in most areas. The relative differences between age groups decreased slightly in terms of economic resources, as the positions of young and old people improved compared to those in the intermediate age groups. Differences between the social classes decreased slightly.

These are some examples of how variations and changes in the level of living in Sweden have been described. What, then, is the theoretical rationale behind this approach? I will first discuss the concept, then continue with some of the problems of operationalization, and finally consider some aspects of its presentation. In doing this I will return to the questions previously mentioned.

2 The Concept

The individual's command over resources in the form of money, possessions, knowledge, mental and physical energy, social relations, security and so on, (p.73) through which the individual can control and consciously direct his living conditions.’ This was the definition of level of living given in the first discussion of the concept in connection with the 1968 survey.8 The central element is the individual's ‘command over resources’, which was extracted from Richard Titmuss's writings on welfare,9 but a discussion of command over resources can be found within economics as well.10 The emphasis on several different components of welfare was taken from the writings of the United Nations expert group, referred to above. To judge the level of living of an individual or a group, we have to know their resources and conditions in several respects, which are not transferable between each other. To have knowledge about, for example, economic conditions is thus not enough; we also have to know about health, knowledge and skills, social relations, conditions of work, etc., in order to determine the level of living. There is no common yardstick through which the different dimensions could be compared or put on a par. No objective or impartial way exists by which it would be possible to decide which of two men is better off if one of them has, for example, worse health but better economic conditions than the other.

Welfare or level of living seems, at least in the European tradition, to be based either on people's needs or on their resources.11 If needs are made central, then the concern is with ‘the degree of need‐satisfaction’.12 If resources are made central then the concern is rather with man's capacity to satisfy those needs or, more generally, to ‘control and consciously direct his living conditions’; the individual's level of living will thus be an expression of his ‘scope of action’.13 Resources, as understood here, seem to be very close to Sen's concept of capabilities. And, as Sen points out, well‐being freedom, that is, a capability to achieve satisfaction in many respects—or, as termed here, a large scope of action—is not only a means to achieve a high level of satisfaction, it is of value in itself.14

To base the concept of level of living on resources rather than on needs has some advantages. We then look upon man as an active being who uses his resources to pursue and satisfy his basic interests and needs. We do not necessarily have to decide what these needs are, the individual is assumed to use his resources in his best interests. On the other hand, we do have to decide what the most important resources are and, in so doing, we have to consider for what purpose they can be used. Thus, one way or another, we must take (p.74) a stand on what the most central areas of human life are, the areas where it is most essential that the individual should be able to determine his living conditions.

However, it does not seem sufficient to restrict the concept of level of living to resources alone. Some conditions, especially good health, certainly are important resources, but this does not exhaust their significance. Their most important aspect may be as ends in themselves. Furthermore, some circumstances, such as the quality of the work environment or the amenities and space in the home, are important for the individual's well‐being, but can only be regarded as resources in a very remote sense. Thus the level of living concept would be too restrictive if we based it on resources alone without adding essential conditions. Moreover, the same set of resources is not of equivalent value regardless of the context. A certain education, say in law, may be of high value on the labour market in the country where it was acquired, but of very limited value in another country. We must therefore consider the arenas in which the resources are to be used.15 Individuals' resources and the characteristics of the arenas where they are to be used together determine the scope of individuals for directing their own lives.16

In essence, then, the position taken in Swedish welfare research is that the individual's resources, the arenas in which they are to be used, and his most essential living conditions, make up his level of living. This position, although independently arrived at, seems to be very close to Sen's as he writes that ‘the central feature of well‐being is the ability to achieve valuable functioning’.17

There is no universal theory which could guide us in deciding what the most important resources and conditions are. We therefore have to base our choice on mainly general considerations. The nine components referred to above (see Table 1) do not constitute a self‐evident choice, but similar lists of the essential areas relevant to level of living look very much the same the world over (possibly to some extent because of communication and mutual influence). The list is to some extent influenced by the situation and culture of Sweden; in a developing country such a list would probably include, for example, access to food and nutrition. It is also obvious that such lists have a political character: they only include items which are at least in principle possible to influence. Thus, for instance, talent and climate are excluded in spite of being quite important for the individual's action potential. The components refer to conditions and problems which we all meet during our lives (p.75) and which are of such importance that there are collectively organized attempts to cope with them in all societies.18

One consequence of the multidimensionality of the level of living concept and of the incommensurability between dimensions is that no simple ordered indicator of level of living can be constructed, either on an individual or on an aggregate level. Differences and changes in the level of living must be described for each component. A total picture of variation and change will thus of necessity be rather complicated, and no satisfactory solution has yet been found to the problem of how to present such a picture. The indicator based on the number of components for which problems were registered, shown in Figure 3, was used as part of an attempt to describe the coexistence and accumulation of welfare problems. The total number of measured ‘problems’ is a crude indicator of the total situation of an individual, which gives each different type of problem an equal weight. This is of course quite problematic, and loads the indicator with implicit value judgements.

However, I do not believe that empirical research in this area is possible if we do not take such decisions, based on explicit or implicit value judgements.19 Descriptions necessitate a choice of indicators, and parsimonious description often also necessitates the amalgamation of indicators into indices.20 Such decisions already have to be taken at the component level. Within the health component we have to decide which symptoms we should consider and how they should be put together in constructing one or several indicators for health. We make similar judgements when we decide which aspects of a total work situation we should measure, and corresponding decisions must be made for the other components.

It then seems as if the question is not whether we should make value judgements or not but rather when we should make them and when we should leave them open. No general answer has been given to this question, but some principles have, overall, been followed. First of all, indicators relating to different components have not been merged into common indices—except for the single case in Figure 3 here. Within areas, indicators have in some cases been put together to create indices, but often only for subareas and in many cases only after some type of dimensionality test like factor analysis. On the whole, indicators relating to clearly different areas of the level of living have not been merged into summary measures but kept apart.

The drawback to this approach is that the total number of indices needed for a complete description of the level of living becomes quite large. It is thus difficult to get an overall picture of the level of living, although it is possible to see how the different components are related to one another. Can (p.76) we then do nothing more than take our large number of indicators and present them one at a time?

I think that within components, we can in many cases order conditions in such a way that we get a small number of ordered scales, or in some cases even only one, which everybody, or nearly everybody, would accept. This would be especially likely if we were working with very broad classes of problems, or perhaps only with dichotomies, distinguishing problematic conditions from others.

But even so we will end up with a fairly large number of incommensurable indicators, say at best with a scale or dichotomy of problematic/not problematic conditions for each of the nine components mentioned before. It would still not be possible to construct a summary ordered measure of welfare, but it would be possible to distinguish between different types of total welfare and to find out how frequent these types are. Even if we cannot order all types in relation to each other, we can find orders within subsets of types. A type which includes all the problematic conditions of another type, but which also includes some additional ones, could clearly be regarded as more problematic than the other, according to a common Pareto criterion.

Of course we would still face quite serious problems. If we dichotomize each of nine components—which is in itself a questionable solution: we lose a lot of information by the dichotomization and will have to put very different conditions into the same category—we will get 512 possible combinations. Thus, we have by that operation not come very far towards a manageable empirical concept, which is feasible for purposes of analysis and presentation.

But it should be possible to reduce such the number of combinations into a smaller number of welfare types. Welfare components are correlated with each other so some combinations would probably be very rare indeed. In the Swedish level of living study we found, for example, a tendency for problems with health, few social contacts, and a lack of leisure activities to go together. These were problems which were especially common among older people. In a similar way economic problems and housing problems went together, especially within the working class and among the old and the young, and a low level of political activity, which is strongly related to educational level, appeared most common among women. Because of such correlations we could probably delimit a smaller number of types of welfare problems, and, furthermore, we would probably find that different types are differently located within the social and demographic structure.

3 A Descriptive or an Evaluative Approach?

The question about who should judge the level of living—the individual or the observer—is partly connected to that about needs or resources. If the concept of welfare (or well‐being) is based on needs, it seems quite natural (p.77) to measure its level by asking people whether they are satisfied or not, while this seems less obvious if the concept is based on resources. The problem with an approach based on people's own assessment of their degree of satisfaction is that it is partly determined by their level of aspiration, that is, by what they consider to be their rightful due.21 This means that measuring how satisfied people are is to a large extent equivalent to measuring how well they have adapted to their present conditions. People who have experienced menial conditions for a long period may turn out to be more satisfied, and thus according to this definition to have a higher level of living, than a person who is used to a very high standard but who has recently experienced a minor lowering of it, an outcome which seems unacceptable. We therefore try to assess the individual's level of living in a way which makes it as little influenced as possible by the individual's evaluation of his own situation. This seems all the more natural as the individual's level of living is to a large extent based on his ‘command over resources’, resources which can be used for the ends which he himself finds most satisfactory.

The empirical question related to whether we should be looking at people's conditions or their satisfaction with these conditions is whether we should use ‘objective’ or ‘subjective’ indicators, a question which was much discussed within the so‐called social indicators movement. Actually, the terms ‘objective’ and ‘subjective’ are slightly misleading, so it seems preferable to use the terms ‘descriptive’ and ‘evaluative’ indicators. With descriptive indicators, the individual is asked to describe his resources and conditions. ‘How much do you receive as monthly salary?’ and ‘What temperature do you normally have indoors in the winter?’ would be typical questions. When we use evaluative indicators the individual is asked to evaluate his conditions. ‘Are you satisfied with your salary?’ or ‘How good is the heating in the winter?’ would be typical questions used. The difference between the two types of indicator should not be exaggerated, however. Descriptive indicators certainly contain evaluative elements, and the indicators used by spokesmen for the different approaches are often quite similar.22

To put the emphasis on descriptive indicators does not mean that whether people are satisfied or not is judged to be of no interest. On the contrary, I would say that it is of very great interest to find out how people's resources influence their level of satisfaction and, of course, how this in turn influences their resources and conditions. But I suggest that welfare—or level of living—should be defined in terms of resources and conditions and is best measured by the use of descriptive indicators. And to the extent that welfare research is coupled to societal planning—which, in Scandinavia, is part of its historical (p.78) legacy but not necessarily of its future constraints—I find it quite essential that descriptive indicators should be used. The data for planning should refer to factual conditions and planning goals should be formulated in terms of such conditions. People's opinions and preferences should influence societal planning through their activities as citizens in the democratic political process, not through survey questions and opinion polls. That is, goals for planning should be set up in terms of factual conditions, not in terms of people's satisfaction with these conditions. It is the assumption that the planning and executive organs of the state act directly to influence people's satisfaction and happiness that is the basis for many of the futuristic hells suggested to us in literary works.

A high level of living, as conceptualized here, is not equivalent to enjoying all the good things in life. This is no drawback, as I see it. There are other good or bad aspects of life, and if we want to study them they must be conceptualized and measured in their own right. A concept which is meant to include everything desirable/undesirable would probably be of questionable value. Moreover, those who enjoy a high level of living are not necessarily satisfied and happy. It is well known that the association between conditions and satisfactions is rather weak.23 Conditions and satisfactions are two different aspects of the good in life. A person who has better amenities at home than someone else is better off in that respect regardless of whether he is more or less satisfied with them. However, if his situation changed so that he had only the amenities of the other, he would probably become even less satisfied and vice versa. In a cross‐sectional sample we would expect to get some association between conditions and satisfactions.24 Over time, however, it is even questionable whether we should expect any relation at all on an aggregate level, at least not if basic needs for food and shelter are covered. People will not usually become more satisfied when the general level of living rises if their relative advantage is the aspect of their conditions that influences their satisfaction, and this seems to a large extent to be the case. Similarly, we cannot with certainty expect any association between conditions and satisfactions among different nations.25

4 Presentation

As described in the introduction, several indicators have been used to describe change over time and differences between various socio‐demographic groups in their level of living. This has been partly the consequence of the indicators (p.79) in most cases being on an ordinal level only—meaning that it is not possible to interpret an indicator value without making some form of comparison—but partly also the result of regarding inequality as a problematic societal condition and, thus, equality as an important political goal.

In these studies inequality has mostly been treated as the variation among socio‐demographic groups rather than over the population at large and, accordingly, measured more through differences and relative rates than through, say, Gini indices or Lorenz curves. This is again a result of the ordinal indicators, but it is also based on the assumption that the socio‐demographic groups are delimited in such a way that people will often be able to identify them in society, and that such a description of conditions in society will therefore be meaningful to a large part of the population. To tie the description of inequality to a recognized social structure, will make the description more pertinent to the political discussion. Moreover, it is a way to try to get round part of the problem of preferential choices in comparisons between individuals. If we compare two people and find that one goes on to higher education while the other does not, this could well be because the first one prefers a higher level of consumption later in life while the other is more interested in earning money immediately. Similarly, if we find that one person is unemployed while another, with the same education, etc., is not, the first may well prefer to work as little as possible and consequently to consume less while the other has different preferences. However, it is more difficult to make such an explanation plausible if we find that people of the same intellectual abilities but of different social origins systematically make different educational choices, or if we find systematic differences in unemployment between persons with similar human capital living in different regions. We cannot, of course, rule out the possibility that the differences in these cases are still due to different distributions of preferences, but those who claim so would have to make plausible their contention that such preferences vary with for example, social class and region.26

The study of inequality in non‐distributable conditions—such as health or knowledge and skills, of which it is not possible to transfer units between persons—also becomes more meaningful when we study inequality between groups rather than between individuals. To study the distribution of handicap over the population at large is, for example, of rather limited interest; it could be supposed only to show some ‘natural’ variation in health. If, on the other hand, as in Figure 1, we show that physical mobility on average and net of age differences varies between social classes, more social explanations must be sought, explanations which in one way or another relate health to the conditions, experiences, and ways of life of people in different classes.

(p.80) 5 Poverty versus Inequality

So far I have not discussed poverty. This is no oversight. As implied by the discussion above, inequality rather than poverty has been the important concept in Swedish welfare research. This follows partly from the emphasis on non‐monetary aspects of welfare, given that poverty refers to economic resources.27 But it is also partly the consequence of an interest in variation over the whole range of a condition and not only over a poverty line. On the other hand, Johansson, in his first discussion of the level of living concept, suggested a concentration on ‘bad conditions’,28 and some indicators are just dichotomies, by which presumed bad conditions are delimited. Many other indicators, however, as mentioned, have the form of ordinal scales. Furthermore, for most dichotomous indicators the dividing line is too arbitrary to be given any other meaning than just being a dividing line. In constructing scales it is regarded as preferable to consider the whole distribution and not to restrict interest to a dichotomy. A desirable result of these considerations is that the intellectually rather empty discussion of whether the proverty line should be drawn here or there has been avoided.

I believe that there is also an ideological ground for this difference in emphasis in welfare research between Sweden and many of the other Western nations. I would suggest that poverty is the main welfare problem for social liberalism, while inequality is the main problem for social democracy. According to classical liberalism, the market is the ‘natural’ mechanism for distributing economic resources. To social liberalism, this is still true, but we have to correct the outcome of the market mechanism in one respect—we must for humanitarian reasons take care of those who end up destitute, that is to say, we must take the poor out of poverty. This can done through what Titmuss called the residual welfare model of social policy, the pursuit of which results in a marginal welfare state,29 in which through governmental activities, the deficiencies of the market are corrected by money transfers to people below the poverty line. According to social democracy state activities are not merely a supplementary mechanism, but one on a par with the market. In an institutional welfare state a redistributive model of social policy should cover the basic needs of all citizens. The variation in essential conditions—between different groups in the population and over the life cycle—should thereby also be diminished. Various social provisions are seen as the rights of citizens in this perspective. The state should therefore, provide health care and education to all—the quality of these services should in principle be such that no demand for private hospitals or schools appears—and it should be possible to have a good standard of housing regardless of income and family size. (p.81) Those who can be expected to be in need should be given support as a matter of course. Child allowances and pensions should therefore, be provided to all.

In different political climates differently formulated questions appear most relevant. In this perspective it thus seems natural that poverty should become the central socio‐political issue where social liberalism dominates the political climate whereas inequality becomes the main problem of welfare where social democracy is dominant.

6 A Political Theory for Social Reporting

In an attempt to formulate a political theory for social reporting, Johansson suggested that political decisions require answers to three questions.30 They are: (1) What are the actual conditions? (2) What goals do we have? and (3) What means should be used? The second question can only be answered in political discussion, it has a purely normative character. The answer to the third question should involve the best expert knowledge: given the goal, what is the best way to get there? The first question is of yet another character. It cannot be answered in discussion—although this is often tried. Whether people's health, on average, is becoming better or worse, whether unemployment is going up or down, whether social selection in school is increasing or not, are questions on which people often have views and opinions, but reliable answers cannot be given on the basis of personal experience or found in the mass media, where they would be based on the methods of journalism and, moreover, probably influenced by the interests and expectations of editors and owners.31 Unemployment or health problems may not necessarily be increasing even if more people in our neighbourhood are falling ill or becoming unemployed, and crime rates may not be rising even if newspapers do start to write more about crime. Reliable answers to such questions can only be found if people in different relevant conditions are counted with the help of established scientific methods. When we started to determine the rate of unemployment through counting the unemployed in representative samples of the population, political discussion could move from issues that cannot be resolved by it—how many are the unemployed?—to questions which, at least in principle, can be answered by it—what should we do about unemployment? To give answers to questions about levels and trends of welfare, about what conditions are like and how they are changing is, then, a task for social reporting. That a mechanism for answering the first question is not discussed in the theory of the democratic process can be seen as a lacuna in it, Johansson suggests.32

(p.82) Writing social reports seems to be a task for statistical offices rather than for social research institutes, and the major task of social reporting has in Sweden been given to the Central Bureau of Statistics. The task for welfare research is to develop theories, models, and methods in the field. This would include both developing a theory for social reporting and developing models of how the level of living components hang together, their determinants, casual connections, and interrelations. Welfare research in Sweden seems to have come quite a long way in developing ideas and methods for the description of individual welfare, but it has a long way to go in explaining its variations and changes.


Bibliography references:

Allardt, Erik (1975). Att ha, att älska att vara (‘Having, Loving, Being’). Lund: Argos.

—— (1977). ‘On the Relationship between Objective and Subjective Predicaments’, University of Helsinki, Research Group for Comparative Sociology: Research Report 16.

Campbell, Angus, Converse, Philip E., and Rodgers, Willard L. (1976). The Quality of American Life. New York: Russel Sage.

Coleman, James (1971). Resources for Social Change. New York: Wiley.

Drewnowski, Jan (1974). On Measuring and Planning the Quality of Life. The Hague: Mouton.

Easterlin, Richard A. (1974). ‘Does Economic Growth Improve the Human Lot? Some Empirical Evidence’, in Paul A. David and Melvin W. Reder (eds.), Nations and Households in Economic Growth. New York: Academic Press.

Erikson, Robert (1974). ‘Welfare as a Planning Goal’, Acta Sociologica, 17.

—— and Åberg, Rune (eds.) (1987). Welfare in Transition. Oxford: Clarendon Press.

—— and Tåhlin, Michael (1987). ‘Coexistence of Welfare Problems’, in Erikson and Åberg (1987).

—— and Uusitalo, Hannu (1987). ‘The Scandinavian Approach to Welfare Research’, in Robert Erikson, Erik Jørgen Hansen, Stein Ringen, and Hannu Uusitalo (eds.) The Scandinavian Model: Welfare States and Welfare Research. Armonck: M. E. Sharpe.

Johansson, Sten (1970). Om Levnadsnivåundersökningen (‘On the Level of Living Survey’). Stockholm: Allmänna Förlaget.

—— (1979). Mot en teori for social rapportering (‘Towards a Theory of Social Reporting’). Stockholm: Institutet for social forskning.

Korpi, Walter (1983). The Democratic Class Struggle. London: Routledge and Kegan Paul.

Lebergott, Stanley (1968–72). ‘Income Distribution: Size’, in David Sills, (ed.), International Encyclopedia of the Social Sciences. New York: Macmillan.

Hernes, Gudmund, Rødseth, Tor, Aase, Asbjørn, and Ringen, Stein (1976). Levekårsunderskelsen, Slutapport (The Level of Living Survey, Final Report'). Oslo: Universitets Porlager.

(p.83) Michalos, Alex (1987). ‘What Makes People Happy: Social Indicators and Quality of Life’, in Levekårsforskning (‘Level of Living Research’). Oslo: Rådet for samfunnsvitenskaplig forskning.

Ringen, Stein (1985). ‘Toward a Third Stage in the Measurement of Poverty’, Acta Sociologica, 28.

Selén, Jan (1985). ‘Multidimensional Descriptions of Social Indicators’, Social Indicators Research, 17.

—— (1987). ‘Regression as a Tool for Describing Level of Living’, in Erikson and Åberg (1987).

Sen, Amartya (1980). ‘Description as Choice’, Oxford Economic Papers, 32.

—— (1985a). ‘Well‐Being, Agency and Freedom. The Dewey Lectures 1984’, Journal of Philosophy, 82.

—— (1985b). Commodities and Capabilities. Amsterdam: North‐Holland.

Szulkin, Ryszard (1987). ‘Political Resources’, in Erikson and Åberg (1987).

Tåhlin, Michael (1990). ‘Politics, Dynamics and Individualism: The Swedish Approach to Level of Living Research’, Social Indicators Research, 22.

Titmuss, Richard M. (1958). Essays on the Welfare State. London: Unwin University Books.

—— (1974). Social Policy. London: George Allen and Unwin.

United Nations (1954). International Definition and Measurement of Standards and Levels of Living. New York: United Nations Publications.

—— (1966). The Level of Living Index. New York: United Nations Research Institute for Social Development.

Wilenski, Harold, and Lebaux, Claude (1958). Industrial Society and Social Welfare. New York: Russel Sage Foundation.


I am indebted to Walter Korpi, Martha Nussbaum, and Michael Tåhlin for helpful comments on an earlier draft, and to Caroline Hartnell for improvement of my English.

(1) United Nations, 1954. See also United Nations, 1966.

(2) See Johansson, 1970.

(3) Thus, the subtitle of this paper could as well be ‘The Scandinavian Approach to Welfare Research’, which actually is the title of a paper by Hannu Uusitalo and myself (Erikson and Uusitalo, 1987).

(4) The report was published in Swedish in 1984. It was later published in English in a slightly abridged version: Erikson and Åberg, 1987.

(5) The regression analysis and the diagrammatic technique are described in Selén, 1985 and 1987.

(6) Opinion‐forming activity is defined as having spoken at a meeting, written in a newspaper, or participated in a demonstration. See further Szulkin, 1987.

(7) The other components were not included because in some cases it was dubious to delimit a problematic state (education, leisure) or because the component was not included in all surveys (security). Employment and working conditions were treated in a separate analysis only including the labour force. The delimitation of problematic states for the five included components involved a large number of indicators. See further Erikson and Tåhlin, 1987.

(8) Johansson, 1970: 25. Italics in the original.

(9) See especially Titmuss, 1958.

(10) See Lebergott, 1968–72.

(11) In the discussion of quality of life, mainly of American provenance, happiness has also been suggested as a central element, cf. Michalos, 1987, or Campbell et al., 1976. There are, however, a number of objections to making happiness a central element of level of living. See Sen, 1985a and 1985b.

(12) Allardt, 1977. Compare also Drewnowski, 1974: 7.

(13) Erikson, 1974.

(14) Sen, 1985a: 201.

(15) The arena concept was taken from Coleman, 1971, and was introduced to Scandinavian welfare research by the Norwegian level of living survey. See Hernes et al., 1976. Compare also Erikson, 1974.

(16) The arena concept seems quite important in theory, but has never been much used in actual research. Often the outcome of the use of resources in an arena is taken as an indicator, e.g. inequality of the job rather than a particular skill in relation to the surrounding labour market.

(17) Sen, 1985a: 200.

(18) Johansson, 1979: 139.

(19) This should not be read as making any claim for the simple indicator used in Figure 3: that was a provisional solution which seemed to be feasible at the time.

(20) Compare Sen, 1980.

(21) For a more thorough discussion of these matters, see Tåhlin, 1989. See also Sen, 1985a, and Campbell et al., 1976.

(22) Compare the indicators used in the Swedish level of living surveys with those used by Allardt, 1975.

(23) Compare Allardt, 1975, or Campbell et al., 1976. This low association is actually one of the reasons why those who emphasize conditions do not want to use evaluative indicators and why those who put the emphasis on satisfaction do not want to use descriptive ones.

(24) Compare the association between income and general sense of well‐being in Campbell et al., 1976: 55ff. See also Easterlin, 1974.

(25) Compare Easterlin, 1974.

(26) This obviously does not mean that such explanations are impossible or not tried. An example is the well‐known hypothesis that middle‐class children accept deferred gratification to a greater extent than working‐class children do.

(27) Compare, however, Ringen, 1985.

(28) Johansson, 1970: 29f.

(29) Titmuss, 1974. Compare Wilensky and Lebaux, 1958, and also Korpi, 1983.

(30) Johansson, 1979: 112.

(31) When the results from the 1968 survey were published in 1970 and 1971, the standard reaction in the mass media was astonishment over the problems exposed. The general expectation was that most welfare problems had already been overcome.

(32) Johansson, 1979: 121.