Cognitive models of class structure and explanations of social outcomes

20
European Journal of Social Psychology, Vol. 23,445464 (1993) Cognitive models of class structure and explanations of social outcomes GEOFFREY EVANS London School of Economics and Political Science and Nuffield College, Oxford Abstract In this paper we argue that people’s explanations of a wide range of social outcomes occur within a framework of expectations derived from beliefs about the pervasive influence of social class on individuals’ life-chances. This claim is tested by examining the efSect of varying the social class origins of vignette characters on judgments and explanations concerning their outcomes. Four domains of social activity are examined: occupational attainment, educationa~ achievement, relationship success, and unemploy- ment. In all of these areas, the class background of the characters was found to be associated with different outcome expectations, future expectations, judgments of res- ponsibility and differences in the ways in which outcomes were explained. The results are consistent with the claim that people have cognitive models in which social class background is associated with particular social outcomes across a range of activities, and that these models are reasonably accurate representations of the relationships between social class and life-chances. This suggests that contrary to theories which have stressed the individualistic nature of belief systems in western societies, social class forms an important part of the popular representation of the influences on occupa- tional, educational and relationship success. INTRODUCTION Despite the recent tendency for certain politicians to talk of a ‘classless society’, Britain has often been thought of as very much a class-divided nation. Social class position is associated with inequalities and divisions in many areas of life: educational achievement (Heath and Clifford, 1990); health and mortality (Wilkinson, 1986); income (Reid, 1989); mental health and psychological well-being (Cochrane, 1983); and political attitudes and behaviour (Marshall, et al., 1988; Evans, Heath and Payne, 1991). Moreover, the attainment of occupation or class positions is itself associated with social class origins (Goldthorpe, 1987). Given the importance of social class in providing differential conditions under which individuals conduct their lives and strive for goals, it would be surprising if it did not have a significant role in people’s understanding of social processes. Thus beliefs about social class are likely to be important elements of popular theories about social relations 0046-2772/93/Q50445-20$15.00 0 1993 by John Wiley & Sons, Ltd. Received I June 1992 Accepted 28 September 1992

Transcript of Cognitive models of class structure and explanations of social outcomes

European Journal of Social Psychology, Vol. 23 ,445464 (1993)

Cognitive models of class structure and explanations of social outcomes

GEOFFREY EVANS London School of Economics and Political Science and Nuffield College, Oxford

Abstract In this paper we argue that people’s explanations of a wide range of social outcomes occur within a framework of expectations derived from beliefs about the pervasive influence of social class on individuals’ life-chances. This claim is tested by examining the efSect of varying the social class origins of vignette characters on judgments and explanations concerning their outcomes. Four domains of social activity are examined: occupational attainment, educationa~ achievement, relationship success, and unemploy- ment. In all of these areas, the class background of the characters was found to be associated with different outcome expectations, future expectations, judgments of res- ponsibility and differences in the ways in which outcomes were explained. The results are consistent with the claim that people have cognitive models in which social class background is associated with particular social outcomes across a range of activities, and that these models are reasonably accurate representations of the relationships between social class and life-chances. This suggests that contrary to theories which have stressed the individualistic nature of belief systems in western societies, social class forms an important part of the popular representation of the influences on occupa- tional, educational and relationship success.

INTRODUCTION

Despite the recent tendency for certain politicians to talk of a ‘classless society’, Britain has often been thought of as very much a class-divided nation. Social class position is associated with inequalities and divisions in many areas of life: educational achievement (Heath and Clifford, 1990); health and mortality (Wilkinson, 1986); income (Reid, 1989); mental health and psychological well-being (Cochrane, 1983); and political attitudes and behaviour (Marshall, et al., 1988; Evans, Heath and Payne, 1991). Moreover, the attainment of occupation or class positions is itself associated with social class origins (Goldthorpe, 1987). Given the importance of social class in providing differential conditions under which individuals conduct their lives and strive for goals, it would be surprising if it did not have a significant role in people’s understanding of social processes. Thus beliefs about social class are likely to be important elements of popular theories about social relations

0046-2772/93/Q50445-20$15.00 0 1993 by John Wiley & Sons, Ltd.

Received I June 1992 Accepted 28 September 1992

446 G. Evans

(Ossowski, 1963; Bulmer, 1975; Kluegel and Smith, 1981), and to influence the ways in which a variety of social outcomes are explained.

Nevertheless, although this argument has prima facie plausibility, there is also an alternative thesis, which emphasizes the individualistic ideology which has been thought to be pervasive in western societies (Ichheiser, 1949; Huber and Form, 1973; Kluegel and Smith, 1986; Nisbett and Ross, 1980). The individualism argument holds that people believe society to be relatively open and that equality of opportunity, if not of outcome, is widespread. Therefore, class-related social constraints do not influence outcomes: individuals are responsible for their own fates. Accordingly, causal responsibility for outcomes of success and failure should be attributed to the individuals concerned, regardless of their social origins. This set of beliefs, which Huber and Form (1973) refer to as ‘the dominant ideology’, and which Ichheiser (1949) talks of as ‘the success ideology’, has not only been thought to account for some of the attributional biases discovered in traditional attribution theory research (see Nisbett and Ross, 1980), but also to provide a legitimation of widespread social inequalities (Huber and Form, 1973; Kluegel and Smith, 1986).

In the study reported here we test whether individualism is as pervasive as many authors have claimed, or whether people have representations of social processes which give an important explanatory role to social structure and its impact on indivi- dual life-chances. To this end, we examine how beliefs about the effects of social class on individuals’ lives, whether explicit or tacit, provide a context in which expla- nations of individuals’ social outcomes occur.

THE THEORETICAL FRAMEWORK

Cognitive models and explanations

Many of the most well-known approaches to attribution research have focused on the formal processes by which individuals may combine information to come to attributions (for reviews see Jaspars, Fincham and Hewstone, 1983; and Hewstone, 1989). However, systematic evidence on covariation between possible causes and consequences for an event is often not available for people to use in their causal reasoning and they must rely instead on their theories and models, whether implicit or explicit, of social processes and human nature (Lalljee, Lamb, Furnham and Jaspars, 1984; Furnham, 1988). Such theories - conceptualized as cognitive struc- tures involving both implicit and explicit knowledge of a particular area of social activity - are given a central role in the writings of several authors who have adapted the notion of scripts from Schank and Abelson (1977), and who argue that explana- tions are embedded in knowledge structures derived from personal experience, the media or other sources of information (e.g. Lalljee and Abelson, 1983; Read, 1987)’. Most commonly, however, they have been conceptualized as schemata: packets of knowledge - concepts and their relationships - which are stored in memory and through which stimuli are interpreted (see Rumelhart, 1984).

Research on the properties of schemata has shown that when one is activated, people introduce assumptions about relationships between phenomena into their

I Similar themes are also to be fo,und in Moscovici’s (1984) discussion of the concept of social represen- tations.

Class structure and social outcomes 447

understanding of events (see Wyer and Srull, 1984, Vols 1-3, for a review of such research). Moreover, people are usually unaware of these assumptions, because in everyday life conscious explanatory activity tends to occur when events and expec- tations do not match (e.g. Weiner, 1985). This suggests that people may often be unaware of the most central aspects of their cognitive models of social processes. Consequently, these will not be easily accessible through direct questioning. However, we can examine the presence of such models indirectly, for if social structural regulari- ties are central components of people’s implicit theories, they should form a frame- work in which explanations of social outcomes occur. Their presence can therefore be inferred from the pattern of explanations and reactions generated by the correspon- dence between information presented to respondents and the expectations derived from their cognitive models.

The structure and patterning of explanations

Consistent with the findings of previous research (i.e. Feagin, 1975; see Furnham (1988) for a review), we predict that explanations can be divided into two main types: One employing sociological and social structural concepts - which we shall refer to as social structural explanation - and the other - individualistic explanation - being of a more psychological nature. Thus, on the one hand there is a sociological or social deterministic perspective, and on the other hand there is a view of human nature in which individuals act, choose and obtain their outcomes as a result of their own attributes (see Lukes (1973) for a more general discussion of these con- cepts)2. The two types of explanation are to some degree alternative ways of explain- ing outcomes. However, they are not mutually exclusive - the same person could emphasize different explanations when accounting for different outcomes. It is in the conditions of application of the explanations that differences in emphasis are likely to appear.

For example, the attainment of upward social mobility by a working class person might be explained in terms of individual attributes. However, this does not necessar- ily indicate the existence of a general preference for individualistic explanations. If the outcome is contrary to the expectations generated by a belief that class origins are an important influence on outcomes, it will be accounted for by an explanation that leaves the belief in social structural regularities intact. This may be achieved, for instance, by emphasizing the ability, luck or personality of the upwardly mobile person. Consequently, an individualistic explanation of upward mobility could indi- cate that the upwardly mobile individual was thought to be successful despite his class origins. So it is the belief that social structural origins have a strong influence on people’s life-chances, which produces an individualistic account of this particular outcome.

In contrast, consider the case of the same middle class occupational outcome being explained individualistically for someone from middle class origins. In this instance it is less plausible to claim that class background is a ‘constraining factor’ on the outcome obtained. Consequently, it is somewhat more likely that such a response indicates the existence of an individualistic or meritocratic theory of occupa-

It is also possible that fatalistic explanations may form another dimension. However, previous research indicates that luck and chance do not usually form an important dimension of explanations (Furnham, 1988).

448 G. Evans

tional attainment. Nevertheless, even in this case, if the same outcome for a person from a working class background was explained even more individualistically, it could be taken to indicate awareness of class-based constraint on individual out- comes; otherwise an extra emphasis would not have been given to the individual characteristics of the character from working class origins.

To summarize the argument: explanations occur within a pre-existing framework of expectations derived from the theories people have about social processes. How- ever, these theories are not simply applied indiscriminately to explain outcomes. The explanations used to understand social outcomes are a product both of prior expectations and the relation of empirical evidence to those expectations. These two sources of influence interact, resulting in an emphasis on particular explanations in particular circumstances. Thus the nature of a person’s cognitive model of social structural constraint on individual outcomes can only be understood by considering explanations within particular contexts. In particular, it is important to compare explanations of outcomes which are consistent with the expectations generated by the person’s cognitive model of social structure, with explanations of outcomes which deviate from what is expected. It follows that we can only confidently infer the existence of a structural or individualistic understanding of social outcomes by consi- dering explanations of the same outcomes obtained by people from different social origins and examining differences in the way the outcomes are accounted for.

HYPOTHESES

From the above reasoning we can derive and compare two distinct models of the representation of the relations between social class position and individuals’ out- comes. First, there is the class-constraint model. This assumes that people have distinct representations of the members of different classes and have different sets of outcome expectations associated with them. These expectations are generated when a person who closely corresponds to a typical class member becomes a focus of explanatory attention. This results in the activation of class-related schemata. As a consequence, this person’s outcomes are interpreted within a cognitive model that causally relates class origins to various outcomes. There are several predictions which can be derived from the class-constraint model.

One implication of the model is that the influence of class origins is not commonly overcome. Therefore, the expectations associated with the model should be indicated by significantly higher levels of surprise at outcomes incongruent with class origins than for those which are congruent (hypothesis 1). Even in instances where outcomes do deviate from class-based expectations, the influence of class origins is likely to be re-asserted in the long run. So congruent outcomes should also be associated with higher degrees of expected future stability than deviant outcomes, thus indicating that the latter are thought to be less stable and more likely to change (hypothesis 2). A further prediction is that outcomes which are associated with social structural causation should also be associated with attributions of personal responsibility to the individuals involved, whereas individuals who obtain outcomes that deviate from the expectations of the class-constraint model should be seen as more responsible for their situation (hypothesis 3). As attributions of personal ‘responsibility’ are moral judgments that may be to some degree distinct from causal attributions, we

Class structure and social outcomes 449

also examine attributions of internal causality and compare responses to the two questions. This allows us to check whether responsibility attribution is in some way distinct from causal attribution.

The final prediction derived from the class-constraint model is that when account- ing for behaviour or outcomes that are congruent with the expectations derived from the model, people should use explanations referring to notions such as social class, socialization and opportunities, which are indicative of a sociological or social structural model of causal influence. However, when an incongruent outcome occurs, attempts to account for it will involve a search for alternative explanations involving notions of individual character, special skills or situations, and personal preferences. As the explanations are expected to form social structural and individualistic dimen- sions, this change in explanatory focus should correspond to a shift of emphasis from the social structural to the individualistic dimension of explanation (hypothesis 4).

In contrast to the class-constraint model is the meritocratic model. This represents an individualistic belief in an ‘open society’, where social origins do not influence the outcomes attained by people. Thus it is assumed that there is equality of oppor- tunity between social classes for the attainment of more or less desirable social out- comes, and that social barriers between classes do not exist. In such a model, social class origins should not influence outcomes. Therefore, there should be no differences in explanations, judgments of surprise, future expectations and responsibility, as a result of the social origins of the actor. In other words, the meritocratic model predicts that the vignette characters’ social class origins will not have any effect on the way their outcomes are explained, or on any of the other judgments elicited from the respondents.

Finally, following from previous research we expect that fate explanations should be of limited importance and should not form part of the two main dimensions of explanation. They should therefore be (a) distinct from social structural and indivi- dualistic dimensions of explanation, and (b) they should not have the same pattern of association with the origin- outcome combinations (hypotheses 5 and 6).

THE EMPIRICAL ANALYSIS

Information was presented to respondents in the form of vignettes. Respondents were asked to rate the importance of explanations of the outcomes obtained by vignette characters with either middle or working class backgrounds and attributes, and to make certain judgments about the situations depicted in the vignettes. In order to assess the range of scenarios to which expectancies derived from cognitive models of class constraints are relevant, we examine patterns of explanation across four domains of social activity.

Occupational attainment

In this domain we compare explanations of middle class and working class occupatio- nal destinations by characters from middle and working class origins. Given the acknowledged importance of individuals’ class locations for their life chances, occu- pational (i.e. class) attainment is fundamental to an understanding of beliefs about

450 G. Evans

inequalities in opportunities and outcomes. Thus, the explanation of why an indivi- dual ends up in one class or another is likely to be associated with many other cognitive, evaluative and behavioural responses.

Educational achievement

Educational success has an important influence on future occupational destinations (Heath, 1981). It is therefore an important aspect of inequality of outcome, related to both class origins and destinations, and is integral to an analysis of beliefs about occupational attainment. In order to assess the role of social structural expectations in this sphere of activity, we compare explanations of educational success and failure for pupils from educational environments and backgrounds characteristic of the middle and working classes.

Unemployment

The assessment of causal explanations for unemployment is a topic that has already received considerable attention (e.g. Furnham, 1982; Schaufeli, 1988), but there is no research which focuses on the role of beliefs about social class in affecting those explanations. In reality, the probability of being unemployed is related both to a person’s social class position (working versus middle) and regional location (North versus South). These two factors are correlated in real life, and probably in people’s beliefs about the causes of unemployment. We therefore examine whether people give more weight to class position or regional location, or if both factors are important elements of their cognitive models of who becomes unemployed, and why they do so.

Personal relationships

The above domains are all of a potentially political and ideological nature. In contrast, personal attraction and relationships are activities not usually discussed with regard to class conflict, or by politicians in their electoral rhetoric. Nevertheless, the degree to which people from different classes inter-marry is an important indicator of class formation (Laumann, 1966). Therefore, in order to see if people’s models of social structural influences on outcomes extend to this less politicized domain of activity, we shall examine explanations of success and failure in marital relations. We do this by examining differences in our respondents’ explanations of within-class and cross-class relationships that are successful or unsuccessful. In Britain successful cross-class marriages are extremely rare (McRae, 1986). Consequently, if people have accurate class-based expectations of relationships, we propose that successful cross-class marriages should be inconsistent with respondents’ expectations, whereas unsuccessful outcomes for cross-class relationships should be consistent with them.

PROCEDURE

The pilot study

A pilot experiment was conducted using as stimulus materials vignettes involving stories about stereotypical representatives of different social classes in conjunction

Class structure and social outcomes 45 1

with expected or unexpected life events. These sets of information were presented to a group of 20 respondents obtained through the Oxford University Department for External Studies, who were asked to generate explanations for the outcomes described in the biographical information they received. Subsequently, a set of expla- nations was derived from the responses obtained. Relevant questions concerning judgments of the characters in the stories and reactions to the events were added to the explanations, and all items were transformed into a 7-point Likert scale format. These questionnaires were then presented to a diverse group of volunteers, ranging from semi-skilled manual workers with no educational qualifications to people with degrees, in order to assess the suitability of the wording and task requirements for people with differing experiences and levels of education. Following this further revisions were undertaken, which were then checked again for consistency of interpre- tation.

Respondents for the main study

The respondents were drawn from the Oxford University Subject Panel. They com- prised 160 adults, 80 males and 80 females, aged 18 to 65, with the majority in the 30 to 40 age group. The educational level of the sample was high: Half had higher education qualifications and most of the others had ‘0’ or ‘A’ levels. Infor- mation on occupations indicated that they were nearly all respondents, or their partners were engaged in professional or administrative work. The questionnaires were distributed by post and were returned using pre-addressed envelopes.

The questionnaire

The questionnaire contains a between-subjects manipulation of the relationship between class background and outcomes in each of the four domains. There are therefore four types of vignette for each domain, each of which has a particular combination of social class origins and outcome: (1) Working class origin-working class type of outcome, (2) working class origin-middle class type of outcome, (3) middle class origin-working class type of outcome, and (4) middle class origin-middle class type of outcome. Class position is indicated by information about parents, education, and occupation. For example, the following vignette was the working class origin-working class outcome story for occupational attainment:

John Burton, aged 39, was born into a poor family in Salford. He went to a local Secondary School before starting work at the age of 15. After many different work experiences he is now a lathe operator in an engineering firm.

The other vignettes in the occupational attainment domain included a middle class background: ‘a well-off family in Petersfield, Hampshire’, or a middle class outcome: ‘he is now the managing director of an engineering firm’. Similar manipulations of class background were used for all of the other areas of activity examined, although in those cases the outcomes focused on employment versus long-term unemployment; educational success or failure at school; and relationship success (marriage) versus failure (breakup). Details of the rest of the vignettes and exact wording of the other explanations and judgments are available from the author.

The explanations of the vignette outcomes consist of everyday wording of the

452 G. Evans

following eight types of explanation: Socialization (i.e. upbringing); opportunities and circumstances; innate attributes and ability; fate (‘It’s just one of those things’); explanations referring to personal problems or particular positive factors (i.e. special skills); social class; personality (or, in the case of the relationships vignette, compatibi- lity of the couples’ personalities); and personal preference. The types of explanation are similar throughout the various domains, although there is not exact comparability, as the pilot study indicated that certain explanations were not seen as applicable in all domains of activity. The explanations were also chosen so as to maintain comparability between different types of vignette within the four domains, although again exact equivalence in wording was not always possible. Finally, respondents were asked several questions about the characters in the vignettes. These questions included judgments of surprise at the outcome, future expectations for the vignette character; and the perceived personal responsibility and ‘internal’ versus ‘external’ causality (internality) of the character.

After reading each vignette respondents were required to indicate the amount of importance they attributed to each of eight explanations for the outcome depicted. No respondent received more than one origin-outcome combination for each domain, or received a particular type of origin-outcome combination more than once. Responses were therefore elicited for a different origin-outcome combination for each of the four domains, with 160 (four groups of 40) respondents for each domain, and 40 respondents in each condition.

Analysis

In the analysis of variance, the four types of vignette for each domain are the indepen- dent variables, and the explanations and surprise, stability, internality, and personal responsibility judgments are the dependent variables. The three degrees of freedom resulting from the four types of vignette are partitioned into three contrasts. These are: (1) Congruence with class-based expectations. This is indicated by comparing means for conditions in which outcomes are consistent with expectations derived from knowledge of class origins, with outcomes which are inconsistent with class origins. (2) Type of outcome. The mean of one type of outcome is contrasted with the mean of the other. (3) Type of origin. This is indicated by the interaction of outcome and congruity effects.

The contrasts for the unemployment vignette are a little different from the others, as there is only one type of outcome. In this case the contrasts are: (1) The effect of social class. This contrasts the means for working and middle class unemployed characters. (2) The effect of region. This contrasts the means for characters who live in the North East versus those who live in the South East. (3) The interaction of class and region.

RESULTS

Judgments of surprise and future expectations

Ratings of surprise

In Table 1 it can be seen that for all domains of activity incongruent outcomes result in higher surprise ratings. However, there are minor variations in the results.

Class structure and social outcomes 453

In the occupational attainment and personal relationships vignettes, incongruent outcomes for characters from middle class origins are more unexpected than for characters from the working class. This is also the case with educational attainment, where the academically successful working class pupil is not significantly more sur- prising than the unsuccessful working class pupil. In the unemployment vignette, both class and region influence ratings of surprise.

Table 1. Judgments of surprise and future stability by type of vignette

Domains Means by conditions Significance of effects Origin: Working Middle Working Middle

class class class class

Occupational attainment Outcome: Working class

Surprise 2.16 4.21 Future stability 5.23 3.67 Educational achievement

Surprise 2.38 3.66

Unemployment

Surprise 2.36 3.23 Future stability 5.18 4.90 Personal relationships

Outcome: Fail

Future stability 5.31 4.34

Region: North

Origin: Working Mixed class class

Surprise 2.27 3.74 Future stability 6.00 4.95

n= (40) (40)

Outcome: Marry

Total n = 160

Middle class 3.46 2.10 5.92 6.00

Succeed 2.35 2.18 4.85 5.73

South 2.78 4.05 5.19 4.93

Working Mixed class class

Part 3.11 2.85 5.68 5.90 (40) (40)

CON OUT ORIG * * * ns ns

t

CON OUT ORIG t t $ * ns *

CLA REG CXR t ns

ns ns ns *

CON OUT ORIG * * ns $

ns $

CON=congruity; OUT = outcome; ORIG=ongin. CLA=class; REG=region; CXR=class xregion. *p < 0.00 1. tp < 0.01. $p < 0.05.

Expectations of future stability

In the occupational attainment, educational achievement and personal relationships vignettes, we find the predicted association between class-related expectations and stability attributions. However, in the occupational attainment vignette, although the downwardly mobile actor is seen as more likely to change his present position than the inter-generationally stable working class character - thus indicating that the middle class origins of the downwardly mobile actor are seen as militating against continued working class status - the situation of the upwardly mobile actor is thought to be more stable.

In the unemployment vignette there are no differences between the various con- ditions in judgments of the stability of unemployment. However, an inspection of

454 G. Evans

the correlations between explanations and future stability ratings shows that an expectation of continued unemployment for the working class character in the North is strongly predicted by ‘lack of job opportunities’ ( r = 0.52, p < 0.001). However, for the middle class person in the South, stability of expectations is predicted by the ‘belief that being unemployed is not too bad’ (r = 0.44, p < 0.01), and ‘a lack of job opportunities’ is negatively associated with such an attribution (r = -0.31), p < 0.05). This suggests that the expectation of continued unemployment for the middle class person was thought to reflect his personal choice, whereas for the working class person it was thought to be related to a lack of job opportunities.

Summary

The analysis of surprise and stability ratings confirms that respondents have differing expectations for outcomes congruent with class origins in comparison with outcomes that are not. Most outcomes that contradict class-related expectations are seen as more surprising and less stable or, in the case of unemployment, stability for unexpec- ted events is associated with an explanation in terms of personal preference, whereas for more expected outcomes stability is associated with a lack of job opportunities.

Judgments of internality and responsibility

Table 2 shows that in the occupational attainment domain judgments of personal responsibility and internality are affected by both the congruence of outcome with class origins and by the outcomes being explained. Upwardly mobile characters are seen as more responsible, and stable working class characters are seen as less responsible for their outcomes than are both characters from middle class origins. Downwardly mobile characters are not attributed a greater degree of responsibility than are those who remain stable in the middle class. These patterns indicate that for judgments of characters from middle class origins, the effect of unexpectedness is countered by the effect of outcome: to be deviant from an unexpected middle class outcome increases attributions of responsibility and internality, but this effect is countered by the tendency for working class outcomes in general to be given lower ratings of responsibility and internality. Inversely, in the case of the upwardly mobile actor the effects of both incongruity with class origins and (middle class) outcome accentuate responsibility and internality judgments.

As with occupational attainment, there are significant congruence and outcome effects for educational achievement: Successful outcomes are given higher ratings of internality and personal responsibility than are unsuccessful outcomes. Conse- quently, the unsuccessful pupil from a working class background produces considera- bly lower scores on measures of responsibility and internality than other origin- outcome combinations and the educationally successful working class pupil is attri- buted greater responsibility and internality than any of the middle class characters.

In the unemployment domain we find a similar pattern to the above, with both class and region significantly affecting responses (although class has stronger effects). As a consequence, the Northern working class character is attributed less responsibi- lity and internality than are the others; the two moderately incongruent conditions (as indicated by the surprise ratings) - Southern, unemployed working class and

Class structure and social outcomes 455

Table 2. Judgments of responsibility and internality by type of vignette ~

Domains Means by conditions Significance of effects Origin: Working Middle Working Middle

class class class class

Occupational attainment Outcome: Working class Middle class CON OUT ORIG

* ns * ns

Responsibility 4.60 5.46 5.97 5.20 t Internality 4.68 5.67 6.10 5.35 t

Responsibility 4.23 4.66 5.13 4.65 * * ns Internality 4.38 4.85 6.15 4.95 * * ns

Responsibility 2.89 3.95 3.81 4.40 $

Educational achievement Outcome: Fail Succeed CON OUT ORIG

Unemployment Region: North South CLA REG CXR * ns

Internality 3.27 4.15 4.11 4.75 t $ ns Personal relationships

Origin: Working Mixed Working Mixed class class class class

Outcome: Marry Part CON OUT ORIG Responsibility 5.30 5.82 5.30 5.47 $ $ ns Internality 5.41 5.92 5.46 5.45 ns ns ns

CON=congruity; OUT=outcome; ORIG=ongin. CLA=class; REG=region; CXR=class Xregion.

n= (40) 140) (40) (40)

*p< 0.001. tp< 0.01. $p<O.OS.

Northern, unemployed middle class - are similar in terms of both responsibility and internality attribution; and the highly unexpected condition (Southern, middle- class unemployed) is higher on both judgments. The effect of the vignette characters’ social class on responsibility and internality attributions is more noticeable in the Northern than Southern unemployed conditions. This is due to the larger effect of region on attributions for working class actors.

Finally, the characters in the personal relationships domain are judged to be more responsible for their outcomes than in the other domains, and the effects of the congruity and outcome contrasts are weak. However, a post hoc test shows that the characters in the successful cross-class marriage are rated as more responsible and as having played a greater part in the success of their relationship than the other successful couple.

Summary

There are strong effects due to the actor’s class origins. However, there is also an effect of outcome: Respondents give lower responsibility and internality judgments to ‘failure’ outcomes, compared with ‘successful’ ones. Attributions of personal res- ponsibility and internality were related to the types of origin-outcome in similar ways.

456 G. Evans

Importance ratings of the explanations

The means of the importance ratings of explanations by origin-outcome combi- nations and the significance of the contrasts are shown in Table 3.

Occupational attainment

‘Socialization’, ‘social class background’ and ‘opportunities’ are all rated as less important for incongruent outcomes, while the importance of ‘personality’, ‘innate ability’ and ‘problems/skills’ increases. There is also a significant outcome effect on the explanations, with middle class outcomes generating higher importance rat- ings. Only the ‘fate’ explanation fails to follow this pattern - it is given slightly greater importance for working class outcomes. Overall, the class-constraint model accurately predicts the relationships between explanations and types of class origin- outcome combinations. However, there is also an outcome effect, with middle class outcomes resulting in higher importance ratings than working class outcomes for seven out of eight explanations.

Educational achievement

‘Quality of school (opportunities)’, ‘class background’ and ‘personality’ are signifi- cantly related to the congruence manipulation in directions which are consistent with the class-constraint model. ‘Socialization’, ‘opportunities’, ‘personality’, ‘prefer- ences’ and ‘innate ability’ are also related to outcome, with successful pupils receiving higher importance ratings on these explanations. There are also significant origin effects for ‘quality of school’ and ‘preferences’. The combination of congruence, outcome and origin effects results in a pattern of explanation in which the successful working class pupil is not given lower importance ratings than the unsuccessful working class pupil for ‘quality of school’, ‘class background’ and ‘socialization’. However, he is given higher ratings for ‘innate ability’ and ‘personality’, and signifi- cantly lower scores on ‘fate’, than either the unsuccessful working class pupil or the successful middle class pupil. The middle class academic failure is the most unex- pected outcome; ‘quality of school’. ‘personal preference’, ‘social class background’ and ‘socialization’ are considered less important in this case compared to the other conditions. However, there is no noticeable re-focusing of explanation on psychologi- cal factors - even in comparison with the working class academic failure - which leaves this outcome relatively unexplained compared with the others.

Unemployment

‘Social class’, ‘job opportunities’ are rated as being more important for working class unemployment than for middle class unemployment, whereas this emphasis is reversed for ‘personality’ explanations. There are class differences in the importance of ‘socialization’ in the South, but not in the North. ‘Fate’ is clearly seen as more relevant to unemployment in the North than in the South. Post hoc tests indicated that ‘job opportunities’ are seen as uniformly important in three of the vignettes, but in the case of the unemployed middle class character in the South - the least expected origin-outcome combination - there is a significant decline in ratings of their importance. In contrast, the ‘preferences’ explanation is considered less

Class structure and social outcomes 451

Table 3. Importance ratings for explanations by type of vignette

Explanations Means by conditions Significance of effects Origin: Working Middle Working Middle

class class class class

Occupational attainment Outcome: Working class

Socialization 4.76 3.69 Opportunities 5.76 4.13 Innate ability 4.24 4.27 Fate 2.45 2.90 Problems/skills 3.00 4.13 Social class 4.26 2.46

Preferences 4.61 4.90 Educational achievement

Socialization 5.00 4.44 Opportunities 4.95 3.98 Innate ability 4.85 4.90 Fate 2.23 2.07 Problems/skills 4.87 4.83 Social class 4.21 2.71 Personality 5.31 5.51 Preferences 5.87 4.78 Unemployment

Socialization 3.23 3.16 Opportunities 6.1 1 5.80 Ability 4.61 4.67 Fate 3.16 2.56 Personal

problems 3.30 3.16 Social class 3.41 2.46 Personality 4.00 5.23 Preferences 3.05 3.87 Personal relationships

Personality 4.84 5.44

Fail

Region: North

Working Mixed class class

Socialization 5.19 3.43 Circumstances 5.32 3.62 Physical

attraction 5.03 5.79 Fate 2.22 3.21 Problems/

attraction 4.73 5.31 Social class 5.05 2.41 Personality 5.76 6.29 Preferences 5.43 5.72

n= (40) (40)

Outcome: Marry

Middle class 4.41 5.73 4.95 6.08 5.56 4.95 2.23 1.98 5.23 4.53 2.56 5.13 6.10 5.65 5.62 5.08

Succeed 5.45 5.65 5.36 5.40 5.58 5.20 1.78 2.28 4.78 4.70 2.80 4.70 6.08 5.35 6.20 5.85

South 3.76 2.85 5.97 4.83 4.87 4.43 2.05 2.28

3.75 3.13 3.87 2.53 4.14 5.03 4.00 3.80

Working Mixed class class

Part 2.51 5.38 4.47 5.13

4.14 2.85 3.43 2.30

4.91 4.50 2.75 5.13 5.43 4.73 4.98 5.20 (40) (40)

CON * * j: ns * * t ns

CON ns j: j: ns ns

$ ns

CLA ns t ns ns

ns

*

* *

ns

CON * *

* *

ns

t ns

*

OUT t t t $

$ t %

OUT t

3 ns ns ns j: t

REG ns $ ns t ns ns ns ns

*

*

OUT ns ns

* ns

ns ns

3 *

ORIG ns ns ns ns ns ns ns ns

ORlG

t ns ns ns ns ns t

CXR t ns ns ns

ns ns ns ns

ns

ORIG $ ns

t ns

ns ns ns t

CON=congruity; OUT=outcome; ORIG=origin. CLA =class; REG= region; CXR =class X region. *p<O.OOl. tp< 0.01. $p < 0.05.

458 G. Evans

important than the others for the most expected outcome - the unemployed, working class character in the North.

Personal relationships

Differences between the two cross-class relationships and the within-class relation- ships clearly indicate strong class-related expectations in this domain. Outcomes congruent with class origins result in higher ratings of importance for social structural explanations and lower ratings for individualistic explanations than do unexpected outcomes. Outcomes incongruent with class origins are more strongly associated with explanations involving ‘personality’, ‘fate’ and ‘physical attraction’. Also, there are outcome effects, most noticeably in the case of ‘personality’: With successful relationships receiving higher ratings than unsuccessful ones. The results are thus consistent with the class-constraint model, but with evidence also that individualistic explanations are rated as more important for successful outcomes.

The patterning of the underlying dimensions of explanations

The pattern of relations between the explanations and the vignette manipulations can be more clearly summarized by reducing the eight explanations for each outcome to a set of dimensions. Through such an analysis we can test whether there are underlying dimensions of structural and individualistic explanation, and whether these covary with the congruence of origins and outcomes in the manner predicted in hypothesis 4. To assess the structure of the explanations we conducted a principal components analysis3. The dimensions were rotated using the Varimax technique (Oblimin and Quartimax rotations produced similar patterns). The structure of expla- nations in each of the domains was similar, with three components corresponding to social structural, individualistic and, less clearly, fatalistic dimensions. In all domains the items load reasonably clearly onto single components (details available from the author). In each domain the structural and individualistic components were distinct: In no case did an item which loaded onto an individualistic component in one domain load onto the structural component in another. Only in two instances did an explanation load similarly on more than one compartment: For occupational attainment, ‘innate ability’ loaded onto both the individualistic and fate components, and for educational achievement, ‘problems/skills’ loaded onto the fate component.

The principal components analysis also allows us to test hypothesis 5 - which proposed that fatalistic explanations are distinct from the two main dimensions of social structural and individualistic explanation - by examining the communalities of the items after extracting the first two principal components. For all domains, the fate item has clearly the lowest communality (details available on request), whereas all other explanations have quite high communalities. Thus although ratings of fate explanations (‘it’s just one of those things’) were associated with those of ‘job opportu- nities’ in the unemployment vignette; with ‘personal problems/skills’ in the edu-

Initially the structure of the explanation items for working class and middle class outcomes was assessed separately, thus enabling a check on variation in the inter-item structure due to slight differences in the item wordings for the two types of outcome. The structure of responses was found to be very similar for both types of outcome; consequently the main principal components analysis was conducted using the whole sample.

Class structure and social outcomes 459

cational vignette; and with special characteristics of the couples’ relationship in the personal relationships vignette, the fatalistic explanation is, on the whole, distinct from the dimensions onto which the majority of the social structural and individualis- tic explanations load.

We then used the principal components (standardized) as indicators of the social structural and individualistic dimensions of explanation4. We also estimated the relative importance of these dimensions by computing a variable (Soc-Ind) which subtracts the individualistic component from the social structural component. On average Soc-Ind is correlated 0.71 with the structural dimension and -0.71 with the individualistic dimension.

It can be seen from Table 4 that, on the whole, the results are consistent with hypothesis 4. For the domains of occupational attainment, educational achievement

Table 4. Social structural and individualistic dimensions of explanation by type of vignette

Means by conditions

class class class class

Significance of effects Origin: Working Middle Working Middle

Occupational attainment

SOC. struct. 0.30 -0.39 -0.72 0.76 t ns Individual -0.67 0.59 -0.01 0.07 ns Soc-Ind 0.94 -0.99 -0.70 0.70 ns ns Educational achievement

SOC. struct. 0.12 -0.12 -0.75 0.53 ns Individual -0.21 0.83 -0.15 -0.17 * f * Soc-Ind. 0.33 -0.41 -0.60 0.70 ns ns Unemployment

SOC. struct. -0.03 -0.22 0.50 -0.27 t ns ns Individual -0.26 0.21 0.02 0.04 ns ns ns Fate/

opportunities 0.50 -0.05 0.09 -0.55 t ns Soc-Ind. 0.50 0.27 -0.44 -0.34 ns ns Personal relationships

Outcome: Working class Middle class CON OUT ORIG * * * *

Fail Succeed CON OUT ORIG * * *

Region: North South CLA REG CXR

* *

Working Mixed Working Mixed class class class class

Outcome: Marry Part CON OUT ORIG * SOC. struct. 0.61 -0.71 -0.59 0.65 ns ns Individual 0.36 -0.33 0.57 -0.60 ns ns Soc-Ind. 0.55 -0.65 -0.90 0.96 ns

* * *

n = (40) (40) (40) (40) CON=congruity; OUT =outcome; ORIG=origin. CLA=class; REG=region; CXR=class xregion, *p < 0.001. tp<O.Ol. $p<0.05.

The standardized principal component scores were used in preference to unweighted unit Likert scales, as the component weights provide a better estimate of the relationship between the underlying concepts and the items used as indicators. The use of unweighted or weighted scales makes no substantive difference to the findings.

460 G. Evans

and personal relationships, the structural dimension is emphasized significantly more when explaining outcomes congruent with class origins; and, in the case of occupatio- nal attainment and educational achievement, the individualistic dimension is pre- ferred for outcomes incongruent with class origins. In all three of these domains the individualistic dimension is also significantly associated with outcome, with suc- cessful outcomes receiving higher scores than unsuccessful ones. Finally, Soc-Ind is strongly related to the congruence between origins and outcomes - but not to outcome - for both occupational attainment and educational achievement, thus indicating that the relative preference for the two dimensions is not significantly affected by outcome.

The unemployment vignette is a little different from the others, as the ‘Qob) oppor- tunities’ explanation forms a separate dimension with ‘fate’, whereas in all other domains it forms part of the social structural dimension, along with ‘class’ and ‘socialization’. This distinction is important, as Table 4 shows that the fate/opportuni- ties dimension is related to both class and region, whereas the social structural dimen- sion and Soc-Ind are related only to social class. Clearly, region is seen as an important influence on job opportunities5. Also, there is some evidence that the effect of class on the social structural dimension appears to be conditioned by region, with stronger class effects in the South than in theNorth ( F = 3 . 6 1 , ~ < 0.06). Nevertheless, these details should not obscure the main finding, which is of a clear change in emphasis from individualistic to social structural explanations when accounting for working class rather than middle class unemployment.

DISCUSSION

We have compared two models of beliefs about the social processes involved in the achievement of various goals. One, the class-constraint model, represents a cogni- tive model in which individuals’ life-chances are thought to be substantially affected by their positions in class structures. The second - meritocratic -model, represents an individualistic view of an open society, where individuals dictate their own fates. Clearly, the meritocratic model fails to account for the results: the categorization of a character as either working or middle class not only had a considerable effect upon the way in which his outcomes were explained, and on the degree of personal responsibility attributed to him, it was strongly associated with surprise and future expectations. The class-constraint model, it appears, is a more adequate account of the pattern of explanations and judgments. Nevertheless, there were two deviations from its predictions.

First, the ratings of surprise show that educational success and upward mobility are less surprising to respondents than are educational failure and downward mobility; the upwardly mobile man is also seen as unlikely to change his class position. This is, it should be noted, what would be expected from research into real-world rates of mobility and educational attainment, which show that far more people are upwardly mobile than are downwardly mobile and that working class are more likely to succeed than middle class pupils are to fail (Goldthorpe, 1987; Heath, Halsey and Ridge, 1980). Even so, the other indicators of the existence of a class-constraint

Although there has been some reduction in regional disparities in unemployment during the latest recession, which suggests that people’s beliefs regarding North-South differences in this respect may also have changed.

CIuss structure and social outcomes 46 1

model were as predicted. Thus, despite the lack of surprise at the success of the working class actor his success was still explained in highly individualistic ways, indicating that success for people from working class origins is believed to be deter- mined more by their personal characteristics than it is for those from middle class origins. Also, the class-constraint model is supported by the high level of personal responsibility and internality attributed to the upwardly mobile and educationally successful characters. So although the low surprise rating indicates a belief that upward mobility is not as rare as downward mobility, it is still seen as requiring special effort and ability on the part of the individual.

Secondly, there were significant outcome effects in all three domains in which success of outcome was manipulated. So that successful (middle class) outcomes were explained more individualistically than unsuccessful (working class) outcomes. A similar effect was also found for ratings of internality and responsibility. This indicates that we need to take into account differences in explanations of success and failure in order to adequately represent the patterns of explanation. To some degree this effect may result from the small differences in the framing of the explana- tions for the different outcomes. Also, it would be surprising if explanations referring to ‘preferences’ did not differ for outcomes that differ in their general desirability. Nevertheless, before reading too much into this finding, it should be noted that for occupational attainment and educational achievement, both individualistic and structural dimensions were given more importance for successful outcomes. There- fore, when the relative importance of explanations was examined (Soc-Ind), the out- come effects became insignificant. Only for personal relationships did successful and unsuccessful outcomes produce differences in Soc-Ind ratings. For the other domains the relative emphasis on individualistic versus structural explanations is related speci- fically to the congruence between class origins and outcomes.

We further proposed that it was useful to distinguish between different types of ‘external’ explanations (hypotheses 5 and 6). Across all domains except unemploy- ment the fate item was found to be somewhat distinct from the main dimensions of explanation and - with the exception of the relationships vignette - it does not covary with the congruence of outcomes with class origins. On the whole therefore hypotheses 5 and 6 are supported. It is perhaps possible that respondents were uncer- tain of the meaning of the ‘it’s just one of those things’ item. However, the low importance attributed to fate is consistent with research reported elsewhere (i.e. Furnham, 1982, 1988); and in other studies we have used the terms ’luck or chance’ instead, and found more or less the same results (Evans, 1988, forthcoming).

We conclude that, in general, the ‘class-constraint model’ is a plausible account of the respondents’ causal understanding of social outcomes. On the whole, the findings are consistent with the claim that the categorization of a vignette character as working or middle class activates different outcome expectations, and correspon- dence with these expectations affects the emphasis given to social structural and individualistic explanations. There is some variation in the structure of explanations in different domains, and in the strength of association between the explanations and the different origin-outcome combinations, but on the whole the results support the proposition that explanations occur within a framework of assumptions about social structure and its effects on life-chances. They do not support the theory that people in western societies have individualistic beliefs about human nature, and that such individualistic understandings of social achievement serve to justify conti-

462 G. Evans

nued inequalities. Moreover, they do not confirm the claims of sociologists who have argued that class-based ‘images of society’ are to be found primarily among working class rather than middle class respondents (see various chapters in Bulmer, 1975).

Finally, two possible shortcomings of the study are (i) the limited range of the explanations presented to the respondents and (ii) the use of a non-representative, middle class sample. However, we believe that these limitations are not too great a problem. First, although only eight explanations were used, they represent a con- siderable spectrum of interpretations covering the range of explanations generated by respondents in the pilot studies. Therefore it is unlikely that giving respondents a greater number of explanations to consider would improve the validity of the study. In addition, pre-testing of the questionnaires did not indicate noticeable dissa- tisfaction from respondents concerning the range or form of explanations with which they were presented. Secondly, although the sample used in this study is not represen- tative of the population as a whole, we have repeated some of the analysis of explana- tions of occupational attainment with a representative national sample and found a very similar pattern of results (Evans, 1988, forthcoming).

CONCLUSION

Despite recent talk of ‘a classless society’, it is clear that expectations associated with cognitive models of social class have a pervasive influence on people’s under- standing of who gets where in society, who becomes unemployed, and even upon who marries whom. These findings were obtained without explicit reference to social class in the vignettes that respondents were presented with: Thus the inferences drawn by respondents were not a result of explicit cueing by the researcher. Consequently, if they did not have both class-constraint models of the causes of people’s outcomes, and stereotypes of different class members, these relationships would not have been obtained. Of course, it should be emphasized that the patterns we have discovered are a matter of degree: we are proposing a model which shows that people take class into account when judging outcomes, not one in which class is seen as the only determinant of explanations. Certainly, individualistic explanations are given considerable importance for many outcomes. Even in the case of long-term unemploy- ment in the North of England, for example, there was some degree of personal choice attributed to the character in the vignette. Nevertheless, the patterning of explanations produced by the manipulations of class origins and outcomes shows that, at least in Britain, outcomes in several important areas of social activity are explained in a way which indicates that people have cognitive models in which class origins are significant influences on individuals’ life-chances. Further work in this area could now usefully be addressed to examining the influences on the social distri- bution of such models, and their implications for perceptions of social justice.

ACKNOWLEDGEMENTS

I would like to thank Michael Argyle and Mansur Lalljee for helpful comments on earlier versions of this paper.

Class structure amd social outcomes 463

REFERENCES

Bulmer, M. (Ed.) (1975). Working Class Images of Society, Routledge, London. Cochrane, R. (1983). The Social Creation of Mental Illness, Longman, Harwell. Evans, G. A. (1988). ‘Causal explanation, perceived efficacy and social class’. Unpublished

Evans, G. A. (Forthcoming). Class, Merit and Justice: Beliefs about Social Stratijication and

Evans, G. A., Heath, A. F. and Payne, C. (1991). ‘Modelling the ClassParty relationship

Feagin, J. (1975). Subordinating the Poor, Prentice Hall, New York. Furnham, A. (1982). ‘Explanations for unemployment in Britain’, European Journal of Social

Furnham, A. (1988). Lay Theories; Everyday Understanding of Problems in the Social Science,

Goldthorpe, J. H. [in collaboration with Llewellyn, C. and Payne, C.] (1987). Social Mobility

Heath, A. F., Halsey, A. H. and Ridge, J. (1980). Origins and Destinations, Clarendon Press,

Heath, A. (1981). Social Mobility, Fontana, Glasgow. Heath, A. and Clifford, P. (1990). ‘Class inequalities in education in the twentieth century’,

Journal of the Royal Statistical Society, Series A , 153: 1-16. Hewstone, M. (1989). Causal Attribution, Blackwell, Oxford. Huber, J. and Form, W. (1973). Income and Ideology, The Free Press, New York. Ichheiser, G. (1949). ‘Misunderstandings in human relations; a study in false social perception’,

Jaspars, J. M. F., Fincham, F. and Hewstone, M. (Eds) (1983). Attribution Theory, Academic

Kluegel, J. and Smith, E. (1981). ‘Beliefs about stratification’, Annual Review of Sociology,

Kluegel, J. and Smith, E. (1986). Beliefs About Equality, Aldine De Gruyter, New York. Lalljee, M. G. and Abelson, R. (1983). ‘The organization of explanations’. In: Hewstone,

M. (Ed.) Attribution Theory, Blackwell, Oxford. Lalljee, M. G., Lamb, R., Furnham, A. and Jaspars, J. M. F. (1984). ‘Explanations and

information search: Inductive and hypothesis testing approaches to arriving at an explana- tion’, British Journalof Social Psychology, 23: 201-212.

Laumann, E. (1966). Prestige and Association in an Urban Community, Bobbs-Merrill, Indiana- polis.

Lukes, S. (1973). Individualism, Blackwell, Oxford. Marshall, G., Newby, H., Rose, D. and Vogler, C. (1988). Social Class in Modern Britain,

McRae, S. (I 986). Cross-Class Families, Oxford University Press, Oxford. Moscovici, S. (1984). ‘Social representations’. In: Farr, R. M. and Moscovici, S. (Eds.) Social

Nisbett, R. E. and Ross, L. (1980). Human Inference, Prentice Hall, New York. Ossowski, S. (1963). Class Structure in the Social Consciousness, Free Press, New York. Read, S. (1987). ‘Constructing causal scenarios: A knowledge structure approach to causal

Reid, I. (1989). Social Class Diflerences in Britain, 3rd edn, Fontana, Glasgow. Rumelhart, D. E. (1984). ‘Schemata: the building blocks of cognition’. In: Wyer, R. S. and

Srull, T. K. (Eds) The Handbook of Social Cognition, Vol. 1, Academic Press, New York. Schank, R. and Abelson, R. (1977). Scripts, Plans, Goals and Understanding, Erlbaum, Hillsdale

NJ. Schaufeli, W. (1988). ‘Perceiving the causes of unemployment: An evaluation of the causal

dimensions scale’, Journal of Personality and Social Psychology, 54: 347-356. Weiner, B. (1985). “‘Spontaneous” causal thinking’, Psychological Bulletin, 97: 7&84.

Doctoral thesis, University of Oxford.

Equality of Opportunity in Britain, Cambridge University Press, Cambridge.

1964-1987’, Electoral Studies, 10: 99-1 17.

Psychology, 12: 335-352.

Pergamon, Oxford.

and Class Structure in Modern Britain, 2nd edn, Clarendon Press, Oxford.

Oxford.

American Journal of Sociology, 55: 1-70.

Press, London.

7: 29-56.

Hutchinson, London.

Representations, Cambridge University Press, Cambridge.

reasoning’, Journal of Personality and Social Psychology, 52: 288-302.

464 G. Evans

Wilkinson, R. (Ed.) (1986). Class and Health, Tavistock, London. Wyer, R. S. and Srull, T. K. (Eds) (1984). The Handbook of Social Cognition, Vols 1-3,

Academic Press. New York.