Make your own free website on Tripod.com
**************************************************************************************

Australian and New Zealand Journal of Sociology, 1971, Vol. 7 (April): 58-64

The Questionnaire Measurement of Social Class



JOHN J. RAY

SOCIAL class has been measured in two main ways. The first requires a study of the community structure in which the subject is placed (Lloyd Warner, 1960). The second seeks details by a standard questionnaire which will enable any respondent to be given a class 'score' without referring to any surrounding community (except perhaps the nation as a whole). It is the latter which concerns us here. Yet more narrowly it is vertical divisions only which are focused on. Horizontal divisions are obviously important but fall outside the scope of this paper.

To enable class placement, two different types of information may be sought -- called 'subjective' and 'objective' (Alford, 1962). In the subjective approach one simply asks the person what class he thinks he is by giving him a class schema and asking him to identify his position in it. This approach, as generally used, has the difficulty of taking little account of the fact that the subject himself may customarily use an explicit schema very different from the one the sociologist presents to him (but note Davies, 1967). In the objective approach, one obtains information about several properties said to be critical for social class and simply sums them. Alford (1962) and Broom et al. (1968) sum scores on occupation, education and income. This approach fails to take into account the full range of factors which may affect class. Things like manner of speech are very difficult to measure by questionnaire. Although it is possible that such omissions might not be important, they will always need to be kept in mind.

Alford believes that the objective method is the best predictor of theoretically related variables, and, of the three elements usually employed, occupation alone is the best. This is the first hypothesis that will be tested here: that occupation is a better predictor than class self-assignment.

Alford also believes that a simple dichotomy of occupation into manual versus non-manual is equally as powerful as a rating of occupations on a multilevel status scale. This second hypothesis will be examined using Congalton's (1969) 'Scale of Occupational Status in Australia'.

The measure of 'goodness' in a class index will be taken to be its power to predict other theoretically related variables-variables of both a social and psychological nature. In order to make the test more decisive, both the subjective and objective measurements will first be 'powered up' by the use of psychometric scaling techniques.

METHOD

DATA COLLECTION

A questionnaire was administered by students in a second-year psychology course at Macquarie University. Each student was given three copies of the schedule and told to seek subjects 'of as low a social class as possible'. By these instructions it was hoped to counterbalance the usual tendency of surveys so administered to tap a largely middle and upper class population. The second usual tendency of this type of sampling to tap people of similar age level to the students (Christie, 1956) was guarded against by two requirements:

1. That no students may be included.

2. That the first two questionnaires be given to one person in the 15-25 age group and one person from the 40-55 age group.

The third questionnaire, if administered, was permitted to be from either age group.

The questions upon which this report is based are given in Table 1. Also included were several Likert-type attitude scales.

TABLE 1

Summary of social class questions administered in this survey

Q. 1 Occupationa! information-scored on Congalton's (1969) 7-point scale.
Q. 2 Level of education (17-point scale).
Q. 3 Number of classes respondent believes to exist in Australia.
Q. 4-8 (Preamble) I have here a sheet on which are listed several different ways of describing the classes that exist in Australia. Would you indicate where you think you fall into each of them by circling the appropriate number?
Q. 4. 4. Upper
.........3. Upper middle
.........2. Lower middle
.........1. Lower
Q. 5. 2. Upper
.........1. Lower
Q. 6. 2. Middle
.........1. Working
Q. 7. 3. Upper
.........2. Middle
.........1. Lower
Q. 8..... 7. Upper upper
.............6. Lower upper
.............5. Upper middle
.............4. Middle middle
.............3. Lower middle
.............2. Upper lower
.............1. Lower lower
Q. 9-11 (Preamble) People often express different ideas about what influences the class a person belongs in. What things do you think influence social class?


------------------------------------------------------------------------------------------------------------

Three things that are often suggested as influencing class membership are: a. Income b. Occupation c. Education

These three things themselves may obviously often go together but what we would like you to do is indicate how important you think each of these things are in making up what class a person belongs in. Assume that they are the only three and allocate percentages out of 100.

Q. 9 Class is : ____________% Income
Q. 10 ___________________% Occupation
Q. 11 ___________________% Education
___________________% Total

Q. 12 Disposable income (14-point scale).
Q. 13 Political party preference.

DATA ANALYSIS

As a preliminary, all variables except No. 3 were multiplied by a scaling constant in order that each might have a uniform maximum value of seven (e.g. answers to question 4 were multiplied by 1.75 and answers to question 2 were divided by 2.43). This step facilitates comparisons without discarding any information.

To compute the new subjective measure, each subject's scores on questions 4-8 were summed. The correlation of each question with the total score was computed and adjusted for spurious overlap (Guilford, 1954). The reliability coefficient 'alpha' for the 5-item 'scale' was also computed (Cronbach, 1951). Only one of the schemas offered in questions 4-8 included the term 'working class'. Although 'working class' is a common voluntary self-identification, it was not given prominence because it was felt to be at least partly an anti-stratification label. In other words, a person unwilling to see himself in a hierarchical system might choose the 'working class' label because: 'We all work, don't we?' This view of how the 'working class' label is used was formed in a study of New South Wales state public service clerical workers during 1968, wherein fifty-two out of sixty-two respondents so described themselves.

To compute the new objective measure, the mean scores on questions 9-11 were found and used to weight the answers to questions 12, 1 and 2 respectively (e.g. if the mean percentage importance of income is given as 40 per cent, scores on question 12 are multiplied by 0-4). After this adjustment, scores on questions 12, 1 and 2 are then summed and scaled as before.

The correlation between the subjective and objective indices was then found. To compute the final composite index, the scores on the subjective index were divided by 5, the subjective and objective indices were summed and the result divided by 2.

A second form of the objective index was also produced. In this form, the subject's scores on questions 12, 1 and 2 were weighted by his own answers to questions 9, 10 and 11. This procedure meets the argument that we might in many cases be more interested in a subject's own view of things (because this is what will influence him) rather than the mean or average view.

Alford's index was produced by reclassifying the answers to question 1. 'Manual' was scored as 3-5 and 'Non-Manual' as 7-0. Service occupations such as police, soldier, etc., were scored 3-5.

A grand intercorrelation matrix of all variables was then produced.

RESULTS

The means and standard deviations of the social class variables are given in Table 2. They are intended to act as norms for the weights and indices proposed. The correlations in Table 3 include a selection of attitude variables to which the social class index might be expected to predict. This enables the psychological usefulness of the various measures to be compared.

The scale characteristics of the subjective and objective indices appear in Table 4. Note the high reliabilities obtained. The correlations between all the social variables appear in Table 4.

TABLE 2 Means and S.Ds. of social class variables and indices. (All scores on Congalton's scale were 'reversed' by subtracting from 8)

Variable...........................................................................Mean.........S. D.

Occupation (Q. 1)..............................................................2.8.............1.3
Education (Q. 2 )...............................................................3.1..............0.9
No classes (Q. 3 )..............................................................3.7.............1.5
Subjective class (Q. 4)......................................................4.1..............1.2
Subjective class (Q. 5)......................................................5.0..............1.7
Subjective class (Q. 6 ).....................................................5.2..............1.7
Subjective class (Q. 7 ).....................................................4.4..............1.1
Subjective class (Q. 8)......................................................3.8..............1.2
Percentage income (Q. 9 )...............................................40.4............17.6
Percentage occupation (Q. 10)........................................29.4............13.5
Percentage education (Q. 11)..........................................30.2............15.0
Income (Q. 12)..................................................................2.1..............1.3
Total subjective index.......................................................4.5..............1.1
Unweighted objective score..............................................2.7..............0.9
Self-weighted objective score...........................................2.7..............1.0
Mean-weighted objective score........................................2.6..............0.9
Subjective-unweighted objective composite.....................3.6..............0.8
Subjective-meanweighted objective composite................3.6..............0.8
Manual vs. non-manual.....................................................5.0..............1.7
Manual-non-manual subjective composite........................4.7..............1.2


TABLE 3

Correlations of class indices with other variables

Column 10 is the sum of entries in Columns 2-9 with the sign disregarded. Def = Deference; PC = Politican conservatism; SC = Social conservatism; AC = Aesthetic conservatism; EC = Economic conservatism; RC = Religious conservatism; MC = Moral conservatism; AA = Attitude to authority; SUM = Summed correlations. Decimal points are omitted.

Predictor............................................Def.......PC.....SC.....AC.....EC.......RC......MC

Occupation........................................227.....-058...-108....060.....241.....024.....-013
Education..........................................140.....-135...-231....072.....136....-040.....-139
Income..............................................132......093....197....122......255...-048......043
Question 8.........................................335......244....123....103......311....198.....-012
Subjective index................................307......225....071....132......306....228......057
Unweighted objective index...............219.....029....-036...112.......284...-026....-034
Self-weighted objective index............222......023....043....125......283...-004.....-008
Mean-weighted objective index..........213......013....003....117......288....030....-024
Unweighted composite index.............324......136....028....149......358....140.....020
Mean-weighted composite index.......323......145....046....152......363....137......025
Manual vs. non-manual occup..........181.......030...087.....150......179....081....-054
Manual-non-manual composite..........285......133....030.....176......283....172...-012

TABLE 3 Continued:

Predictor...........................................AA........SUM

Occupation.......................................-020.......751
Education.........................................-124.......1017
Income..............................................173.......1063
Question 8.........................................180.......1506
Subjective index................................168.......1494
Unweighted objective index...............060........800
Self-weighted objective index............116........824
Mean-weighted objective index..........087........775
Unweighted composite index.............145......1300
Mean-weighted composite index........160......1351
Manual vs. non-manual occup.............047.......809
Manual-non-manual composite............117.....1208


TABLE 4 Intercorrelations of the class indices. n = 203 but for correlations with conservatism of political party preferences, n= 154

.............................Ed......Inc.....Q.8....Subj. U.Obj....S.Obj....M.Obj....U.Comp

Occ. ( Cong. ) .......47......43.......25.....35.........84........76.........80........68
Education........................22.......22.....33.........68........63........67.........58
Income......................................13......17........76.........77........82.........52
Q. 8.....................................................86........26.........29........25.........71
Subjective index..............................................36........39........34..........87
Unwt. objective............................................................95.........99.........78
Selfwt. objective.......................................................................95.........77
Meanwt. objective..................................................................................76


Table 4 Continued:

.................................U.Comp.....M.Comp....Man.N.M.....MNM.Comp....Pol Pref

Occ. ( Cong. ) .............68..............67..............56..............59..................28
Education....................58..............57..............54..............56..................19
Income........................52..............56..............28..............29..................04
Q. 8.............................71..............71..............16..............54..................19
Subjective index..........87..............86..............28..............70..................25
Unwt. objective...........78..............78..............59..............62..................23
Selfwt. objective.........77..............95..............52..............57..................18
Meanwt. objective.......76..............77..............56..............59..................21
Unwt. composite...........................99..............51..............80..................29
Meanwt. composite.........................................49..............79..................28
Manual vs. non-man..........................................................88..................31
M-N.M. composite...................................................................................35
Political preference


TABLE 5 Scaling characteristics

I. The Subjective Class Scale

Item................Correlation with Total Score

Q. 4....................80
Q. 5....................70
Q. 6....................61
Q. 7....................63
Q. 8....................79

Reliability ('alpha') = .86


II. The Objective Class Scale

Item.................Correlation with Total Score

Income...............40
Occupation.........56
Education...........41

Reliability ('alpha') = .64


DISCUSSION

The idea of adding scores on social variables to produce an index of class is not new. It could not, however, be accepted on a priori grounds that each variable added was of equal importance for social class. The ipsatised weights found here do present evidence on the question. Not only have they been shown to be unnecessary but their contribution is, in fact, slightly negative.

The idea of an additive index for subjective class placement does, however, appear to be new. Its advantages are in the higher reliability and greater generality of measurement that the Likert scaling method is designed to achieve. It is also one of the few ways in which account might be taken of the fact that different respondents use different schemas. Presumably all subjects would find at least one of the schemas congenial and hence each subject would be giving maximum information on each occasion. Some subjects would not be favoured above others because of a fortuitous correspondence between the experimenter's schema and the subjects' schema.

The third step of adding and averaging the subjective and objective indices to produce one composite index is a logical further application of the basic Likert dictum that by adding scores on two correlated variables we produce a better measure of what is general to the two. That this has been achieved can best be seen if it is borne in mind that the 'alpha' statistic is at once an index of success in measuring a general concept and also a very useful and accurate estimate of the test-retest reliability obtainable with that scale or index.

Given these measurement refinements it is all the more notable that a single simple class self-assignment question is approximately twice as good a predictor as any single, composite, or weighted objective measure (see column 10 of Table 3). An even more impressive difference would be shown if the correlations were squared before adding. (This is the normal procedure to find the 'variance explained' by a correlation statistic.) For the sake of simplicity, however, the table uses raw correlations only. The results are, in any event, a clear disconfirmation of Alford's first hypothesis.

Conversely, Table 3 shows Alford's second hypothesis to be just as strikingly confirmed. The dichotomous system of scoring occupation is in fact slightly better than the multilevel status scale. Since any system of dichotomous scoring has inherently less power to predict than a multilevel version of the same thing, this must mean that the division proposed (which cuts across Congalton's status division) is inherently more meaningful.

When these findings are rechecked in terms of their power to predict other social variables (Table 4), slightly different findings emerge. As might be expected, occupation is a better predictor of education and income than is question 8. Education, occupation, and income are known to be causally very closely related. The issue between Alford's index and Congalton's index is equivocal. Alford's predicts education better but Congalton's predicts income and subjective class better. The matter of closest concern to Alford, of course, is political party preference, and, in this field, both his hypotheses are in fact clearly supported. As in the findings with psychological data, the subjective index is still a slightly better predictor than any version of the objective index but Alford's index is better again. Best of all (r = .355) is the combination of the two strongest indices provided by summing together Alford's index and the subjective index.

In conclusion, it is clear that the best all-round predictor of psychological and social data is the (summated) subjective index. It is almost the same as question 8 in its ability to predict class related psychological variables but is markedly superior to question 8 in the field of sociological prediction. Although not the very best predictor in either field, it is the only index which is efficient in both.

SUMMARY

What has been indicated above then, in an admittedly preliminary way, is in what areas the various possible indices of class are strong predictors among Australians. Any general statement about one index of class being superior to another is shown to be misleading. Given the alternatives he considered, Alford's judgment about the most suitable predictor of voting is vindicated. It is of importance to know however that this superiority cannot be generalised to other prediction tasks. Where generalisability of prediction is however desired for either theoretical or practical reasons, a new index that is suitable for the task has been provided (the summated subjective index). More sophisticated measurement even offers improved measurement in Alford's own prediction task, i.e. the best voting predictor of all is shown to be the composite of subjective score and manual-non-manual score.

The third measurement innovation (weighting the components of the objective index) was shown to be not worthwhile. It is however an advance to know this rather than having to assume it a priori.

REFERENCES

{Articles below by J.J. Ray can generally be accessed simply by clicking on the name of the article. I am however also gradually putting online a lot of abstracts, extracts and summaries from older articles by other authors so if an article not highlighted below seems of particular interest, clicking here or here might just save you a trip to the library}

ALFORD, R. R. (1962) 'A suggested index of the association of social class and voting.' Public Opinion Quarterly, 26: 417-25.

BROOM, LEONARD, F. LANCASTER JONES and JERZY ZUBRZYCKI (1968) 'Social stratification in Australia'. Pp. 212-33 in J. A. Jackson (ed.), Social Stratification. Sociological Studies 1. Cambridge University Press.

CHRISTIE, R. (1956) 'Eysenck's treatment of the personality of Communists'. Psychological Bulletin, 53: 411-38.

CONGALTON, A. A. (1969) Status and Prestige in Australia. Melbourne: F. W. Cheshire.

CRONBACH, L. J. (1951) 'Coefficient alpha and the internal structure of tests.' Psychornetrika, 16: 297-334.

DAVIES, A. F. (1967) Images of Class. Sydney: University Press.

GUILFORD, J. P. (1954) Psychometric Methods. 2nd ed. New York: McGraw-Hill.

LLOYD WARNER, W. (1960) Social Class in America. New York: Harper.




Go to Index page for this site

Go to John Ray's "Tongue Tied" blog (Backup here or here)
Go to John Ray's "Dissecting Leftism" blog (Backup here or here)
Go to John Ray's "Australian Politics" blog (Backup here or here)
Go to John Ray's "Gun Watch" blog (Backup here or here)
Go to John Ray's "Education Watch" blog (Backup here or here)
Go to John Ray's "Socialized Medicine" blog (Backup here or here)
Go to John Ray's "Political Correctness Watch" blog (Backup here or here)
Go to John Ray's "Greenie Watch" blog (Backup here or here)
Go to John Ray's "Food & Health Skeptic" blog (Backup here)
Go to John Ray's "Leftists as Elitists" blog (Not now regularly updated -- Backup here)
Go to John Ray's "Marx & Engels in their own words" blog (Not now regularly updated -- Backup here)
Go to John Ray's "A scripture blog" (Not now regularly updated -- Backup here)
Go to John Ray's recipe blog (Not now regularly updated -- Backup here or here)

Go to John Ray's Main academic menu
Go to Menu of recent writings
Go to John Ray's basic home page
Go to John Ray's pictorial Home Page (Backup here)
Go to Selected pictures from John Ray's blogs (Backup here)
Go to Another picture page (Best with broadband)