COR 689 A Note on Factor Weighted SES Scores The respondent's social class is operationalized in this study as a weighted combination of four items; namely, (1) father's education, (2) mother's education, (3) father's occupation, and (4) average yearly income of the parents during the period from 1957 to 1960. In the 1957 statewide questionnaire survey, the respondents were asked to check the following item: "Education of father and mother (check highest level attained)" Father Mother High School: did not attend attended graduated from Trade or Business School: attended College: attended graduated from has Master's or Ph.D. degree Do not know: This information obtained from this item was coded as shown below: Code Category 1 Elementary school, None, N.A. 2 Some high school education 3 Completed high school, or attended trade or business school 4 Some college education 5 Completed college 6 Some graduate work Most of the data on father's occupation and average yearly income of the parents were obtained from the tax returns filed by the parents of the respondents and, in some cases, from the tax rolls in the Wisconsin State Department of Taxation. The information on father's occupation reported in the tax returns was very descriptive in terms of the type of industry or firm in which the respondent's father was employed, as well as in terms of his occupational rank. Whenever a general category of occupation was mentioned in a tax return, a detailed occupational title was determined on the basis of other information provided in the tax return. Consequently, father's occupation was coded into about 500 occupational titles given in the 1950 classification of occupations made by the Bureau of the Census. In some cases, where the tax returns filed by the respondents' parents could not be located in the tax department, information on father's occupation, as reported by the respondents in the 1957 statewide questionnaire survey, was utilized. The respondents were asked to check the following item: "My father is engaged in the type of occupation checked in the left hand column below" Office work (cashier, clerk, secretary, bookkeeper, etc.) Professional (doctor, lawyer, minister, teacher, etc.) Executive (manages large business, industry, firm) Factory worker (laborer, janitor, farm hand, etc.) Salesman (insurance, real estate, auto, store, etc.) Owns, rents, manages small business (store, station, newspaper, cafþ, etc.) Owns, rents, manages farm Other occupation (be specific) It has become clear that the information obtained from the questionnaire could be coded into only gross categories of father's occupation. Thus, the information on father's occupation obtained from the tax returns was much more detailed and consistently reliable than the similar information obtained from the questionnaires. But in about 10 percent of the cases for which tax returns filed by the respondents' fathers could not be located in the Wisconsin State Taxation Department, it was preferable to use the questionnaire information rather than to treat these cases as "information not available." Each occupational title was assigned a score on Duncan's Socioeconomic Index. Since we are primarily interested in determining the impact of the respondents' families's socioeconomic status on their educational and occupational aspirations when they were seniors in high school, and on their educational and occupational achievement following graduation from high school, it seemed appropriate to use the average yearly income of their parents from 1957 to 1960, which provides a fairly constant and reliable indicator of the family's economic status. Also, income data were not complete for any one year and by using an average of parental income it became possible to reduce the number of cases with missing income data. Thus, data on average yearly parental income were available for 8,971 or 87 percent of the cases in the one-third sample. The missing data for the remaining 13 percent cases were estimated by regressing the available income data on father's occupation on which complete data were available. The regression equation X = a + bY was estimated, where X = income and Y = father's occupation (Duncan's SES scores), and this equation was used to fill in the missing data.* The means, standard deviations, and intercorrelations of father's education, mother's education, father's occupation, and average yearly income of the parents are given below: Intercorrelations Item Mean S.D. 1 2 3 4 (1)Father's education 2.193 1.333 1.000 0.496 0.487 0.309 (2)Mother's education 2.346 1.292 1.000 0.334 0.227 (3)Father's occupation 29.772 22.081 1.000 0.411 (4)Parents' Average Income (in hundreds of dollars)62.056 56.807 1.000 The method of principal component analysis was used to determine the weights for each of the above-mentioned four items. According to this method, for each scale the selected matrix of correlations, say R, is analyzed. Its largest latent root, say (lambda), and corresponding latent vector, say (nu) are determined. Then the loadings are (nu) multiplied by SQRT [(lambda)] and the weights are (nu) divided by SQRT [(lambda)]. The percentage of variance accounted for is equal to 100 times divided by the number of selected items. In order to make a scale fall in the range 0-99, a linear transformation of the form S = aF + b is performed, where F is the factor score, S is the stored value in the range 0-99, and a and b are constants determined so that the highest F is transformed into 99 and the lowest F is transformed into 0. For a detailed illustration of this method, see Hagood and Price (1952: 523-547). The principal component accounts for 53.61 percent of variance of the items; the weight and correlation of each item with the principal component are as follows: Correlation with Item Weight the Principal Component (1) Father's education 0.37546 .805 (2) Mother's education 0.32700 .701 (3) Father's occupation 0.36154 .775 (4) Parents' Average Income 0.29620 .635 *NOTE: This passage was revised for accuracy and clarity by R.M. Hauser, 7/27/99.