COR 074c

August, 1970
Revised: July, 1999

On the Coding of Occupational Aspirations

1.  COR 74D describes the dual coding of occupational aspirations 
    into "Intended Occupation Class," (OCCL57) and "Intended      
    Occupation Prestige (NORC)." (OCPR57)

2.  Of the two codings, the latter contains far more detail,   
    while the former has more of the character of a true 
    occupation code. While the question in the 1/6 sample 
    conducted by Professor J. Kenneth Little (see note) has only 
    seven precoded categories plus an open-ended category, 
    respondents apparently volunteered a detailed response in a 
    large number of cases. Among male respondents in the 1/3 
    sample conducted by Professor William Sewell, for example, 54 
    prestige levels are represented in the NORC coding, and 27 
    levels were chosen by 25 or more respondents. In contrast, 
    the "intended occupation class" code contains only 18 
    occupational categories, and 10 of these are subdivisions of 
    the professions. The marginals for these two variables 
    (labelled 'approximate' in the codebook) for 1/3 sample males 
    are shown on Attachment B.

3.  The "NORC" codes are not actually prestige scores. Rather, 
    they are Duncan's curvilinear transformations of his SEI for  
    occupations into the metric of the 1947 NORC scores.

4.  There is a third entry for occupational aspirations, 
    "R's Occupational Aspiration: Duncan's SES index." (OCST57) 
    While each of the other codes has a category for "no          
    information," this variable does not. There is no reference 
    to missing data for this variable in any published papers, 
    nor is there any documentation on how it was coded. A simple 
    matching exercise has established that this variable is 
    a recode of "intended occupational class". That is, it loses  
    whatever detail was captured in the "NORC" coding. For males  
    only, N's reported in OCCL57 were  matched with those in 
    OCST57 yielding a set of corresponding codes. Then, each set 
    of codes was used to list the N's for females.  These matched 
    perfectly, except females reporting no future plans other
    than marriage were coded "19" on the Duncan SES scale.

5.  In the Duncan SES recode the missing data were eliminated by  
    assigning nonresponses (and marriage only responses) code 
    "19", the code based on the socioeconomic characteristics of  
    employed men not reporting an occupation in 1950.

6.  The recode to the Duncan HS scale, rather than the more 
    detailed NORC prestige scores, has been used in published 
    regression analyses. The rationale for this choice is 
    suggested by the comparison of correlations involving the  
    two variables which were abstracted from output used in       
    William H. Sewell, Archibald O. Haller, and George W. 
    Ohlendorf. 1970. "The Educational and Early Occupational   
    Status Attainment Process: Replication and Revision." 
    American Sociological Review 35 (December): 1014-27. The 
    correlations involving the Duncan SEI were higher by a 
    substantial margin in every case.

7.  This result is wholly attributable to the treatment of        
    nonresponses in the two recodings of occupational aspiration. 
    While nonresponses were coded "19" in the Duncan scale, they 
    were assigned the value "00" in the NORC scale, a value which 
    does not actually occur on the prestige scale and which is 
    well outside the range of scale values. When non-responses 
    for occupational aspiration are eliminated, the Duncan and 
    NORC codings produce nearly identical correlations. 

8.  Our own preference for a socioeconomic code for occupational  
    aspirations is a retransformation of the detailed NORC 
    code back into the Duncan metric. While these results do not  
    differ substantially from the other two codings, in 
    most instances the correlations in this third set are 
    the highest of the three.

9.  It is apparent that the combination of recoding and deleting  
    missing data does not substantially change the pattern 
    of correlations for occupational aspirations for either
    subsample of males. Again, in most instances the correlations 
    involving the retransformation of the NORC scores are larger.

Note On 1/6 and 1/3 Samples:
     Professor J. Kenneth Little conducted follow-ups in Fall,
1957 and Spring, 1968 using a 1/6 random sample of all the
Wisconsin 1957 high school graduates.  Professor William Sewell
used a 1/3 sample that consists of Little's 1/6 sample and
another random sample of 1/6 of the graduates.  This 1/3 sample
is the sample that all subsequent WLS follow-ups have used.