Quality and Other Technical Information for SESTAT Fields Using Occupation Codes

After applying the "best" code editing rules, it was found that more managers were coded into the management category than were coded out. CATI resulted in a higher percentage of respondents being coded into management than did the mail questionnaire. Examples of common occupation recodes included:
• Respondents coded out of a scientific occupation field were most often coded into teacher/professor.
• Clinical psychologists tended to miscode themselves as diagnosing/treating practitioners. "Clinical psychologist" will be cross-referenced in 1995.

The mode effect study found respondents who reported a science and engineering occupation in the Decennial Census were more likely to select non-science and non-engineering occupations in the NSCG interview. CATI reported higher non-scientist and non-engineer responses than those completed by mail:

• Physical scientists: mail - 30.2 percent; CATI - 39.7 percent
• Mathematicians/computer scientists: mail - 37.3 percent; CATI - 45.0 percent
• Psychologists/social scientists: mail - 71.0 percent; CATI - 84.1 percent
• Total scientists/engineers: mail - 32.1 percent; CATI - 38.0 percent
• Management: mail - 9.7 percent; CATI - 11.3 percent

Since NSCG respondents designated as "S&E" would be retained for potential follow-up in 1995, NSF examined the extent to which respondents were switching into and out of S&E occupations. To be designated "S&E" a respondent must be working in a scientific or engineering field (as defined by NSF) or have majored in such a field. S&E occupations were determined by three possible occupation codes: the occupation code chosen by the respondent (initial interview), the code selected by best coding, or the code the respondent chose during the reinterview.

Some switching existed between technicians and professionals in engineering. This is a problem because technologists are considered non-S&E. This switching may be attributable to the fact that the job code list did not include the word "engineering." In related questions concerned with work activities, respondents did not interpret "management and administration" consistently. Because of the variation that appears to have existed, this category cannot be used as a management measure. On the basis of respondent-provided information, 20 percent changed their occupation response between the initial interview and the reinterview.