top banner

Key Terms


Unemployed: Either on layoff or not employed but searching for work (during the four-week period prior to the reference date). NOTE: Individuals who are not working and not employed are defined as being out of the labor force. This group includes most individuals who are voluntarily not employed.

Labor Force: The labor force consists of unemployed plus employed individuals.

Unemployment Rate: The percent of the labor force that is unemployed.

Standardized Unemployment Rates: A number of techniques can be used to estimate the effect of one (independent) variable on another (dependent) variable, while "controlling" for other variables. The most straightforward technique is to construct a three-way table. For example, the average age of women in the doctoral labor force is less than that of men. To determine whether age and/or gender are related to the unemployment rate, it is logical to look at unemployment rates within sex-age groups (for example, men under age 30 compared with women under age 30, men aged 31--35 compared with women aged 31--35, etc.). Although cross-tabulations can be extremely helpful in understanding how two or more independent variables affect a single dependent variable, usefulness is limited by the fairly large sample sizes needed to estimate accurately subgroup unemployment rates. This becomes an especially serious problem when controlling for more than one variable. (For example, to understand whether observed differences in unemployment rates for individuals with disabilities can be explained by the fact that individuals with disabilities tend to be older than individuals without disabilities, and the fact that the incidence of disabilities tends to be higher among men than among women).

Instead of using cross-tabulations for control purposes in this report, a multivariate technique known as logistic regression analysis was used to estimate the simultaneous effect of a large number of variables on unemployment. The independent variables used in the logistic regression model are: degree field; place of employment or residence; years since receipt of Ph.D.; age when doctorate received; years of part-time experience; years of full-time experience; whether or not employed in April 1988; occupation in April 1988 (for employed individuals); employment sector in April 1988 (for employed individuals); parents' level of education; disability status; percent involuntarily out-of-field in the 1988 occupation; foreign research experience; marital status; interaction between gender and marital status; interaction between gender and whether children are in the home; interaction between gender and race/ethnicity; and interaction between marital status and race/ethnicity.

The logistic regression model was used to estimate the unemployment rate for a group of individuals who exhibited the same values on all of the independent variables except the one under consideration. For example, the observed unemployment rate for individuals with hearing disabilities was 3.0 percent, compared with a rate of 1.6 percent for non-disabled doctoral scientists and engineers; the respective standardized rates were 2.5 percent and 1.6 percent. Thus, factors listed above (other than whether the person had a hearing disability) explained some but not all of the observed difference between those with hearing disabilities and those without disabilities.

More detailed information about the standardization process is included in the Technical Notes. (See p. 47.)

 


top toc help SRS Homepage