Appendix 7D: Estimating the Probability of a Person with Given Background Traits Being Underutilized in 2013–2014

The logistic regression coefficients can be used to estimate the probability of a person with given characteristics being underutilized in 2013–2014. The procedure for estimating the probability of a person being underutilized with given traits is relatively straightforward. The probability that a given person being underutilized is equal to the following:

To calculate the values of Pi, we begin by calculating the value of α + βx for an individual with given traits, Xi (e.g., gender, race-ethnic origin, age, education, nativity, disability, family income level). The values of the α and β's are those generated by the logistic regression model. We then calculate the value of eα+βxi. The value of the denominator is simply equal to 1+ eα+βxi. The ratio of these two values would then yield the estimated probability of college attendance for this individual.

Appendix 7E: Logistic Probability Model of Labor Underutilization for Labor Force Participants Under 30

The following are details regarding estimates of a logistic probability model of labor underutilization among labor force participants under 30 in 2013–2014 (see Table 7E.1). The base group for this analysis is a 25to 29-year old White nonHispanic male who was not disabled, held a bachelor's or higher degree and lived in a family with an income over $150,000. [1]

Similar to our findings for all working-age adults (16 and over), gender had only a very modest impact on the labor underutilization rate of teens and young adults. Holding all other demographic and socioeconomic traits constant, young women were slightly under one percentage point less likely than males to be underutilized. [2] Teens and young adults (20–24 years old) faced much higher rates of labor underutilization than their older peers (25–29 years old). A teen labor force participant (or a member of the labor force reserve) was nearly 11 percentage points more likely than his or her peers 25–29 years old to be underutilized, while a 20–24 year old was about six percentage points more likely to be underutilized than his older peers.

Members of each race-ethnic group were significantly more likely than White non-Hispanics to be underutilized. The estimated sizes of these independent impacts of race-ethnic group varied from lows of two to three percentage points among Asians and Hispanics to a high of nine percentage points among Black non-Hispanic youth. The educational attainment of these youth also had frequently strong impacts on the probability of being underutilized at the time of the 2013–2014 surveys. Relative to their base group peers with a bachelor's or higher degree, those young adults who lacked a high school diploma or GED were nearly 14 percentage points more likely to be underutilized. High school graduates were 10 percentage points

Table 7E.1 Findings of the logistic probability model of the underutilized status of individual members of the young adult labor force under age 30 in 2013–2014

Variable

(A) Logit

coefficient

(B) Sig. of coefficient

(C) Marginal

probability at the mean

Constant

−2.777

0.01

Female

−0.038

0.01

−0.005

Asian

0.206

0.01

0.026

Black

0.713

0.01

0.090

Hispanic

0.197

0.01

0.025

Native American/Other

0.443

0.01

0.056

Native Born

0.162

0.01

0.021

Disabled

0.798

0.01

0.101

Age 16–19

0.859

0.01

0.109

Age 20–24

0.457

0.01

0.058

High school student

0.947

0.01

0.120

High school dropout

1.117

0.01

0.141

High school graduate

0.790

0.01

0.100

13–15 Years, no degree

0.381

0.01

0.048

Associate's degree

0.233

0.01

0.029

FAMINC < $20,000

0.923

0.01

0.117

FAMINC $20,000–$39,000

0.365

0.01

0.046

FAMINC $40,000–75,000

0.130

0.01

0.016

FAMINC $75,000–$99,000

−0.002

0.000

FAMINC $100,000–$149,000

−0.045

0.05

−0.006

Note: Implies not statistically significant −2 Log likelihood =364601, Nagelkerke R Square = .142, Chi Square = 36761, Sig. = .01, DF =19,

N =377,096

more likely to be underutilized than bachelor's degree holders. The impact drops to only 5 percentage points for those with 13–15 years of schooling but no degree and to under three percentage points for those with an associate's degree.

Family income of respondents also affects an independent impact on the probability of young adults being underutilized in the labor market, but the negative impacts are primarily concentrated among low-income and low-middle-income youth. Those young adults with household incomes under $20,000 had a probability that was 12 percentage points higher of being underutilized than their affluent peers, and those young adults with incomes between $20,000 and $40,000 had a five to six percentage point higher probability of experiencing an underutilization problem. There were no significant differences between upper-middle-income youth and the most affluent families.

The above findings illustrate quite dramatically that among the young as well as among all workers, age, race-ethnic origin, educational attainment, and family income status played jointly large roles in shaping the incidence of underutilization problems in 2013–2014.

  • [1] The expected probability of labor underutilization among the base group was only 5.9 percentage points.
  • [2] Male teens and those 20–24 were heavily hit by changing employment developments over the 2000–2014 time period, including the high loss of blue-collar jobs that impacted young men more than young women.
 
< Prev   CONTENTS   Next >