Part I -The Influence of Educational Attainment, Occupational Status, Sex, Age and Race on Personal Income
What accounts for the variance in personal income? To provide an answer I downloaded the recent IPUMS dataset (2011) with the following variables: personal income, educational attainment, occupational status, age sex, and race. I constrained the sample to include only individuals between the age of 22 and 64. The sample included 1,277,325 individuals but when sampling weights are applied the population size is equal to 134,476,073 individuals.
Two major analyses were undertaken: a multiple regression analysis and a logistic regression analysis.
I. Multiple Regression Analysis
Personal income was regressed on 7 levels of educational attainment, occupational status, age, sex, 3 categories of race (white black and other) and the interaction of educational attainment with age squared. The multiple regression output is found in Appendix A.
I believe predictive margins graphs provide the most succinct manner to interpret the results.
A. Predictive Margins of Seven Levels of Educational Attainment
With age on the horizontal axis and the predicted personal income on the vertical axis we can observe:
- The higher the level of educational attainment the higher personal total income,
- Individuals with the the top three levels of higher educational attainment (Master’s degree, Bachelors degree and Professional/Doctorate degree) enjoy greater returns on their educational investments.
- Total personal income increases for the top five levels of higher educational attainment, flattens out and then decreases somewhat, albeit at different rates of decline.
B. Predictive Margins of Three Levels of Educational Attainment
The seven levels of educational attainment were collapsed to three levels of attainment and the regression was rerun with the following outcome.
Again the higher the level of educational attainment the higher the predicted total personal income.
C. Predictive Margins of Sex
Throughout the age distribution males make more than females. (On average, more than $19,300.)
D. Predictive Margins of Sex By Race
The ranking of total personal income (highest to lowest) is:
- Male, White
- Male, Other
- Male, Black
- Female, White
- Female, Other
- Female, Black
E. Predictive Margins of Education by Sex
Collapsing education to three levels we can observe the predictive margins of education by sex on total personal income.
The ranking of total personal income (highest to lowest) is:
- Bachelor’s Degree or More, Male
- Bachelor’s Degree or More, Female (Note that bachelor degree females don’t reach this ranking until after age 33.)
- Some College, Male
- High School or Less, Male
- Some College, Female (This is a surprise in the ranking. One might expect that ‘females with some college’ would rank higher than ‘males with high school or less’.)
- High School or Less, Female
F. Predictive Margins of Occupational Standing
Not unexpectedly, the higher one’s occupational ranking the higher total personal income.
Watch for Part II of this analysis. We will examine the influence of educational attainment, occupational status, age, sex and race on the odds and probabilities of earning an above median income through the use of logistic regression.
Appendix A – Regression Analysis Output
Data Source
Steven Ruggles, J. Trent Alexander, Katie Genadek, Ronald Goeken, Matthew B. Schroeder, and Matthew Sobek. Integrated Public Use Microdata Series: Version 5.0 [Machine-readable database]. Minneapolis: University of Minnesota, 2011.