De Zhong Gao (260421995)
Professor Elisabeth Gidengil
Techniques of Empirical Research
November 24, 2011
Education and Income: A Case Study
There is a popular saying that all Yalies earn at least $110 after graduation. One can wonder whether a high level of education necessarily guarantees a high income in the future. In this paper, I will argue that the higher the level of education attained, the higher will be the earned income. In order to validate such causal relationship, I will bring three control variables in my testing hypothesis: the spurious variable of gender, the conditional variable of racial background, and socio-economic status—being in the middle or working class. As a spurious variable, gender affects people’s level of education attained and income. Being a woman or man predetermines people’s level of education and income in the future. Indeed, women have less access to education than their male counterparts, and thus they will earn a smaller income. A possible factor that can explain this is the dominance of paternalistic family in most Western societies that encourage men to pursue higher levels of education.
Social class and race also predetermine people’s level of education they will receive later in life as well and income. Because of systematic inequality in the United States, members of the minority groups and members from poor social class are less likely to have access to higher education than Whites. For the conditional variables of socio-economic status—being in the working or middle class—and racial background, my hypothesis claims that working class, Blacks and Hispanics tend to have a higher level of education than Whites and middle class people, for Blacks and Hispanics and people from the working class value education more than Whites do, since education is regarded as a golden life opportunity which will help them escape from the cycle of poverty. This essay will strive to provide an insightful analysis of how gender, socio-economic and racial background affect people’s level of education and income using cross-tabulations.
Because I have used data from the American Voting Election, the selection of subjects in the survey is not random. The ex post facto experiment is used by researchers when the assignment of subjects to the control and experimental group is not random. Researchers attempt to mimic post-test design by using statistical method, which substitute for the lack of random assignment to the control and experimental group. The amount of money earned annually as an indicator of the variable of income. Indeed, the annual salary evaluates one another’s capability to earn income. The level of educational attainment is an indirectly indicator. Educational attainment is representative of the concept of education in that it illustrates the amount of education one another has received over the course of his or her life.
For the ordinal level of measurement, Kendall’s tau is one of the most commonly used measures of association for ordinary variables. Unlike tau-c that is used when the table is not dichotomous, tau-b is used only if the table is has the same number of categories in the dependent variable as in the independent variable. Tau-b is thus used. For the original table, the tau-b is 0.35, indicating a moderately strong relationship.
The p-value is crucial in assessing the probability that any relationship between the independent variable—in our case, the level of education—and dependent variable—income—in sample occurred by chance. In this case, the p-value is 0.000, which means that there is less than 1 chance out of 1,000 that the relationship in the sample has simply occurred by chance. Thus, the relationship is deemed statistically significant since it is unlikely that there is no real relationship between education and income in the population. Because our data have been collected from our sample, there are inevitably sampling errors, which make the sample not perfectly representative of the studied population. Chi-square tells researchers whether an apparent nominal-level association between two variables is likely to result from chance. It does so by comparing the observed result with the expected results has the relationship not existed.
Re-coding my Variables
I have recoded the independent variable by collapsing the initial six categories of education, which were less than high school, high school, some college, associate degree, college degree and advanced degree into three: less than high school and high school, some college and associate degree, and a college and advanced degree. Getting some education lesser than high school and a high school diploma can be regarded as a single category. Earning some college education can be the equivalent of an associate degree, so that some college education can equate an associate degree. Receiving a college diploma and an advanced degree both fall under the category of people who have a higher education, and can be regrouped under a single category. I have also recoded my dependent variable of income into three categories: the first category includes between less than $15K and $30K, for people with relatively low income, the second category comprises the categories of $30K and $75K, and the last category refers to the categories of $75K and more than $110K. Finally, for the category of race, I have decided to only focus on the cases of Whites, Blacks and Hispanics, for the other cases contain too few cases for analysis.