Topic

Hypothesis Testing Assignment - Difference in college attendance between male and female.

I perform hypothesis testing, using the t-test to establish whether there is a significant difference in college attendance based on gender. The data used is CSP March 2017 by the U.S. Census Bureau.

In conclusion, there is a statistical significance of college attendance in favor of females, but there is no economic significance.

Main learning components

• Statistical significance
• Economic significance
• t-test
• Rejection regions

How to access the data

According to the U.S. ​Department of Commerce​, the Current Population Survey (CPS) is the primary resource of the labor force for the population of the United States. The data we will be looking at today is the “cpsmar2017” data, detailing information of the U.S. labor force in March 2017. The data is collected from 60,000 households, identified by address or person, in the States every month with voluntary participation. Those ineligible to the survey are individuals under 15, member of the U.S. military, and people in institutions (United States Census Bureau, n.d.)

Interviewers use a computer-assisted survey instrument, visit the household personally or perform telephone interviews. Each household is representative for thousands of other households. The same sample is surveyed for four months consecutively, with every eight months not surveyed in between. More specifically, approximately seventy five percent of participants are included in the sample from one month to the next; fifty percent is kept constant as the same month of the year prior. The data set is a cross section data due to the fact that the monthly census does not record variables on a timeline, as surveys are performed each month. Thus, the analysis will focus on the differences in the current states of households interviewed, not taking into account any changes over time. The survey also applies a two-dimensional matrix data structure.

Visualizations

Table 1. Joint probability table on the number and percentage of a random person to fall into any of the four categories. The number on top indicates the number of people, the decimal numbers indicates the probability (unit: %).

Conclusion

I began the hypothesis test by creating a graph of female and male college attendance. My two graphs indicate that there is a difference in gender for higher education level, but is not reliable to show any significance. With a t-test, I was able to conclude, as there is a 95% confidence that the sample mean will fall within the critical values of [0.034, 0.045], which indicates that there is a statistical significance in college attendance between male and female. We can also say that there is a ~4% chance of getting a college education in favor of females compared to males. However, this result is not economically significant and does not prove that a female has a significantly more advantage than her counterpart in being accepted for college.