The Glass Ceiling A Study on Annual Salaries Group 4 Julie Shan Brian Abe Yu Ting Cheng Kathinka Tysnes Huan Zhang Andrew Booth Agenda Introduction Exploratory Analysis Linear Regression Analysis Conclusion Further Analysis Introduction What Why A sample of 1980 s managers salaries To determine factors that affect the salary How Linear regression Introduction Data Set Analyzed A subsample of a large data set from the early 1980s from a study investigating potential gender bias in determination of professional salary differentials The individuals come from several large corporations Data was organized by Management Level Gender Education Level Years in Job Salary Exploratory Analysis Exploratory Analysis Affects of the independent variables on the dependent variable SALARY Independent Variables Years in job Management level Education level Gender Exploratory Analysis Salary vs Years in Job 30000 SALARY 25000 20000 15000 10000 0 4 8 12 YEARS 16 20 24 Positive Relationship Between Years in Job and Salary Exploratory Analysis Upper Management Earns More Than Lower Management Salary vs Management Level 30000 25000 SALARY 20000 15000 10000 0 2 0 0 0 2 0 4 0 6 0 8 MANAGEMENT 1 0 1 2 Exploratory Analysis Salary vs Education Level 30000 SALARY 25000 20000 15000 10000 0 1 2 EDUCATION 3 4 More Educated Managers Earn More Outliers May Skew Regression Results Exploratory Analysis Female 0 if Male Female 1 if Female Note Many More Males than Females in Data Set Females Seem to have Cap Lower Max Salary Salary vs Female 30000 25000 SALARY 20000 15000 10000 0 2 0 0 0 2 0 4 0 6 FEMALE 0 8 1 0 1 2 Exploratory Analysis New Variable Female management 1 and 2 correspond to men and women in lower management respectively 3 and 4 correspond to men and women in upper management respectively Again females earn less have a cap on salary Salary vs Female Management 30000 25000 SALARY 20000 15000 10000 0 1 2 3 4 FEMALE MANAGEMENT 5 Linear Regression Analysis A regression of salary vs the other variables Ed1 3 are dummy variables for education level Ed1 high school Ed2 bachelors Ed3 graduate degree Linear Regression Analysis All variables except female are significant at a 5 level R2 0 94 so it is a good fit The Durbin Watson is less than 2 but greater than 1 Linear Regression Analysis Histogram of Residuals 10 Series Residuals Sample 1 43 Observations 43 8 6 4 2 0 2000 1000 0 1000 2000 Mean Median Maximum Minimum Std Dev Skewness Kurtosis 1 45e 12 6 465925 3085 931 2363 230 1280 438 0 299525 2 570329 Jarque Bera Probability 0 973732 0 614549 3000 Jarque Bera statistic is greater than 0 05 indicating normality of the residuals Linear Regression Analysis Updated regression excluding variable FEMALE Linear Regression Analysis R2 0 93 still a good fit The Durbin Watson statistic is once again less than 2 but greater than 1 Linear Regression Analysis Histogram of Residuals 8 Series Residuals Sample 1 43 Observations 43 7 6 5 4 3 2 1 0 3000 2000 1000 0 1000 2000 Mean Median Maximum Minimum Std Dev Skewness Kurtosis 2 44e 12 148 6636 2850 258 2681 484 1338 158 0 193549 2 286455 Jarque Bera Probability 1 180694 0 554135 3000 Jarque Bera statistic is greater than 0 05 indicating normality of the residuals Linear Regression Analysis Wald Test for equivalency of intercepts for various education levels Ho ED2 ED3 Ho ED1 ED2 Linear Regression Analysis Final Model SALARY 615 0378 YEARS 7509 9807 MANAGEMENT 7352 3861 ED1 10907 4441 ED23 Linear Regression Analysis Conclusion The variable FEMALE was not statistically significant No gender bias at a 5 significance level There is gender bias at a 10 significance level Other variables played important role in determining salary The number of years worked in a job add to salary level The higher one s education level the higher the salary level Upper management has higher salaries than lower management Further Analysis Newer Larger Data Set Allows Removal of Outliers Additional Independent Variables Company Size Industry Age of Company More in Depth Analysis of Potential for Gender Bias At 10 it was Significant Fin Any Questions
View Full Document
Unlocking...