#75461 - 4 Pages - Research project of computer gaming

1 | P a g e

pLAB-3 REVISION

Q1: Based on the plot, gender has significance for BS and Bcom and not for Ba.

True or False. Why?

Q2: Ask yourself, if you think gender, race, and GPA can help predict salary, could you use a two-way ANOVA? Why?

2 | P a g e

Q3: On average, economists published 4 academic articles in their entire life. True or False. Why? INTERPRETATION OF MEAN DIFFERENCE AND CONFIDENCE INTERVAL This specific test tells us that students in the Econ field will publish less than those in other fields. In fact, by the end of PhD graduation non-economics PhDs have published, on average, 3.6121 papers more than those in the Econ field. Looking at the 95% confidence interval of difference we can assume that 95% of these cases will fall between 2.7939 or 4.4303 more papers published by those outside of econ field for their PhD.

Q4: The effect of holding a BSc degree compare to the other types of degrees, BA and Bcom, has positive effect on annual salary of the individual.

Based on Coefficients Table below, is this statement true or false? Why?

3 | P a g e

EXCELLENT UNDERSTANDING OF CONTROL VARIABLE Q4.4: Looking at the Coefficients and Summary Table, what can you conclude on the effect of gender on salary? Also, did introducing gender to the regression model changed the overall effect of degree-type? (10 pts) As shown in the tables, the unstandardized coefficient for the dummy variable “Gender of Graduate” is -16412.081 which suggests the average annual salary for female graduates is 16412.081 units of salary less when compare to that of male graduates while holding all other variables constant. And the p-value (0.000) of this coefficient indicates the salary difference between male and female is statistically significant. Moreover, the adjusted R square for this model is 0.710 which is larger than the adjusted R square (0.538) of the model without the variable “Gender of Graduate”. It means the control variable “Gender of Graduate” helps to explain the variation in the dependent variable “salary”. By comparing the unstandardized coefficients of degree type in two regression tables, we can see the coefficients of the variable “degree-type-BA” and “degree- type-BSC” in the model with the variable “Gender of Graduate” are smaller than that of the model without the variable “Gender of Graduate”, which means the variable “Gender of Graduate” reduced the explanatory power of degree-type on salary. Therefore, we could conclude that the control variable “Gender of Graduate” changed the overall effect of degree-type on average annual salary.

4 | P a g e

NO CONTROL

WITH GENDER AS CONTROL

HOW TO INTERPRET UNSTADARDIZED COE. FOR CONTINUOUS VARIABLE INTERTATION OF REGRESSION TABLE Q5.4: Interpret the results. (10 pts) As shown in the tables, the unstandardized coefficient of the variable “# of Job Experience” is 1276.059 (p=0.033), we expect 1276.059 units increase in salary of graduates for every unit increase in the number of job experience while holding all other variables constant. Similarly, we expect 306.390 (p=0.090) units decrease in salary of graduates for every unit increase in the number of skills while holding all other variables constant, but the p-value (0.090) indicates the coefficient is not statistically significant. Besides, the R square (0.040) suggests only about 4% of the variation in the dependent variable (salary) are explained by these two independent variables (# of Job Experience and # of skills).

5 | P a g e

Q5: Why is Degree Type the only independent variable recoded into multiple dummy variables? (4 pts) Because it is a categorical variable. In linear regressions all independent variables that are categorical need to be recorded into multiple dummy variables. What is missing in the statement above?

GOOD EXAMPLE – USE OF DUMMY VARIABLES

6 | P a g e