Suppose we are interested in find out whether increasing the availability of computer for student could help to improve their test scores. We start from two variables: the number of computers per student and test scores for 5th graders, comp stu and testscr.

Download the data set caschool.dta and read it into R. You could use the code in "ps2.R" to read in the dta dataset to R. It contains observations on 420 California school districts in 1999 on 14 variables. Description on these variables is as the following:

?district: District code.

? school: School name.

? county: County name.

? gr span: Grade span of district.

? enrl tot: Total enrollment.

? teachers: Number of teachers.

? calw pct: Percent qualifying for CalWorks (income assistance).

? meal pct: Percent qualifying for free lunch.

? computer: Number of computers.

? testscr: Average test scores.

? comp stu: Number of computers per student.

? expn stu: Number of expenditure per student.

? str: Student-teacher ratio.

? avginc: Average income.

? el pct: Percent of English learners.

? read scr: Average reading score.

? math scr: Average math score

(a) Draw a scatterplot of test scores versus number of computers per student. Describe in words what you see.

(b) Run the following bivariate regression of test scores on number of computers per student. Report the constant and slope coefficient.

test scoresi = ? + ?comp stui + ei

(c) What is the predicted test scores of a district with number of computer per student as 0.094? What about 0.164?

(d) What is the expected test score difference if the number of computer per student increase from 0.09 to 0.10? What about from 0.16 to 0.17?

(e) What is the predicted test score of a district with number of computer per student as 10? Do you think this number is economically meaningful? Why or why not? (Note: There is no "correct" answer for this question, just think based on common sense).

3. Suppose we are interested in find out whether increasing the availability of computer for student could help to improve their test scores. We start from two variables: the number of computers per student and test scores for 5th graders, comp_stu and testscr. Download the data set caschool . dta and read it into R. You could use the code in "ps2.R" to read in the dta dataset to R. It contains observation on 420 California school districts in 1999 on 14 variables. Description on these variables is as the following: . district: District code. . school: School name. . county: County name. . gr_span: Grade span of district. . enrl_tot: Total enrollment. . teachers: Number of teachers. . calw-pct: Percent qualifying for CalWorks (income assistance). . meal-pct: Percent qualifying for free lunch. . computer: Number of computers. . testscr: Average test scores. comp stu: Number of computers per student. . expn stu: Number of expenditure per student. . str: Student-teacher ratio. . avginc: Average income. . el_pct: Percent of English learners. . read scr: Average reading score. . math scr: Average math score. (a) Draw a scatterplot of test scores versus number of computers per student. Describe in words what you see. (b) Run the following bivariate regression of test scores on number of comput- ers per student. Report the constant and slope coefficient. test scores; = a + ycomp stu; + e; (c) What is the predicted test scores of a district with number of computer per student as 0.094? What about 0.164? (d) What is the expected test score difference if the number of computer per student increase from 0.09 to 0.10? What about from 0.16 to 0.17? (e) What is the predicted test score of a district with number of computer per student as 10? Do you think this number is econommically meaningful? Why or why not? (Note: There is no "correct" answer for this question, just think based on common sense)