Solution: In order to predict how a school district would have - #80101
This is the rough introduction to the problem.
In order to predict how a school district would have scored when accounting for poverty and other income measures. The Cincinnati Enquirer gathered data from the Ohio Department of Education’s Education Management Services and the Ohio Department of Taxation. First, the newspaper obtained passage-rate data on the math, reading, science, writing, and citizenship proficiency exams given to fourth, sixth, ninth, and 12th graders in early 1996. By combining these data, they computed an overall percentage of students that passed the tests for each district.
The percentage of a school district’s students on Aid for Dependent Children (ADC), the percentage who qualify for free or reduced-price lunches, and the district’s median family income were also recorded. A portion of the data collected for the 608 school districts follows. The complete data is in the file attachment.
The data have been ranked based on the values in the column labeled % passed; these data are the overall percentage of students passing the test. Data in the column labeled % on ADC are the percentage of each school district’s students on ADC, and the data in the column labeled % Free lunch are the percentage of students who qualify for free or reduced-price lunches. The column labeled Median Income shows each district’s median family income. Also shown for each column labeled district is the county in which the school district is located. Note that in some cases the value in the % Free Lunch column is 0. Indicating that the district did not participate in the free lunch program.
Written below is what we were given on how this assignment should be completed.
One change is that I have coded county as an independent variable. You will now have 4 independent variables.
Your task is to develop the best possible regression equation for predicting scores. First you will run a correlation analysis of all independent variables and discuss significant findings.
You will run more than one regression; the first should set all independent variables against the dependent. At this point you will choose to throw out a variable that shows no significance, after running it as a simple regression to make sure of no significance. Then you will start combining the three variables left to get the best adjusted R square. (There should be four of these)
This is to be a comprehensive well written case that discusses your reasoning behind everything that you do; including the data points (F tests, T tests) to back up your reasoning. You should have an introduction and conclusion to your case. You should discuss descriptive statistics for each of the variables at the beginning of the case.
In your conclusion you will give the equation that you picked from your regressions. At this point you are to tell me what you would do differently if you ever tried to make a model on % passing the test again. Hint: you have two highly correlated variables in your equation, next time you made a model you would only use one of them, which one? (Two more regressions). Then you should recommend what variables to use and make suggestions for other independent variables to consider, for your next model
Your report should include each regression that you run (total of 8) and the correlation table. DO NOT include the raw data as there are over 600 data points. Present a summary of your analysis, including key statistical results, conclusions, and recommendations, in a managerial report. Include any technical material you feel is appropriate in an appendix.
Deliverable: Word Document
and pdf
