Sunteți pe pagina 1din 5

Statistics 3500 SP2012

Exam #2 (a) March 16, 2012

Ellebracht 100 points

Name: ______________________________

Student#:______________________________ =============================================================== Instructions DO NOT separate the exam form. You must turn in the entire exam paper. The last problem on the exam is Q5(B). It is on page 4. Some work must be shown in order for you to receive credit. (No work=no credit.) If a space is not provided, indicate your final answer by circling it. Carry all computations to at least 3 decimal places. If a question asks you to perform a hypothesis test, be sure to include the null and alternative hypothesis, a conclusion, and some supporting work for that conclusion. Point values for problems are marked. The entire exam is worth 100 points. Work that is neat and easy to read tends to be graded easier. Any questions or complaints about the scoring or grading of the exam must be brought to the instructors attention during or immediately following your first class following the exam. If you miss that class without prior approval, you will forfeit any possibility for grade appeals or corrections.

===============================================================

Grading use only:

TOTAL : _________________ / 100

Q1. Data was gathered for all teams (28 total) from the 1994 National Football League (NFL) regular season. First we will look at a simple linear regression using the number of interceptions (INT) to try and predict the total number of points a team scored (PTS). The MiniTab output is below. Regression Analysis: PTS versus INT
The regression equation is PTS = 360 - 2.13 INT Predictor Constant INT S = 59.2081 Coef 360.16 -2.130 SE Coef 43.67 2.493 T 8.25 -0.85 P 0.000 0.401

R-Sq = 2.7%

R-Sq(adj) = 0.0% Predicted Values for New Observations

Analysis of Variance Source Regression Residual Error Total DF 1 26 27 SS 2557 91145 93703 MS 2557 3506 F 0.73 P 0.401 Obs 1 Fit 311.2 SE Fit 18.8 95% CI (????) 95% PI (183.5, 438.9)

a) (6 points) Compute the correlation (r) between the two variables. Would this be considered strong or weak correlation? Why?

b) (10 points) In the MiniTab output there is a 95% prediction interval for INT=23 (x=23) given. Compute the corresponding 95% confidence interval for the mean value of y.

Q2. (2 points each) Write down the assumption about the residuals (errors) that each plot corresponds to. a) Normal Probability Plot b) Residuals vs. Fitted Values

c) Histogram

d) Residual vs. Order

Q3. In 1994 each NFL team belonged to one of six divisions (AFC East, AFC Central, AFC West, NFC East, NFC Central, NFC West). The division a team belongs to would be considered a qualitative (categorical) variable. a) (10 points) Write down the proposed regression model for estimating the mean number of wins per division using the AFC West division as the baseline for comparison. BE SURE TO CLEARLY DEFINE ALL INDICATOR VARIABLES.

b) (4 points) Assuming you did a) correctly, what would the value of 0 be? NOTE: You are NOT able to determine the exact numerical value, but should be able to define it in the context of the problem.

Q4. A multiple linear regression model was fit using the following 6 independent variables: points (PTS), yards (YDS), passing touchdowns (P_TD), interceptions (INT), running touchdowns (R_TDS), and fumbles lost (FL). The dependent variable for the regression model was wins (WINS). Answer the following questions using the provided MiniTab output. a) (6 points) Based on the scatterplots provided, put the independent variables in the appropriate column based on what you would expect each variables contribution to the regression model to be. Positive Negative Insignificant

b) (10 points) Perform the global utility test at an alpha-level of 5%. INTERPRET YOUR TEST CONCLUSION.

c) (6 points) Which independent variable in the model appears to be the most significant? Justify.

d) (4 points) How much of the variability in y is explained by the current model, after adjusting for the number of independent variables?

e) (10 points) Construct a 95% confidence interval for the (slope) parameter for P_TD.

Q5. Attached are four models along with the residual plots. Answer the following questions. a) (18 points) Determine which model you believe to be the best model. Make sure you tell me why you chose your model as well as why you did NOT choose the other models. Make sure that you comment on the assumptions and all things that made you decide on your model. Use an alpha-level of 5% in your analysis.

b) (8 points) The following data is from the 1994 Kansas City Chiefs: Variable PTS Value 319 YDS 5692 P_TD 20 INT 14 R_TD 12 FL 12

Using the model YOU chose in part a), determine the number of predicted wins for the Kansas City Chiefs.

S-ar putea să vă placă și