In their paper on peer effect in golf, Guryan, Kroft and Nodowidigdo investigate
ID: 1214940 • Letter: I
Question
In their paper on peer effect in golf, Guryan, Kroft and Nodowidigdo investigate whether a player’s score is lower if his playing partners are better golfers. Ability is measured by the average score per round over the season. Other influences on score are controlled using fixed effects. (Don’t worry about this term – we will study it later.) Standard errors appear in parentheses below the coefficient estimate. The data involve scores from rounds played during several Professional Golf Association seasons. The paper is here:
http://www.nber.org/papers/w13422.pdf
You should not have to read the paper to answer this question.
Consider this regression table from their paper.
1. Consider the first regression (left hand column).
a. What is the dependent variable?
b. What is the hypothesis being tested?
c. Which variable captures the hypothesis?
d. What is its coefficient?
e. What does the coefficient mean?
f.Is it significant at the 5% level?
g. What does this mean?
h. Would you say that there is good evidence for peer effects? Explain.
i. Is the coefficient on own ability significant at the 5% level?
j. What type of variable is this?
k. What does this estimate mean (in English)?
2. Bobo, who is from Ghana, likes to wrestle. He wrestles on Tuesday and Friday. (By the way, his name means, “born on Tuesday.”) The following table gives his results this year on the two days.
Tuesday Thursday
Win 20 15
Lose 8 13
a. He wants to know if he wrestles better on Tuesday. What statistical test should he use?
b. What is the hypothesis to be tested?
c. What is the null hypothesis?
d. He performs the test that you recommend and finds a p-value = 0.167. Interpret this number.
e. What is learned from this test? Explain.
3. Fred, like all normal people, loves to bowl. He never has an open frame. That is, each frame is either a strike (x) or a spare (/). Strikes are better. Here is his data.
xxx/xxx/////xxxx////x
a. What is the conditional probability of a strike in the frame following a strike?
b. How many runs for Fred?
c. The runs test for this data gives the following.
Expected Number of Runs: 11.5; sd: 2.2279
z-value= -2.0091
Explain the results of this test. Does it show evidence of the hot hand?
d. Draw a nice picture of the distribution of runs and show the actual value, the z-statistic.
TABLE 4 -THE EFFECT OF PEERS' ABILITY ON OwN SCORE Own ability Average(ability), partnens Tournament x category fixed effects Measurement error correction Plaver fixed effects 0.672 (0.039) 0.035 (0.040) Yes 0.949 (0.057) -0.036 (0.063) -0.032 (0.040) Yes No Yes es No Yes No 17,492 No 17,492 17,492 Notes: Results in column 1 are from baseline specifications as specified in equation (3). Column 2 reports a measurement-error-corrected estimate using the estimator described in Appendix B. Column 3 reports results using player fixed effects instead of own ability. The dependent variable is the golf score for the round. The ability variable is measured using the player's handicap. Standard errors are in parentheses and are clustered by playing group. All regressions include round fixed effects and weight each observation by the inverse of the sam pling variance of estimated ability of each player.Explanation / Answer
1.
(a)The own score of the individual is the dependent variable.
(b)The hypothesis is: Whether peer's ability affects own score?
(c)Average ability,(partners) variable captures the hypothesis.
(d) -0.035
(e)It estimates suggests that improving the average ability of one playing partners by one stroke actually increases (i.e. worsens) own score by 0.035 strokes.
(f) It is not statistically significant.