Scribd
Upload
99 views3 pages

Simple Regression Case: London Olympics

The key regressions and findings were: 1) A regression of weight on height for basketball players found height was a statistically significant predictor of weight, with each additional cm in height corresponding to a 1.06 kg increase in predicted weight on average. 2) A regression found Chinese male swimmers were on average 2.03 years younger than non-Chinese swimmers, providing evidence they may be too young. 3) Regressions of weight on age for different sports found a statistically significant increase in predicted weight with age for

Uploaded by

Kathy Thanh PK
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
99 views3 pages

Simple Regression Case: London Olympics

The key regressions and findings were: 1) A regression of weight on height for basketball players found height was a statistically significant predictor of weight, with each additional cm in height corresponding to a 1.06 kg increase in predicted weight on average. 2) A regression found Chinese male swimmers were on average 2.03 years younger than non-Chinese swimmers, providing evidence they may be too young. 3) Regressions of weight on age for different sports found a statistically significant increase in predicted weight with age for

Uploaded by

Kathy Thanh PK
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 3

Worksheet 6

Simple Regression Case: London Olympics


You have been hired by the Chinese Olympic Committee as a statistical advisor to their coaches. Your
first task is to give advice about the ideal weight and age for athletes by sport and gender. You
download data (“Worksheet 6 2012 Olympic Data.xlsx”) available from The Guardian on the weight,
height, gender, country of origin and sport of most athletes who competed in the 2012 London
Olympics.
Task A: Understand the relationship between weight and height for basketball players

1. You start with the following regression:

Regression 1:

Predicted Weight=b + m*Height

a. Run the regression and report the estimated equation:

Predicted Weight=−112.65+ 1.057∗Height


b. Does it make sense to interpret the intercept? Explain.

No, especially because there is no such thing as a negative weight, but the
literal interpretation would be that the average weight for an Olympic
basketball player who was 0 cm tall is approximately -112 kg.

c. What is the interpretation of the coefficient on Height?

For every additional cm, the predicted weight for an Olympic basketball player
increases by 1.06 kg.

2. Chinese players are on average 4 centimeters taller than Americans.


a. What would the model predict their average difference in weight to be?

1.0574 * 4 = 4.23 kg

b. Give a confidence interval for this difference.

[0.908*4, 1.206*4] = [3.632, 4.824]

3. You have heard of two Australian players, David Andersen and Patrick Mills. One’s hobby is to
lift weights and the other is to run in the mornings. Use your model to determine who is who.
Lifts weights:_____ Patrick Mills ________ (He weighs 85 kg, we predict
weight= - 112.65 +1.057*180 (since he is 180 cm tall) = 77.61 so there is a
7.49 kg residual/error. Since he weighs more than average for his height, he probably
lifts weights and has a lot of heavy muscles.)

Runs:________ David Andersen ________ (Weighs 102, predict weight to be 111.43, so


there is a -9.43 kg residual/error. Since he weighs less than average for his height, he
probably runs off any extra weight.)

Task B. Figure out if Chinese swimmers are too young

4. In the past, the Chinese male swimming team has been criticized for being too young and
inexperienced.

a. What regression would you run to test if, indeed, Chinese male swimmers are younger
than average? (hint: to run the regression you need to construct a dummy variable for
being from China).

Predicted Age = 23.290 - 2.0289 China

b. Interpret the coefficients.

Intercept: The average age for non-Chinese swimmers is 23.29 years.


Coefficient on China: Chinese swimmers are, on average, 2.03 years younger than non-
Chinese swimmers

5. With your model, do you have evidence that the Chinese are, on average, younger?

Yes because the coefficient on the dummy variable is statistically significant (|t-stat| >
2, p-value < 0.05, and the 95% confidence interval does not contains 0) so we reject
the null that Chinese players are same age on average.

Task C. Determine expected weight gain with age, by sport (extra – time permitting)
6. You wonder if swimmers and basketball players gain weight as they age, and what the expected
gain per year is by sport. Run the following regression, separately by sport:

Predicted Weight=m + b*Age

Report your regression equation:

Basketball: Predicted Weight=96.492+0.0936∗Age

Swimming: Predicted Weight=64.960+0.639∗Age

7. The basketball nutritionist states that, on average, players’ weights are flat with respect to age.
Can you prove him wrong with your model?

No: we cannot prove him wrong because 0 falls within the 95% confidence interval for the
basketball age coefficient.

8. On the other hand, the new swimming nutritionist argues that to keep up with the population
average weight change, Chinese swimmers should aim at gaining weight at a rate of 1 kilo per
year. Based on your model, do you agree?

No because 1 falls outside the 95% confidence interval for the swimming age
coefficient. Thus, we can reject the null hypothesis that B1 = 1.

9. What do you think can explain that the point estimate is so much smaller for basketball players?

The average age for basketball players is 27, for swimmers it is 23. The corresponding
age histograms are left and right-skewed respectively (i.e. there are far more teenage
swimmers than teenage basketball players). And since younger athletes are still
growing, it is no surprise that the age coefficient for swimmers is statistically
significant.

You might also like