Statistics: Comparing z scores across distributions

In summary, the conclusion that player A is overall better is based on assumptions of normality and may not accurately reflect the true differences between the two players' abilities. It is important to consider the limitations and assumptions of the statistical methods used and to be cautious in drawing conclusions based on them.
  • #1
Quantum1990
7
0

Homework Statement



This problem is taken from the 2011 AP Statistics Exam, which we reviewed in class. The exact problem can be found here, and solution after:

http://apcentral.collegeboard.com/apc/public/repository/ap11_frq_statistics.pdf
http://apcentral.collegeboard.com/apc/public/repository/ap11_statistics_scoring_guidelines.pdf
Its Question 1, part c.

However, my problem is more general.

Suppose I have two players, each with measured running times and weight lifting amounts. It has already been concluded that at least one of these distributions is not normal. The z score table is reproduced(approximately) below:
A B
Amount Held 2.4 2.6

Running Time -1.2 -.2

According to the solution, A is slightly worse at weight lifting, but much greater at running, thus A is the overall better player. However, in concluding that, we saw that the B led A in amount held by .2, and A led B in running time by 1.0. We compared the two z scores, and determined A's ability in running overshadows B's ability in amount held. But by doing this, aren't we comparing z scores(or differences in them) across distributions? Specifically, how do we know that a 1.0 lead in running is more impressive than a .2 lead in Amount held, if we don't know the skewness of the distributions?

This was a long question, and I wasn't quite sure how to phrase it. Any help in understanding this would be great.

Homework Equations





The Attempt at a Solution

 
Physics news on Phys.org
  • #2


it is important to always consider the limitations and assumptions of any statistical analysis. In this case, the conclusion that player A is overall better is based on the assumption that the distributions of weight lifting and running times are both approximately normal. However, if this assumption is not met, then the conclusion may not be valid.

Additionally, the z-score table only allows for comparisons within a single distribution, not between different distributions. Therefore, comparing the z-scores of weight lifting and running times may not accurately reflect the true differences between the two players' abilities.

Furthermore, as you mentioned, the skewness of the distributions is also an important factor to consider. If the distributions are highly skewed, then the z-scores may not accurately represent the players' abilities. In this case, it may be more appropriate to use non-parametric methods of analysis that do not rely on the assumption of normality.

In conclusion, while the analysis presented in the solution may provide some insights, it is important to consider the limitations and assumptions of the statistical methods used and to be cautious in drawing conclusions based on them. It may be beneficial to explore alternative methods of analysis and to further investigate the distributions of the players' abilities.
 

FAQ: Statistics: Comparing z scores across distributions

1. What is a z score?

A z score, also known as a standard score, is a statistical measure that indicates how many standard deviations a particular data point is above or below the mean of its distribution. It is calculated by subtracting the mean from the data point and then dividing by the standard deviation.

2. How do you compare z scores across distributions?

To compare z scores across distributions, you must first standardize the data by converting each data point to its corresponding z score. Once the data is standardized, you can compare the z scores across distributions to see how each data point relates to its own distribution's mean and standard deviation.

3. Why is it important to compare z scores across distributions?

Comparing z scores across distributions allows you to see how a particular data point compares to the rest of its distribution and other distributions. This can help you understand the relative position and significance of the data point within its distribution and in comparison to other distributions.

4. Can z scores be negative?

Yes, z scores can be negative if the data point is below the mean of its distribution. A negative z score indicates that the data point is below the mean and a positive z score indicates that the data point is above the mean.

5. How do outliers affect z scores?

Outliers can have a significant impact on z scores, especially if they are extreme. Outliers can increase or decrease the mean and standard deviation of a distribution, which in turn can greatly affect the z scores of other data points in the distribution. It is important to identify and address outliers when comparing z scores across distributions.

Similar threads

Replies
1
Views
4K
Replies
3
Views
748
Replies
5
Views
4K
Replies
4
Views
4K
Replies
6
Views
2K
Replies
3
Views
1K
Back
Top