Write down your research question (from component 1)
How does the number of coaches assigned to US college sports teams vary by the gender of the team?
Write down your hypothesis (from component 1)
I predict that for sports in which there is both a men’s and a women’s team, coaching levels will be equivalent. I expect this because universities are required to adhere to Title IX, which would prohibit different levels of resources for equivalent sports.
Write down the name of your data set (from component 1)
Write down the names of the variables you will use in your analysis (i.e., what you need to type to access them in R)
ncoaches, teamgender, sport
For each variable, write:
A brief description of what it measures
Whether it is an explanatory (independent) variable, a response (dependent) variable, or something else (explain what)
What kind of variable it is (categorical or numeric and which subtype within these)
ncoaches: This measures the number of coaches on each team. It is my response variable. It is a discrete numeric variable.
teamgender: This records the gender of the players on the team. It is an explanatory variable and it is nominal categorical.
sport: I will use this variable to limit my data frame to the sports that have both men’s and women’s teams. It is a nominal categorical variable.
Part 2: Univariate distributions
For each of your variables:
Look at its distribution using the summary() or table() function (as appropriate to its type).
Indicate how many missing values (NAs) there are.
In ~2-3 sentences, describe what you see in the results. Note anything that surprises you.
ncoaches:
The minimum number of coaches per team is one, the median is four, and the maximum is 14. There are no missing values. The fourth quartile is much wider than the other three (spanning from 4-14), and the mean (3.9) is larger than the median, suggesting that the data may be skewed to the right. In other words, there may be a small number of teams with a large number of coaches relative to the rest.
The median and the cutoff point for the third quartile are the same (both equal 4), indicating that four is a really common number of coaches for teams to have.
summary(sports$ncoaches)
Min. 1st Qu. Median Mean 3rd Qu. Max.
1.000 3.000 4.000 4.058 4.000 14.000
teamgender:
There are 418 men’s teams and 461 women’s teams in these data. There are no missing values. I am surprised that there are so many more women’s teams than men’s teams. I would have expected the numbers to be about the same.
table(sports$teamgender, useNA ="always")
men women <NA>
418 461 0
sport:
There are 17 sports represented in my data (though it is likely that “All Track Combined” and “Track and Field X Country” are different words for the same thing, which would make it 16 sports). There are no missing values. Most sports have either 76 or 38 distinct teams in my data. It is possible that the sports with 38 teams are those that are available only to one gender, and the ones with 76 (38x2) teams have both men’s and women’s versions.
table(sports$sport, useNA ="always")
All Track Combined Baseball Basketball
72 38 76
Fencing Field Hockey Football
76 38 38
Golf Gymnastics Lacrosse
76 19 76
Rowing Soccer Softball
38 76 24
Swimming and Diving Tennis Track and Field X Country
76 76 4
Volleyball Wrestling <NA>
38 38 0
Part 3: Preparing your data set
Will you analyze all the observations in your data set? Explain why or why not. If not, which will you include and which will you exclude?
I will not analyze all of the observations in my data set because analyzing all of them would make for an unfair comparison. There is a different set of sports available for men than for women. I am interested in comparing the number of coaches for comparable men’s and women’s teams, so I will exclude sports for which there are only men’s teams or only women’s teams (e.g., football, softball).
Do you need to create any new variables? Explain why or why not. If so, what are they?
I do not need to create any new variables. My explanatory variable is teamgender and my response variable is ncoaches, and they already fit my research question well in their current forms.
If necessary, use the code chunk below to make any needed changes (filtering observations and creating variables) to your data.
# first, I need to figure out which sports have teams for both genderstable(sports$sport, sports$teamgender)
men women
All Track Combined 36 36
Baseball 38 0
Basketball 38 38
Fencing 38 38
Field Hockey 0 38
Football 38 0
Golf 38 38
Gymnastics 0 19
Lacrosse 38 38
Rowing 0 38
Soccer 38 38
Softball 0 24
Swimming and Diving 38 38
Tennis 38 38
Track and Field X Country 2 2
Volleyball 0 38
Wrestling 38 0
# from that table, I can see that I need to remove baseball, field hockey, football, gymnastics, rowing, softball, volleyball, and wrestling.sports_filtered <- sports |>filter( sport !="Baseball", sport !="Field Hockey", sport !="Football", sport !="Gymnastics", sport !="Rowing", sport !="Softball", sport !="Volleyball", sport !="Wrestling" )# did it work? Let's make the same table with the new data settable(sports_filtered$sport, sports_filtered$teamgender)
men women
All Track Combined 36 36
Basketball 38 38
Fencing 38 38
Golf 38 38
Lacrosse 38 38
Soccer 38 38
Swimming and Diving 38 38
Tennis 38 38
Track and Field X Country 2 2
# now all the sports have both men's and women's teams--we're good to go!
Part 4: Multivariate visualizations
Make a plot showing the relationship between your explanatory (independent) and response (dependent) variables. Depending on your variable types, this might be a scatter plot, box plot, bar chart, etc. Include meaningful axis labels.
Interpret your plot in a few sentences. Does it appear these variables are associated with one another? Does anything about the relationship surprise you? Is it consistent or inconsistent with your hypothesis?
As suggested by the univariate summary statistics, the distribution of the number of coaches is skewed to the right. The distributions look nearly identical for men’s and women’s teams, so it appears that team gender and staffing levels are probably not associated with one another. This is consistent with my hypothesis and is not very surprising, though I was not expecting the distributions to be as extraordinarily similar as they appear here.
# I have one numeric and one categorical variable, so a boxplot (as in example code in the assignment template) would also be a good choice here--but for variety I will show the same information in histogram formggplot(data = sports_filtered, aes(x = ncoaches, fill = teamgender)) +geom_histogram(position =position_dodge(), # this puts the bars next to each otherbinwidth =1 ) +labs(x ="Number of coaches",y ="Number of teams",title ="Distribution of staffing levels by team gender for Duke and UNC sports",fill ="Team gender")
