As the class begins their research for the Women’s World Cup, I thought I would share some interesting information regarding the determinants of women’s football performance. To start, there is not a lot of information. The lack of research on women’s performances makes it hard to have reliable statistical data. However, what is being looked at, I found to be very interesting. For example, some researches are taking the temperatures from the days of every World Cup game from the past 30 years and comparing it the home country of the national team. I never thought a variable like temperature would have an effect on the performance of a soccer team, but some researchers claim to have found “statistically significant” data. Other have tested for the effect of women’s rights index and GDP per capita, which make more sense to me. The variables relating to economics were not always straight forward though. For many nations, higher GDP were associated with better teams. But this model is nuanced, there are many distributions of data that can be tested, which will yield different results. But it would make sense, a country with more wealth will have more disposable income to spend on things like women’s soccer.
To remedy the issue of lacking information on women’s soccer teams and determinants statistics, I challenge all the sports scientists out there to work on a few passion projects for this topic. Also, to finish the class, my team and I will be exploring new determinants of women’s soccer success to add to the already present literature.
I’m also on this team. The analysis should be relatively straight forward, though there are some considerable nuances worth considering. First, a lot of the interesting factors worth considering lack clear measures. How does one quantify women’s rights? Or skill level? This is where proxying comes into play. For factors like women’s rights, we consider factors such as women’s representation in government, basic civil liberties (voting, speech, etc.) and collective indices compiling all these factors. Similarly, for skill we utilize variables such as ranking (i.e. 3rd best in the world), win rates, or ELO rankings. These variables come close to measuring the total effect but will likely always fall short. As such, it’s worth noting that the results will likely not be fully representative. Second, which factors are worth considering? There are dozens of variables that we could have used: GDP, women’s rights indices, fandom, etc. But the analysis would likely fail if we put in too many variables, as the model would be over-fitted. Many of these variables are correlated with one another, and adding them all to the model will bias the effect each has. As such, we settled on only five or six.
It will be interesting to see what the result reveal and if there are clear-cut answers.