Statistics with Kyle Edmondson
Winter Reflection:
A piece of work I have created this first semester of senior year that I am particularly proud of is my mini correlation project. I learned to create correlations between objects. In order to create correlations, you must go through a strict process. It begins with picking two objects and what you desire to be measured between them. You can find a correlation through running numbers through an excel spreadsheet. What the correlation tells you is how strongly they relate and if they do in fact relate at all.
EVIDENCE:
Does a team’s quarterback rating relate to their number of wins?
The object I measured were the NFC East teams in the NFL: Philadelphia Eagles, Dallas Cowboys, New York Giants, and Washington Redskins). I chose this simply because the Eagles are my favorite NFL team. The two measurements I looked at were quarterback ratings and number of wins between 2006-2013. QBRs did not exist prior to 2006 which is why the data collected starts there. The data I found suggests that when a team’s QBRs (quarterback ratings) increases their number of wins during each season. The sources I used were Wikipedia (to gather each quarterback and the number of games won) and I also used ESPN (to get stats for each quarterback). I used information from seasons between 2006 and 2013 to get the QBR.
Link to my spreadsheet: https://docs.google.com/a/animashighschool.com/spreadsheets/d/1n-zXZ4RvSo_YJkdUCXoPejyHADa4qW4ifKgIslOZ4lY/edit#gid=0
n: 4
r: 0.5864822482
r^2: 0.34396142745
t: 1.03
My r correlation, 0.59, is a strong positive which therefore says QBRs do in fact relate to the number of wins that an NFL team has. The correlation is fairly strong because it is greater than 0.5. The r correlation squared, 0.34, is the percentage of one variable that can be explained by the number of wins. So 34% of the variation in wins is described by QBRs. I found the degree of freedom by doing n-2 so 4-2 giving me the number 2. This tells me that the result is not statistically significant because it is not even in the 95% confidence interval. Therefore, there is a greater than 5% chance this correlation happened by luck.
There are multiple options of what the logic could be to describe this correlation. One being a better QB causes more wins, more wins causes a better QB, or something completely different.The logic that a QB causes more wins makes sense because they are throwing more passing yards and handing off for more rushing yards. The logic that more wins could cause a better QB could make sense because the team is already doing really well that season, they may not be taking as many chances to possibly make a mistake. Something else such as a team with better strategies/systems could be the logic to cause more wins and a better QB. My results are not chance because my stats are very accurate and in the 99% confidence interval. Another method to verify this explanation would be to repeat the same process for every other division in the NFL (AFC and NFC) It would be possible to also verify this through dropping certain people on the same system and focusing on one specific team. I could also test if defense is better at helping cause more wins. When I began this project I assumed that QBRs would relate to more wins and now with my research I am able to know that my hypothesis was correct. I can now compare teams in my favorite division and see how much a better QBR will help the team.
This project has enabled me to see relationships beyond a simple glance. I have realized that I am a strong organizer and do well with communication. This will play really well into my globalization group project. I think that starting documents and keeping them organized will be very beneficial when we have five people all working on the same Google Doc. I think this will also be key when trying to stay on task and meeting deadlines. We will be meeting with Local First soon and I feel that I did a solid job of contacting this group and setting up plans in order to gather information from them. These communication skills will be important when not only contacting outside sources but also within my group to make sure we are all on the same page. I struggled with interpreting data throughout this project. I feel that I have an understanding of how it can be done but I could be more skilled at doing so. Through the next semester I hope that my peers especially can help me interpret the information we get from companies/organizations that we talk to about globalization. I think that also reading articles and anything we find on globalization will be beneficial in moving me towards being better at interpreting data. Moving forward, I am interested in studying food science and human nutrition. Therefore, interpreting data will of test results and things of that sort will be extremely critical in my potential work of being a dietician/nutrionist. When people discuss correlations I will now not only consider only how two things do or do not relate but also the strength of that correlation.
EVIDENCE:
Does a team’s quarterback rating relate to their number of wins?
The object I measured were the NFC East teams in the NFL: Philadelphia Eagles, Dallas Cowboys, New York Giants, and Washington Redskins). I chose this simply because the Eagles are my favorite NFL team. The two measurements I looked at were quarterback ratings and number of wins between 2006-2013. QBRs did not exist prior to 2006 which is why the data collected starts there. The data I found suggests that when a team’s QBRs (quarterback ratings) increases their number of wins during each season. The sources I used were Wikipedia (to gather each quarterback and the number of games won) and I also used ESPN (to get stats for each quarterback). I used information from seasons between 2006 and 2013 to get the QBR.
Link to my spreadsheet: https://docs.google.com/a/animashighschool.com/spreadsheets/d/1n-zXZ4RvSo_YJkdUCXoPejyHADa4qW4ifKgIslOZ4lY/edit#gid=0
n: 4
r: 0.5864822482
r^2: 0.34396142745
t: 1.03
My r correlation, 0.59, is a strong positive which therefore says QBRs do in fact relate to the number of wins that an NFL team has. The correlation is fairly strong because it is greater than 0.5. The r correlation squared, 0.34, is the percentage of one variable that can be explained by the number of wins. So 34% of the variation in wins is described by QBRs. I found the degree of freedom by doing n-2 so 4-2 giving me the number 2. This tells me that the result is not statistically significant because it is not even in the 95% confidence interval. Therefore, there is a greater than 5% chance this correlation happened by luck.
There are multiple options of what the logic could be to describe this correlation. One being a better QB causes more wins, more wins causes a better QB, or something completely different.The logic that a QB causes more wins makes sense because they are throwing more passing yards and handing off for more rushing yards. The logic that more wins could cause a better QB could make sense because the team is already doing really well that season, they may not be taking as many chances to possibly make a mistake. Something else such as a team with better strategies/systems could be the logic to cause more wins and a better QB. My results are not chance because my stats are very accurate and in the 99% confidence interval. Another method to verify this explanation would be to repeat the same process for every other division in the NFL (AFC and NFC) It would be possible to also verify this through dropping certain people on the same system and focusing on one specific team. I could also test if defense is better at helping cause more wins. When I began this project I assumed that QBRs would relate to more wins and now with my research I am able to know that my hypothesis was correct. I can now compare teams in my favorite division and see how much a better QBR will help the team.
This project has enabled me to see relationships beyond a simple glance. I have realized that I am a strong organizer and do well with communication. This will play really well into my globalization group project. I think that starting documents and keeping them organized will be very beneficial when we have five people all working on the same Google Doc. I think this will also be key when trying to stay on task and meeting deadlines. We will be meeting with Local First soon and I feel that I did a solid job of contacting this group and setting up plans in order to gather information from them. These communication skills will be important when not only contacting outside sources but also within my group to make sure we are all on the same page. I struggled with interpreting data throughout this project. I feel that I have an understanding of how it can be done but I could be more skilled at doing so. Through the next semester I hope that my peers especially can help me interpret the information we get from companies/organizations that we talk to about globalization. I think that also reading articles and anything we find on globalization will be beneficial in moving me towards being better at interpreting data. Moving forward, I am interested in studying food science and human nutrition. Therefore, interpreting data will of test results and things of that sort will be extremely critical in my potential work of being a dietician/nutrionist. When people discuss correlations I will now not only consider only how two things do or do not relate but also the strength of that correlation.