Statistics Cann

Sample Statistical Analysis of the question:

Has the NFL become more pass oriented?

The purpose of my study is to determine if the National Football League has become more pass oriented. Meaning that there is a significantly greater amount of plays that are passing as opposed to rushing. I will define a pass play as any play during the course of a game in which a forward pass is attempted.

I chose to do this particular study because over the years I have seen (what seems to me to be) a spike in the amount of passing plays. That and the more and more frequent use of “college style” plays (which emphasize gadgetry) have lead to more coaches thinking outside the box. Since there are only so many types of running plays that can be used, the creativity is usually embodied in the form of a pass play.

The population I will use for my study is the 32 teams in the NFL. I will begin my study during the year that the NFL expanded to 32 teams, 2002. I will be surveying the entire population of NFL teams from the 2002-2008 seasons. There will be no use of randomization in this study because I will be working with the entire population of NFL teams.

I expect that this analysis will indicate that there is a significant difference between passing attempts and rushing attempts during the years indicated. I do not, however, know how that will be manifested within the study.

The descriptive statistics used in this study will be counts of passing plays and rushing plays for each of teams in each of the years in the study. These values can be found in the excel chart attached. There were several outliers (which can also be found on the attached excel spreadsheet). Some were outliers on the high end, others were on the low end. Often times, there would be passing outliers that were on the high end; and in the same year, the same team would be an outlier on the low end. That makes sense because if a team has an unusually high number of pass attempts, that team would most likely have an unusually low number of rushing attempts.

For this study, I will use two levels of hypothesis testing. First, I will use a 2 population mean significance test to test the claim that the mean pass attempts is greater than the mean rush attempts for each of the 7 years at the 0.01 significance level.

2002 Pass / 2002 Rush / 2003 Pass / 2003 Rush / 2004 Pass / 2004 Rush / 2005 Pass / 2005 Rush / 2006 Pass / 2006 Rush / 2007 Pass / 2007 Rush / 2008 Pass / 2008 Rush
/ 540.375 / 440.688 / 515.406 / 453.375 / 511.063 / 450.875 / 514.5 / 449.219 / 512.156 / 451.5 / 532.656 / 437.063 / 516.438 / 441.219
/ 52.0347 / 45.208 / 44.944 / 44.476 / 56.333 / 58.824 / 67.544 / 56.329 / 52.334 / 48.326 / 52.378 / 48.653 / 57.547 / 56.421
P-val / / / / / / /
P>R? / yes / yes / yes / Yes / Yes / yes / yes

As you can see in the table above, there is sufficient evidence (in each year) to reject the claim of no difference between the mean passing attempts and mean rushing attempts, at the 0.01 significance level. Therefore, my claim that there is a significantly greater amount of passes per season than rushes is true.

Then, I will use a One-Way ANOVA to determine if there is a significant difference in passing percentage among the 7 years. For this I will use, STATDISK software.

Source: DF: SS: MS: Test Stat, F: Critical F: P-Value:

Treatment: 6 24953.839286 4158.973214 1.326483 2.886002 0.246422

Error: 217 680368.40625 3135.338278

Total: 223 705322.245536

Fail to Reject the Null Hypothesis

There is not sufficient evidence to reject equality of means

The graph below is a scatterplot that shows the relationship of number of passes over the 7 year period. It does not indicate a trend of growth over the given timeframe.

The vertical axis represents the average number of passes in the NFL.

The horizontal axis represents the year.

This study has demonstrated that while there is a statistically significantly greater number of passes attempted in the NFL on a year-by-year basis, there is not a significant increase throughout the seven-year period. The presence of several outliers indicates that while most teams run a balanced offense, other teams skew towards or away from passes depending on their circumstances, team philosophy, personnel, etc. I did not find there to be any trouble spots in this study. It was long and arduous, but the data was readily available (and accurate) from nfl.com. This study did seem to ask more questions than answer though. A good follow-up would be an attempt to identify the teams that are more pass oriented than others and then to see if that has any impact on their season’s record. The method I used was somewhat convoluted. I worked among three types of statistical analysis software (STATDISK, Excel, TI-84 plus silver edition). If I were to do it all over again, I would make more of an effort to stick to one piece of technology. There was also a lot of copying and pasting from nfl.com. A set of downloadable statistics would have been easier to work with.

Ultimately, the NFL skews more toward the pass than the run on a yearly basis with no end in sight!

1