This chart is incredibly interesting. Ever since Brady pushed for road teams to be able to bring their own footballs (remember Romo's fumble in Seattle?), these are the fumble rates per play for road teams since.
NOTICE HOW NEW ENGLAND'S FUMBLE RATE WENT WAY DOWN COMPARED TO THE LEAGUE AVERAGE. The result of manipulating the psi on the footballs? Before this, New England was pretty comparable to everybody else's fumble rate.
This is a very telling graph. It's difficult not to draw conclusions from this graph alone as there appears to be a direct correlation between ball security and the rule change - marginally, league-wide and substantially for NE.
The first year following the rule change, NE saw a 94.8% increase in its ball security stats while the other 31 teams, on average, saw a 7% decrease. Post 2006, the NFL shows a positive upward trend, punctuated by three seasons (2011, 2013, 2014) that fall significantly above the average for the entire 15-year period. The slopes of the two NE lines are approximately equal to the NFL average plot but prior to the rule change NE was performing at approximately 7.1% to 7.8% above the NFL norm, and this was aided in large part due to year 2005 when NE was +15 (38.5%) better than the rest of the league.
Pre 2007, the difference could be accounted for by better coaching (during the Belichick era the Patriots have had one of the best coaching staffs in the NFL) which emphasizes ball protection better than most other teams. Even with the outlier of 2005, there could have been other NFL teams with more touches per fumble than NE. It would be interesting to see how the other top ball protection teams fared compared to NE both prior to and after the change.
Some of the difference, pre 2007, could be accounted for by play calling (% run plays vs. % pass plays), level of pass protection (i.e., more sacks probably directly correlate to more QB fumbles), the strength of defenses within a given division (e.g., would Cleveland be more prone to lower ball protection stats having to face very physical defenses with strong front sevens in Pittsburgh, Baltimore and Cincinnati as opposed to an NFC South or AFC South team facing less imposing defenses?) and quality/style of running backs.
However, post 2006, the plots are so drastically different that the above variables simply do not seem able to account for the difference in fumble rate. Even taking out year 2010 (+69; +168%), which seems to be an anamoly, NE is so far above the rest of the league that the data must raise flags. The only oddity in the 8-year plot is year 2013 in which NE fell below the NFL average. Otherwise, the Patriots were better by a margin ranging from +20 (+45.5%) to +41 (+83.7%), excluding 2010.
Those results are too statistically-significant to dismiss to coaching, % of run plays, opposing defenses and RBs.
What would be helpful in assessing this graph is to see how the top three teams beside NE fared from 2007-2014; to plot Tom Brady's % completion and QB rating year-by-year; and to track the fumble rate for RBs who have left the Patriots since 2007 to see if there is any difference in their ball security stats with NE as compared to their new teams, as well as tracking the ball security stats of backs who have left other teams to join NE since 2007. If backs have left the Patriots and seen their stats come down more to the league-wide average or if backs have seen much better performance - as statistically significant as the post 2006 plot - since leaving other teams and joining NE, the theory that under-inflated balls accounts for the difference would seem to hold water.