Saturday, 16 June 2012

Maximizing the Returns from Undergraduate Library Fees: A Literature Review of Scholarly Articles on Football

Mired in the doldrums of NFL offseason coverage, football fans looking for insightful analysis or interesting materials on the subject are largely out of luck, and will remain in such a state over the next several weeks. Growing tired of in-depth OTA and minicamp play-by-plays (and the assorted articles arguing how said comprehensive accounts are largely meaningless) and intensive coverage of Twitter kerfuffles, these are tough times for the several football fans possessing reading abilities on or above the grammar school level.  Having already exhausted the few decent football books myself (The Essential Smart Football is definitely worth a read) and slogging through several examples of mediocre literary gridiron fare, I decided to pursue more unorthodox mediums: academic literature.

Every time an NFL coach is faced with a crucial decision on fourth down, the media will inevitably bring up the economist David Romer's article on fourth down conversions and risk averse behavior. I also remember reading about a study that found that the success of the Oregon Ducks football team reduced students' GPAs and tempered their teetotaling tendencies. In light of these scholarly football articles, I decided to search my college's online library archives to see if I could find similar pieces to tide me over until the mindless extrapolation and scrutinization propagated by the football media begins in earnest once players graduate from shorts to knee pads in August and the preseason kicks off. Now I have made such attempts at finding interesting scholarly articles in the past with mixed results, but perhaps the staggering economic and social impacts of King Football on America have stimulated the pens of many an academic. I have presented the best articles I found from my search in the rest of the post. I tried to link to the full-length article if I could, but for the others you will be forced to deal with only an abstract and my sub-par interpretations and summaries. 

Press Pass: Payoffs to Media Exposure Among NFL Wide Receivers by Julianne Treme and Samuel K. Allen (2011). From the Journal of Sports Economics.

Much to the chagrin of long-snappers and coverage gunners everywhere, wide receiver has traditionally been one of the more glamorous positions in the NFL. Plenty of e-ink has been spilled covering the dealings of the recently-released Terrell Owens and the newly-employed Chad Ochocinco and current star receivers such as Larry Fitzgerald and Calvin Johnson enjoy sizable contracts and lucrative endorsement deals. This paper attempted to test the effects of on-field production and off-field popularity on wide receivers' salaries and draft positions. The authors used data from draft prospects at the position from 2001 through 2006. They created regression models that measured a player's draft position and salary while controlling for physical abilities, college production, and other factors.

Media exposure was found to have beneficial effects on a prospect's draft position even after controlling for factors such as touchdowns in college and performance in the Combine. Overall,  every two additional newspaper citations for a player in the year before the draft increased their draft position by two picks, holding all else constant. A 0.10 second reduction in a receiver's 40-yard dash time was found to raise their position by around 11 to 12 picks and their rookie salary by roughly $200,000, conclusions which were probably incredibly skewed by the personnel decisions made by the perpetually confuzzling and speed-loving Al Davis. The authors conclude that the 40-yard dash has a very large effect on draft position and salary while faster times were not associated with better rookie years in terms of receptions, the latter of which isn't all that surprising. Treme and Allen argued that the statistically significant effects of media exposure on draft position and salaries illustrated that coaches are possibly concerned with media coverage and popularity outside. This paper looks especially prescient in light of Chad Ochocinco's recent signing with the Miami Dolphins, though it is likely that exhortations from HBO executives and the frightening prospect of starting Davone Bess and Brian Hartline at wideout certainly had something to do with that as well.

Trying Out for the Team: Do Exhibitions Matter? Evidence from the NFL by Lee A. Craig and Alastair R. Hall (1994). From the Journal of the American Statistical Association

While it may be unfathomable to fans today, preseason games in the NFL used to be a big deal. A 1969 matchup between the New York Jets and Giants attracted 70,874 to the Yale Bowl and then-Giants head coach Allie Sherman was fired before the start of the actual season that year after recording an 0-5 record in exhibition play. During the forty-odd years since, preseason games have largely devolved into uninteresting contests which no fan would willingly travel to Connecticut for. The NFL exhibition schedule had already reached such a state by 1994, the time this article was written, which was influenced by an article by Rick Telander in Sports Illustrated calling for the end of the preseason. This paper tried to defend the exhibition schedule and its usefulness and predictive ability. One interesting hypothesis tested in the paper is the idea that high-performing teams feature more competition among veteran free-agents (though interestingly enough this data is culled from the era before the more restrictive Plan B free agency was struck down by the U.S. federal court) because such "good" franchises are more likely to already have quality veterans under contract and are likely to have fewer spots up for grabs while many free agents will want to play for a winner. These competitive effects will lead to improved preseason performance and cause teams that play well in the preseason to carry their success over to the regular season.

The authors tested this assumption with preseason data spanning from 1971 through 1991. The results indicated that there was a small positive effect with the effect of preseason wins on regular season performance. If a team won 6 preseason games rather than 0 (this uses data from when the preseason was an interminable six weeks) they would be expected to win 2 more games in the regular season. Craig and Hall also discovered that a 1% increase in veterans on a roster increased a team's winning percentage by roughly 2.5%. The veteran competition hypothesis is intriguing but I don't really think it is all that valid in driving preseason performance. There certainly would be competition among marginal free agents to make a good team, but I don't think big-name free agents who would be likely to actually bring in regular season success like Mario Williams really need to sweat their preseason performance all that much. I will agree that the NFL preseason can be a valuable opportunity to coaching staffs to evaluate their personnel, especially backups who may not play much in the regular season, but I remain very skeptical that preseason records have really any association with regular season performance.

The NFL Combine: Does It Predict Performance in the National Football League? by Frank E. Kuzmits and Arthur J. Adams (2008). From the  Journal of Strength and Conditioning Research. 

One slightly more puzzling development in NFL history is the mutation of the annual draft Combine from a no-frills event in Florida to a media event which some people actually willingly attend (in Indianapolis no less). While many fans are quick to mention Combine stars who fizzled in the NFL such as Vernon Gholston, Matt Jones, and Mike Mamula, analysts and coaches still put a lot of stock into how prospects perform during the event. This study attempted to determine whether such scrutiny and attention is warranted and if Combine measurements actually had some predictive value determining NFL success.

Kuzmits and Adams utilized data on quarterbacks, running backs, and wide receivers invited to the Combine from 1999 through 2004. The results indicated that most tests, such as a the 40-yard dash, vertical jump, bench press, and Wonderlic had little forecasting ability for NFL success, at every position. They did find that 10, 20, and 40-yard dash times had a small positive association with running back performance in the NFL, but I should mention the caveat that this paper was published before the miserable 2011 campaign of Chris Johnson and his 4.24 40 time. I can't say that these results are incredibly surprising, given that the 40-yard dash was designed by Paul Brown many moons ago to assess punt coverage ability and the Wonderlic dates back to 1936 and was designed for general usage among workers in many industries rather than just football players, something that I'm sure Morris Claiborne will eagerly point out to you. The study concludes by stating that many of the analyzed tests should be underemphasized in personnel decisions but that the Combine still has practical value in facilitating social interaction between coaching staffs and players trough interviews and position workouts.

Playing Position and Psychological Skill in Football by Richard H. Cox and Ho-Sang Yoo (1995). From the Journal of Sport Behavior.

There are a few generally-accepted stereotypes concerning the mental and emotional composition of various football positions. Offensive linemen, especially centers, are characterized as intelligent field generals who must understand the responsibilities of his teammates, while the football cognoscenti claim that short memories and resiliency are vital to the success of cornerbacks and quarterbacks. Conrad Dobler, a former offensive guard for the Cardinals and widely considered the league's dirtiest player during the pre-Bill Romanowski era claimed that his degree in child psychology was great preparation for dealing with developmentally infantile defensive linemen. (Perhaps it made it easier for him to kick them in the head?) And thanks to the exploits of Mike Vaderjagt, Garo Yepremian, and Bill Gramatica, kickers are often viewed as foreign, quirky, and over-exuberant. Building off of previous research limited to the personality characteristics of individual positions, this study was designed to examine if players at different positions exhibited significantly different psychological makeups. 

The paper uses data on 43 players from a major Division I-A program in the Midwest. Given that Cox has served as a professor for many years at the University of Missouri, we can probably assume that he was dealing with the Mizzou Tigers of his home school. Participants completed  the Psychological Skills Inventory for Sports (PSIS R-5), a 45-item test designed to measure an athlete's psychological skills and abilities. The results indicated that linemen on both sides of the ball demonstrated lower levels of psychological skill compared to members of offensive and defensive backfields. In this case, psychological skill means concentration level, anxiety control, and confidence. Offensive players had slightly lower anxiety scores than their defensive counterparts. Offensive backs displayed significantly higher motivation levels than offensive linemen. The authors hypothesized that these mental discrepancies are largely the result of development resulting football experiences rather than from any inherent characteristics, and they argue that positions such as cornerback require significant mental preparation and discipline due to the negative psychological effects resulting from lapses such as blown coverages.

Catching a Draft: The Process of Selecting Quarterbacks in the NFL Amateur Draft by David J. Berri and Rob Simmons (2009). From Journal of Productivity Analysis.

Manning vs. Leaf. Luck vs. RG3. Russell vs. Quinn (good gravy 2007 was a terrible year for quarterbacks). Every NFL draft features this discourse where draftniks voice their affiliations and take stances on what prospect is the superior quarterback in the draft. The increasing importance of the passing game has elevated such decisions to an even higher magnitude, and getting a franchise quarterback through the draft is incredibly important for NFL teams, a claim that Daniel Snyder would surely have few quibbles with. Berri and Simmons wanted to test whether talent evaluators were any good at actually doing this. The redistributive effectiveness of the NFL draft (something which has been challenged by the academic KG Quinn (to the 2 people who actually click on my (rather thoughtful and informative, if I may say so myself) hyperlinks: Ctrl-F on the page for Quinn) in the past) hinges on the staff of poorly performing teams (who don't mortgage their entire drafts for washed up former Heisman winners) doing just that.The article starts with the interesting tidbit explaining the impetus for the current draft structure. Former NFL commissioner Bert Bell apparently adopted a reverse-order amateur draft in response to a vicious bidding war between the Brooklyn Dodgers and Philadelphia Eagles over the services of the middling fullback Stan Kostkas in 1935, which led to the lucky back earning one of the highest salaries in the league. The redistributive draft order was also quite friendly to Bell, who owned the Eagles, the worst team in the league that year (...and many years to follow as well).

The authors wanted to figure out what factors NFL teams used to choose quarterbacks, how such factors actually predicted NFL production, and if a quarterback's draft status had any impact on their subsequent NFL performance. They tested the relationship between draft selection and NFL production with data on 331 quarterbacks selected in the NFL draft from 1970 through 2007. Berri and Simmons measured NFL performance with a QB score metric, essentially a simplified version of a quarterback rating. They found that quarterbacks picked earlier in the draft had higher aggregate QB scores compared to quarterbacks picked later in the draft, but that quarterbacks drafted in picks 11-50 and 51-90 actually had higher QB Scores on a per-play basis. Essentially, quarterbacks picked earlier weren't necessarily better than their lower-picked peers, they just played more, probably because NFL coaches are loath to admit that they made a mistake with their top draft pick and refuse to acknowledge sunk costs. Right Rex Ryan? The authors tested what factors decided a quarterback's draft position with data on 121 quarterbacks picked between 1999 and 2008. They found that characteristics measured at the Combine such as height, weight, Wonderlic scores and 40-yard dash performance explained about 20% of the variation in a quarterback's selection spot, and adding a performance metric based on college production only increased the explanatory value of the model by 3%, suggesting that teams may weigh measurables over college production to a significant degree. The paper then tested the predictive effects of such factors as they related to NFL performance, and the authors found that Combine results and college production barely had any effects on NFL production, though there was a significant and positive relationship between college and NFL completion percentages.

Honorable Mention: Effect of National Football League Games on Small Animal Emergency Room Caseload  by Rozanski et al. (2009). From the Journal of the American Veterinary Medical Association.

Unfortunately, my university could only give me an abstract and citation for this article. The summary hinted that the basis of the study was to determine whether the games of "popular professional football teams" (I suppose they excluded the Allen Wranglers and Jacksonville Jaguars from their analysis) have any impact on the hospitalizations, euthanizations, and deaths of cats and dogs at emergency veterinary clinics in the teams' local markets. Thankfully the good people at the Cat Expert blog provide a rather comprehensive overview of the piece. The authors compiled data on pet hospitalizations for a Boston-area clinic during the Patriots' 2007 season during game days. They ultimately found that a total of 33.8 pets were hospitalized on non-game day Sundays while 33.3 pets were admitted into the clinic on Sundays when the Patriots played, and a game's significance did not affect hospitalization rates. They conclude that while they cannot definitively make a statement on the impact of NFL games on pet hospitalizations, "staffing alterations may be warranted" in light of these findings.