During the Jason Boltus discussion on the Ti-Cat board I referred to some casual research i had done on football QB's. Here's an interesting bit of research that shows that football powerhouses don't tend to produce top QB's. Note- this is simply a statistical analysis but it says some pretty interesting things.
Whole Article- http://www.holycross.edu/departments/ec ... BDraft.pdf
NFL Productivity Outcomes
Table 11, and Figures 8 and 9 show various measures of passer productivity by
year in the league, as well as the percentage of drafted QBs that made a Pro Bowl roster.
The measures used here are the NFL’s passer rating measure, as well as Football
Outsiders’ DVOA and DPAR indices. All three measures tell essentially the same story:
QB performance peaks around Year 4 or 5, and begins to decline sometime thereafter.
However, it should be noted, as shown by Figure 10, that less than two-thirds of QBs –
even first-rounders - are still with the team that drafted them by in Year 5. As suggested
by Table 12, Super Bowl QBs can come from just about anywhere in the draft, but they
on average have nearly six years of NFL experience.9 These regressions were tested for heteroskedasticity using the White test (White, 1980); none was found
at the 10% level. Furthermore, censored (on the left at zero) TOBIT regressions were also run; they are not reported here, but gave coefficients and R2 values very similar to those indicated in Table 9.
11Tables 13, 14, 15, and 16 show the results of more systematic testing of the
hypothesis that QBs drafted higher and paid higher signing bonuses are more productive.
The OLS model results shown in Table 13 shows no evidence of any statistically
significant relationship between NFL career passer rating (minimum 50 attempts) and
either draft order or signing bonus (in $2007).10
Table 14 repeats these regressions using average season DVOA (minimum three
seasons with minimum 100 pass attempts). While there is some suggestion that a
relationship between overall draft order and DVOA might exist, top 10 picks overall in
the data set are significantly negatively associated with DVOA; i.e., the very top picks
underperformed their peers, ceteris paribus. While the coefficient on the overall draft
order number is also counterintuitively negative, it is not significant. Table 15 reports
similar OLS regressions using DPAR as the dependent variable, and finds similar results,
although DPAR is negatively and significantly associated with signing bonus. Finally,
Table 16 reports LOGIT regressions of a dummy variable =1 if the player ever made a
Pro Bowl roster (0 otherwise). The regressions indicate no significant relationship
between making the Pro Bowl with neither overall draft order nor signing bonus.
Table 17 breaks down QB draft choices and eventual NFL productivity by college
conference. The SEC and PAC 10 conferences saw the most quarterbacks drafted
between 1999 and 2004, but they cannot boast the most productive quarterbacks. Indeed,
draftees from non-BCS schools – comprising about 30% of all draftees – showed
significantly greater average NFL career passer ratings than did draftees from BCS
schools (equality of means t = 2.28). Average seasonal DVOA and DPAR were also
10 All the models in Tables 13-15 and in Table 18 were tested and corrected if necessary for
heteroskedasticity at the 10% level in accordance with White (1980).
12 higher on average for the non-BCS draftee group, but the number of QBs with three or
more seasons of at least 100 pass attempts was too small to find statistically significantly
With that said, no Div. III schools produced an NFL QB for purposes of the study.