(05-02-2021 06:37 PM)Atlanta Wrote: (05-02-2021 04:52 PM)4xGrad Wrote: (05-01-2021 08:44 PM)jedclampett Wrote: [quote='4xGrad' pid='17407020' dateline='1619915350']
[quote='jedclampett' pid='17406989' dateline='1619912658']
5-1-6 model applied to 2020 CFP rankings:
1st round bye: Alabama, Clemson, OSU, ND
1st round games:
#5 TAMU vs. #12 Coastal Carolina
#6 Oklahoma vs. #11 Indiana
#7 Florida vs. #10 Iowa State
#8 Cincinnati vs. #9 Georgia
.
But as you yourself have argued, If that 5-1-6 were in place.... p5 bias would have dropped the two G5's in ranking and Cincy probably would have had the hardest path to the top.
True.
I posted what the match-ups would have been, but offered no comment.
Cincy would have had the roughest path, and it doesn't seem fair, considering that they were undefeated and #6 in the AP polls. The CFP moved a couple of teams ahead of them despite the fact that unlike Cincy, they had lost at least one game.
(05-01-2021 07:29 PM)4xGrad Wrote:
If a G5 has a really good year they can raise to a ranking that will make it just barely into the mix as the season winds down while the committee hopes a last minute toe stub drops them out of contention.
That guaranteed spot is the only thing allowing 1 G5 into the rankings at all.
(05-01-2021 06:44 PM)jedclampett Wrote:
Right, but frankly, it would do them little good if they would end up as the #8 seed, playing the #1 team every year.
That's why I'm starting to wonder if the G5 and FBS independents might be better off if they were to organize an alternative to the abysmal CFP.
.
(05-01-2021 07:29 PM)4xGrad Wrote: I get it. I just really think that (as the G5) going our own way will be detrimental. I really think our common best interest is to outplay them at their own game. For our teams to become more relevant than theirs. That is going to be a long hard path but we have all been taking a long walk down it already. Look at what these teams have done over the last 20 years. 40 years ago which of these teams ever entered the national conversation at all?
Okay... Navy.... but who else?
(05-02-2021 06:37 PM)Atlanta Wrote: Memphis tied #2 Ole Miss in 1963, beat #11 Ms St, So Carolina, Louisville & UH & finished undefeated 9-0-1 & UPI ranked #14.That's probably our best ranking, our only undefeated season.
.
It may be helpful, in this regard, to examine the AAC's track record in more detail.
These have been the highest-ranked AAC teams in the penultimate (second to last) AP and Coaches Poll (CP) Top 25 since 2014:
........Team (AP) (CFP rank)
2013 UCF (#15) (pre-CFP)
2014 No ranked AAC teams (Memphis was #25 in Final AP)
2015 Houston (#14 AP) (#16 CP) (CFP #18)
2016 AP: Temple (#22)^ (CFP #24)
2016 CP: USF (#22)* (CFP: unranked)
2017 UCF (#10) (CFP #12)
2018 UCF (#7) (Coaches: #6) (CFP #8)
2019 Memphis (#15) (CFP #17)
2020 Cincy (#6) (CFP #8)
^ Temple was also ranked in the CP (
average AP/CP rank: #23)
* USF was also ranked in the AP (
average AP/CP rank: #23.5)
https://www.sports-reference.com/cfb/con.../2015.html
Summary:
The AAC has had two top-ten CFP teams in the eight years of its existence (both were ranked #8).
Both of those top-ten CFP teams achieved top ten status in the past three seasons (in 2018 and 2020).
The AAC has had 3 teams with a final CFP ranking of #12 or higher. All three received those rankings in the past four seasons (2017, 2018, and 2020).
After getting off to a somewhat rough start, with Boise State and WMU representing the G5 in the NY6 bowls,
the AAC has had the highest-ranked CFP team for the past four years in a row, a promising indicator of the potential that future AAC teams may have to make the CFP in the years ahead.
However, it should be considered a matter of significant concern that the CFP rankings of the most highly-ranked AAC teams have been consistently lower (2 ranks lower, on average) than the teams' rankings were in the AP and Coaches polls that were conducted at the same point in time.
Implications:
The CFP rankings of the top AAC teams have trended upward over the past 7 years.
The top AAC teams have been consistently (every year in the history of the CFP rankings) been under-ranked by an average of 2 ranks relative to their rankings in the AP and Coaches' polls (e.g., Cincy was ranked #6 in the AP and Coaches Polls, but only #8 in the CFP rankings).
--This is strong evidence of systematic pro-P5 bias in the CFP rankings.
--The probability of a consistent under-estimate of the top AAC team's ranking - relative to the estimates of the sports writers and coaches happening by chance alone is .0078125 (less than 1 in 100):
...................................................................................................
........................................................................
Multiplicative
..........Probability of CFP under-ranking...............l
Probability of
........the top AAC team by chance alone:.............l
under-ranking top
......................................................................l
AAC team seven
AP/CP...(AP)....(CP)..AP/CP..AP/CP..AP/CP..AP/CP l
times in seven:
.2015...2016...2016..2017...2018....2019...2020..l
comparisons
......................................................................l
by chance alone:
. .50..* .50... * .50..* .50..* .50...* .50....* .50 =
.0078125 (p<.01)
...................................................................................................
Note: A p-level below 0.01 (i.e., a finding that would less than once in 100 trials is considered to be statistically (i.e., scientifically) significant.
Scientifically, such a finding would result in a rejection of the "null hypothesis" that there is no pro-P5/anti-AAC bias in the CFP.
But let's remember that we're not just speaking hypothetically in this case - - the data that we are discussing here are, in fact, valid quasi-experimental data.
--
The data were obtained from official records and archives that were reported by organizations and entities that had no prior knowledge of the fact that there would be research conducted with the data they reported.
--
The rankings that were obtained are highly valid, due to the fact that they were conducted by the highly experienced professionals who are tasked with their rankings each season and recognize that their rankings are taken very seriously.
--
Their rankings were 100% blind with respect to the present study, since they had no way of knowing that anyone would be doing research with the data (rankings) that they provided.
--While ranking data are, by nature, somewhat subjective and are influenced by many different factors,
this is a study of precisely that - - their judgements.
----Human judgments can be studied scientifically, just as any other type of human behavior can be investigated. Polls that assess political preferences and marketing research that assess product preferences are other examples.
--Moreover, while AP, CP, and CFP rankings are subjective, it is considered an honor to be asked to report their rankings, and those who provide their rankings recognize that if their rankings depart too far from reality, they might lose the privilege of submitting their rankings. While the possibility of bias cannot be ruled out, that type of discipline instills a certain degree of self-discipline and rigor.
.
Given that the data we're dealing with are well documented, valid, and blind to the purpose of the study, this is in fact, an example of an appropriate quasi-experimental data set.
Thus, the following assertions based on this small quasi-experimental study may be considered scientifically justifiable:
a)
The null hypothesis that there was no pro-P5/anti-AAC bias in the CFP rankings - relative to the contemporaneous AP and Coaches Poll rankings - between 2015 and 2020,
was rejected.
b)
The findings indicate that the CFP rankings between 2015 and 2020 have demonstrated a significant (p<.01) tendency to under-rank the top AAC teams .
c)
the extent of the bias has been relatively modest, averaging only 2 ranks, but not small enough to preclude the possibility of a significantly adverse effect on CFP access.
d)
It was the consistency of the pro-P5/anti-AAC discrepancies in the rankings, and not the magnitude of the differences that provided empirical evidence of pro-P5 bias vis a vis the top AAC teams between 2015 and 2020.
e)
If the discrepancies had been much less consistent than they were, the findings would not have been statistically significant.
.
Q:
But the CFP has only been under-ranking the top AAC teams by only 2 ranks. What possible difference could that make, going forward?
A:
It could easily make the difference between qualifying and not qualifying for the CFP, for example, if the top AAC team were to be ranked #4 by the AP and Coaches polls, but only #6 in the CFP rankings.
--Cincinnati was #6 in the AP and Coaches polls in 2020 in mid-December. If the Bearcats had been #4, the average 2 point under-ranking in the CFP rankings would have kept them out of the playoffs.
--The CFP rankings don't have to under-rank the top G5 by much (only 1 or 2 ranks in order to keep the G5 teams out of the CFP.
Q:
But this can't be considered scientifically valid evidence can it, since a formal experiment wasn't conducted?
A:
It can be and is considered scientifically valid. Scientific papers based on quasi-experimental data have been published in the most highly esteemed journals, including both Science and Nature.
Evidence of this kind (data resulting from what scientists refer to as a quasi-experiment, a.k.a. a "natural experiment")
may be considered scientifically valid if it is the best kind of evidence that can be obtained in circumstances that would not permit a controlled double-blind experiment with random assignment of subjects to experimental conditions to examine - for example - whether the CFP rankings have been biased (a valid double-blind study couldn't be conducted because those who do the rankings would know - and would not be blind to the fact - that it was a study about how they rank teams).
In addition, quasi-experimental studies are very often the only ethically impermissible way to study the effects of adverse events, such as nuclear detonations or pandemics, since the only way to do an experiment such as this would be to cause an epidemic so that the effects could be studied scientifically.
Fortunately, in such cases, the quasi-experimental method makes it possible to conduct scientific tests - for example - to investigate the effects of pandemics on cities that have been infected.
--Notably, such
quasi-experimental research has been extremely beneficial during the Covid pandemic. It has not only (a) provided researchers with a tremendous amount of information about the pandemic, but has also (b) led to important advancements in the treatment and prevention of Covid-19 possible.
----This shows that
quasi-experimental research isn't merely an abstruse field of academic interest. It is a highly important type of research with very practical applications.
.
If anyone would like to learn more about quasi-experimental research and its unique scientific importance, this is the classic text on the topic:
Amazon.com: Experimental and Quasi-Experimental Designs for Research (9780395307878): Campbell, Donald T., Stanley, Julian
https://www.amazon.com/Experimental-Quas...0395307872
.