(07-24-2014 10:53 AM)Topkat Wrote:  

(07-23-2014 10:24 PM)1845 Bear Wrote:  

(07-23-2014 09:41 PM)cuseroc Wrote:  SEC
BIG
ACC
PAC 12
Big !2

Yet another article takes uneven data and draws a wrong conclusion. Apparently the top half of appearances per school in the ACC and PAC drawing a slightly better average than our top 75% where some of our weakest games drag the average down is some fair comparison?

Having games like Raycom's ULM@Wake not drag the ACC average down (not nationally rated due to syndication) and have games like ULM@Baylor on FS1 drag ours down does not say anything about league viewership strength. It's not a fair comparison as it is apples and oranges.

The Big 12 has over 9 appearances per school, most of the other leagues are around 6-7 and the almost all the G5 have less. When you take the Big 12's top 60 appearances (6 per team) it outdoes the top 6 for everyone except the B1G and SEC.

I will post more complete data later (away from my computer right now) but this data is pretty misleading based on very different amounts of games reaching nationally rated networks.

You can only go with games that are televised. He does a pretty good job with the analysis given you are never going to compare apples to apples.

I never understood people wanting to throw out games. Why do you get to throw out ULM/Baylor while the other leagues do not get to throw out televised games (Fl Atl/Miami, Buffalo/Ohio St, UAB/LSU, Kent St/LSU, Troy/Mississippi, Idaho/Florida St)? All those games were on the networks included in the op stats. There are more than this that are questionable using your logic.

I believe he does not include regional games because they only show one rating per time slot. If ABC had regional games Texas/Iowa St (Central and West Time Zones) and Fl St/UNC (Eastern Time Zone), only one rating would show. How you split that is the problem. Obviously Texas/Iowa St would not have captured the east coast like Fl St/UNC; Fl St/UNC would not have captured the Central Time Zone like Texas/Iowa St.

But the point is that it is meaningless if you try to compare schools and conferences while taking out the best rated game of the day for the Big 10, Big 12 and ACC schools. Normally that afternoon time slot has the best ratings during the regular season.

You can compare how schools in the same conference do, but you can't compare the SEC to the rest when you exclude that.

A more valid comparison would be, for example, taking the ESPN games and comparing how the various teams do. But of course, that is complicated by opposing games, the time and the day.

(07-24-2014 01:46 PM)bullet Wrote:  

(07-24-2014 10:53 AM)Topkat Wrote:  

(07-23-2014 10:24 PM)1845 Bear Wrote:  

(07-23-2014 09:41 PM)cuseroc Wrote:  SEC
BIG
ACC
PAC 12
Big !2

Yet another article takes uneven data and draws a wrong conclusion. Apparently the top half of appearances per school in the ACC and PAC drawing a slightly better average than our top 75% where some of our weakest games drag the average down is some fair comparison?

Having games like Raycom's ULM@Wake not drag the ACC average down (not nationally rated due to syndication) and have games like ULM@Baylor on FS1 drag ours down does not say anything about league viewership strength. It's not a fair comparison as it is apples and oranges.

The Big 12 has over 9 appearances per school, most of the other leagues are around 6-7 and the almost all the G5 have less. When you take the Big 12's top 60 appearances (6 per team) it outdoes the top 6 for everyone except the B1G and SEC.

I will post more complete data later (away from my computer right now) but this data is pretty misleading based on very different amounts of games reaching nationally rated networks.

You can only go with games that are televised. He does a pretty good job with the analysis given you are never going to compare apples to apples.

I never understood people wanting to throw out games. Why do you get to throw out ULM/Baylor while the other leagues do not get to throw out televised games (Fl Atl/Miami, Buffalo/Ohio St, UAB/LSU, Kent St/LSU, Troy/Mississippi, Idaho/Florida St)? All those games were on the networks included in the op stats. There are more than this that are questionable using your logic.

I believe he does not include regional games because they only show one rating per time slot. If ABC had regional games Texas/Iowa St (Central and West Time Zones) and Fl St/UNC (Eastern Time Zone), only one rating would show. How you split that is the problem. Obviously Texas/Iowa St would not have captured the east coast like Fl St/UNC; Fl St/UNC would not have captured the Central Time Zone like Texas/Iowa St.

But the point is that it is meaningless if you try to compare schools and conferences while taking out the best rated game of the day for the Big 10, Big 12 and ACC schools. Normally that afternoon time slot has the best ratings during the regular season.

You can compare how schools in the same conference do, but you can't compare the SEC to the rest when you exclude that.

A more valid comparison would be, for example, taking the ESPN games and comparing how the various teams do. But of course, that is complicated by opposing games, the time and the day.

Just read some of HokieMark's post. What he is doing is much more valid.

(07-24-2014 12:38 PM)Hokie Mark Wrote:  Effect of TV Network:

Code:

    home team conference                        
Netwk    ACC    B1G    G5    ND    PAC    SEC    XII
ABC      2.8    3.3    2.5    2.5    2.5    4.2    3.7
CBS      ---    ---    2.0    ---    ---    4.5    ---
ESPN     2.4    2.6    1.4    4.3    2.0    2.6    2.7
ESPN2      0.7    0.9    0.5    ---    0.8    1.3    0.6
ESPNews    0.2    ---    0.1    ---    0.1    0.2    ---
ESPNU    0.3    0.3    0.2    ---    0.4    0.4    0.2
FOX       ---    6.4        2.0    1.3    3.9    1.7
Fox News    ---    ---    ---    ---    ---    ---    1.0
FS1      0.3    0.3    0.1    ---    0.5    0.3    0.4
FS2      ---    ---    ---    ---    ---    ---    0.1
NBC      ---    3.2    1.6    2.2    2.3    ---    2.6
Avg      1.1    2.4    1.1    2.7    1.2    2.2    1.4

ABC-only numbers show Big XII jumping all the way to 2nd place - impressive, HOWEVER... it's only 4 games:
Oklahoma vs. Oklahoma St
Baylor vs. Oklahoma St
Oklahoma vs. Texas
Mississippi St vs. Oklahoma St

No games involving the bottom 60% of the conference at all. That also skews the results - in favor of the Big XII.

ESPN may be the best indicator, since just about everybody plays on ESPN. Based on that, there simply isn't a whole lot of difference between one conference and another within the P5.

Note that the B1G rating on Fox was the Big 10 championship game. That was the only Fox game in the top 25. All the rest were CBS, ABC or ESPN games, showing the advantage of CBS and the mouse. FS1 ratings were similar to ESPNU.

I have gone and made two common sense adjustments and then ranked the leagues based on a similar percentage of games.

1- Split Coverage: If an ABC game had split coverage like the following 2 games I adjusted it based on % of tv households in the markets that got it on ABC (HH x ABC viewers), got it on the mirror station (% of mirror broadcast households X mirror viewers), and teams got no credit for the markets where the game was banished to ESPN3/gameplan like game #2 is here was in the Big 12 footprint. Obviously it's not perfect but it's a much closer estimation than the current data of how many viewers each game got. For instance game 1 between Ohio State and San Diego State had 42% of the nation's tv sets get it on ABC and 58% get it on ESPN2. So they got 42% of the ABC viewers and 58% of the ESPN2 Mirror Coverage Viewers.




2- Rain Delay Adjustment:

Games that had to be moved off of their normal spot for part of the broadcast due to delays show two ratings on the SMW data. I took the rating from the channel they were intended to run on and omitted the alternate channel data. For instance, like Mizzou-Toledo which was supposed to be ESPNU also had a rating for the hour it was on ESPNEWS. For this list only what they averaged on ESPNU is taken into consideration.


From there I compiled every game that a league participated in and sorted them by largest number of viewers to smallest.

The big issue I have with the data as the blog presented it is that it takes different percentages of each league's games.

SEC- 61.61%
B1G- 46.88%
Big 12- 73.33%
ACC- 49.11%
PAC12- 54.95%
AAC- 47.5%
MWC- 30.93%

So comparing very different percentages is not going to yield any good data unless things are at least somewhat similar.

So I took the highest viewed number of games needed to bring everyone down to a similar percentage to what the B1G and AAC had. On all of them it is roughly the top half of their games, which is where the money is for tv contracts anyway.

Everyone else was brought to average around 46-50% of their total games (the highest I could go with the P5 due to limited # of B1G games) not including league title games for the P5 or anyone's bowls as each of those games carries some natural ratings expectation that isn't an even comparison.

SEC: 53 games, average of 4,300,461 viewers per game. 46.8% of 112 total games
B1G: 45 games, average of 2,766,214 viewers per game. 46.8% of 96 total games.
Big12: 37 games, average of 2,181,901 viewers per game. 49.33% of 75 total games.
ACC: 53 games, average of 1,992,091 viewers per game. 47.32% of 112 total games.
Pac12: 45 games, average of 1,974,786 viewers per game. 49.45% of 91 total games. (9 game league slate means 6 fewer than B1G but USC got an extra due to playing at Hawaii)
AAC: 38 games, 938,450 viewers per game, 47.5% of 80 total games.
MWC: Not enough data for a valid comparison.

I'll post school specific data later. I've got it organized where you can see it by each team's top 1, 2, 3, 4, etc all the way up to 11 for those who had that many. I just need to find a clear way to paste it here.

Strain at a gnat to swallow a camel. The only statistics when it comes to these markets that matter is % of saturation within your own footprint and the individual school's market share for all broadcast games. The rest is just a way to game the numbers.

The SEC has deeper penetration into its own footprint than any other conference. But in those numbers (and I haven't seen this year's) the Big 12 and Big 10 are very close behind and in that order if I remember correctly. The ACC has the largest footprint and the poorest penetration of those 4. I don't recall if the PAC was ahead of them or not in this regard.

Outside of that the next relevant stat is the ranking of the most viewed schools.

Every conference is helped by not including bottom dwellers in the data in the OP's piece. Isn't it great when viewers and wins and losses help to cull the dead wood by the absence of those schools from the TV lineup. I imagine the Big 12 was hurt by these numbers because of how many regional games involved statistically smaller fan bases. In a 10 team conference with only two national football brands that should be expected. But the saturation numbers of the Big 12 show a strength almost as strong as the SEC.

The value of being on ABC just dropped dramatically.
New ABC College Football Analyst.
http://espn.go.com/college-football/stor...ball-games

(07-24-2014 04:47 PM)JRsec Wrote:  Strain at a gnat to swallow a camel. The only statistics when it comes to these markets that matter is % of saturation within your own footprint and the individual school's market share for all broadcast games. The rest is just a way to game the numbers.

The SEC has deeper penetration into its own footprint than any other conference. But in those numbers (and I haven't seen this year's) the Big 12 and Big 10 are very close behind and in that order if I remember correctly. The ACC has the largest footprint and the poorest penetration of those 4. I don't recall if the PAC was ahead of them or not in this regard..

Saturation in your market certainly matters but national brand also matters.

Quote:Every conference is helped by not including bottom dwellers in the data in the OP's piece.

The bottom half of the Big 12's ratings performers this season (ISU, KU, KSU, TCU, WVU) had a combined 44 appearances, almost half.

The other leagues had nowhere near that much. We weren't helped by that.

Quote:I imagine the Big 12 was hurt by these numbers because of how many regional games involved statistically smaller fan bases. In a 10 team conference with only two national football brands that should be expected. But the saturation numbers of the Big 12 show a strength almost as strong as the SEC.

The big issue is that early season winning/losing can dramatically affect viewership and UT, KSU, OU, OSU, WVU, ISU, KU, and TCU all had issues by midseason that dampened their draw and BU/TTU were really the only teams with any unbeaten momentum that could have challenged for a spot in the title game.

You are right about the way the Big 12 carries it's own footprint extremely well.

(07-24-2014 11:18 AM)cuseroc Wrote:  It just seems that for the last several seasons the ratings numbers keep showing the conferences in the same order as far as top football ratings. And you Big 12 guys keep supposedly "debunking" the methodology. These ratings are just as legit for fb as they are when they do the basketball ratings. Yet, for some reason, no one ever attempts to "debunk" the bb ratings.

The methodology IS bad. There's no other way to put it. A rigorous statistical method would try to find SOME way to normalize the data (e.g., compare all 3:30 ABC games against each other, and then see how the conferences compare each other that way).

Now, I'm not arguing the normalization would be easy (there are a number of other factors also worth considering --- how many other networks are showing games at the time? Is there another game in our own conference, such that that game is more likely to be cannibalizing viewers from our own broadcast? Et cetera, et cetera).

Boosted regression may be the best. Not easy to set up though ....

(07-24-2014 02:58 PM)1845 Bear Wrote:  I have gone and made two common sense adjustments and then ranked the leagues based on a similar percentage of games.

1- Split Coverage: If an ABC game had split coverage like the following 2 games I adjusted it based on % of tv households in the markets that got it on ABC (HH x ABC viewers), got it on the mirror station (% of mirror broadcast households X mirror viewers), and teams got no credit for the markets where the game was banished to ESPN3/gameplan like game #2 is here was in the Big 12 footprint. Obviously it's not perfect but it's a much closer estimation than the current data of how many viewers each game got. For instance game 1 between Ohio State and San Diego State had 42% of the nation's tv sets get it on ABC and 58% get it on ESPN2. So they got 42% of the ABC viewers and 58% of the ESPN2 Mirror Coverage Viewers.




2- Rain Delay Adjustment:

Games that had to be moved off of their normal spot for part of the broadcast due to delays show two ratings on the SMW data. I took the rating from the channel they were intended to run on and omitted the alternate channel data. For instance, like Mizzou-Toledo which was supposed to be ESPNU also had a rating for the hour it was on ESPNEWS. For this list only what they averaged on ESPNU is taken into consideration.


From there I compiled every game that a league participated in and sorted them by largest number of viewers to smallest.

The big issue I have with the data as the blog presented it is that it takes different percentages of each league's games.

SEC- 61.61%
B1G- 46.88%
Big 12- 73.33%
ACC- 49.11%
PAC12- 54.95%
AAC- 47.5%
MWC- 30.93%

So comparing very different percentages is not going to yield any good data unless things are at least somewhat similar.

So I took the highest viewed number of games needed to bring everyone down to a similar percentage to what the B1G and AAC had. On all of them it is roughly the top half of their games, which is where the money is for tv contracts anyway.

Everyone else was brought to average around 46-50% of their total games (the highest I could go with the P5 due to limited # of B1G games) not including league title games for the P5 or anyone's bowls as each of those games carries some natural ratings expectation that isn't an even comparison.

SEC: 53 games, average of 4,300,461 viewers per game. 46.8% of 112 total games
B1G: 45 games, average of 2,766,214 viewers per game. 46.8% of 96 total games.
Big12: 37 games, average of 2,181,901 viewers per game. 49.33% of 75 total games.
ACC: 53 games, average of 1,992,091 viewers per game. 47.32% of 112 total games.
Pac12: 45 games, average of 1,974,786 viewers per game. 49.45% of 91 total games. (9 game league slate means 6 fewer than B1G but USC got an extra due to playing at Hawaii)
AAC: 38 games, 938,450 viewers per game, 47.5% of 80 total games.
MWC: Not enough data for a valid comparison.

I'll post school specific data later. I've got it organized where you can see it by each team's top 1, 2, 3, 4, etc all the way up to 11 for those who had that many. I just need to find a clear way to paste it here.

So the AAC is getting 10% of the money, but doing about 50% of the viewers. Of course, they may get the same number of viewers showing poker and that may be even cheaper.

(07-24-2014 06:17 PM)Hokie Mark Wrote:  The value of being on ABC just dropped dramatically.
New ABC College Football Analyst.
http://espn.go.com/college-football/stor...ball-games

Guess Butch and Mack won't talk about UNC's academic scandals.

(07-24-2014 01:51 PM)bullet Wrote:  

(07-24-2014 01:46 PM)bullet Wrote:  

(07-24-2014 10:53 AM)Topkat Wrote:  

(07-23-2014 10:24 PM)1845 Bear Wrote:  

(07-23-2014 09:41 PM)cuseroc Wrote:  SEC
BIG
ACC
PAC 12
Big !2

Yet another article takes uneven data and draws a wrong conclusion. Apparently the top half of appearances per school in the ACC and PAC drawing a slightly better average than our top 75% where some of our weakest games drag the average down is some fair comparison?

Having games like Raycom's ULM@Wake not drag the ACC average down (not nationally rated due to syndication) and have games like ULM@Baylor on FS1 drag ours down does not say anything about league viewership strength. It's not a fair comparison as it is apples and oranges.

The Big 12 has over 9 appearances per school, most of the other leagues are around 6-7 and the almost all the G5 have less. When you take the Big 12's top 60 appearances (6 per team) it outdoes the top 6 for everyone except the B1G and SEC.

I will post more complete data later (away from my computer right now) but this data is pretty misleading based on very different amounts of games reaching nationally rated networks.

You can only go with games that are televised. He does a pretty good job with the analysis given you are never going to compare apples to apples.

I never understood people wanting to throw out games. Why do you get to throw out ULM/Baylor while the other leagues do not get to throw out televised games (Fl Atl/Miami, Buffalo/Ohio St, UAB/LSU, Kent St/LSU, Troy/Mississippi, Idaho/Florida St)? All those games were on the networks included in the op stats. There are more than this that are questionable using your logic.

I believe he does not include regional games because they only show one rating per time slot. If ABC had regional games Texas/Iowa St (Central and West Time Zones) and Fl St/UNC (Eastern Time Zone), only one rating would show. How you split that is the problem. Obviously Texas/Iowa St would not have captured the east coast like Fl St/UNC; Fl St/UNC would not have captured the Central Time Zone like Texas/Iowa St.

But the point is that it is meaningless if you try to compare schools and conferences while taking out the best rated game of the day for the Big 10, Big 12 and ACC schools. Normally that afternoon time slot has the best ratings during the regular season.

You can compare how schools in the same conference do, but you can't compare the SEC to the rest when you exclude that.

A more valid comparison would be, for example, taking the ESPN games and comparing how the various teams do. But of course, that is complicated by opposing games, the time and the day.

Just read some of HokieMark's post. What he is doing is much more valid.

I'll agree with that given the regional games excluded for the B1G and ACC.

If anyone has a beef about not including regional games it should be the Big 10 (11 games not included, 9 conference games).

The ACC had 8 regional games not counted (6 conference games).

The Big 12 only 3 (2 conference games).

Far and away, the Big 12 had the highest percentage of their teams games on free tv (27 games in only a 10 team league) that were exclusive broadcasts. I think the Big 12 should do much better given the free (over the air) exposure.

I hate to say it, but HokieMarks data is just ESPN. The free over the air ratings are what propel the SEC above the rest.

(07-24-2014 02:58 PM)1845 Bear Wrote:  I have gone and made two common sense adjustments and then ranked the leagues based on a similar percentage of games.

1- Split Coverage: If an ABC game had split coverage like the following 2 games I adjusted it based on % of tv households in the markets that got it on ABC (HH x ABC viewers), got it on the mirror station (% of mirror broadcast households X mirror viewers), and teams got no credit for the markets where the game was banished to ESPN3/gameplan like game #2 is here was in the Big 12 footprint. Obviously it's not perfect but it's a much closer estimation than the current data of how many viewers each game got. For instance game 1 between Ohio State and San Diego State had 42% of the nation's tv sets get it on ABC and 58% get it on ESPN2. So they got 42% of the ABC viewers and 58% of the ESPN2 Mirror Coverage Viewers.




2- Rain Delay Adjustment:

Games that had to be moved off of their normal spot for part of the broadcast due to delays show two ratings on the SMW data. I took the rating from the channel they were intended to run on and omitted the alternate channel data. For instance, like Mizzou-Toledo which was supposed to be ESPNU also had a rating for the hour it was on ESPNEWS. For this list only what they averaged on ESPNU is taken into consideration.


From there I compiled every game that a league participated in and sorted them by largest number of viewers to smallest.

The big issue I have with the data as the blog presented it is that it takes different percentages of each league's games.

SEC- 61.61%
B1G- 46.88%
Big 12- 73.33%
ACC- 49.11%
PAC12- 54.95%
AAC- 47.5%
MWC- 30.93%

So comparing very different percentages is not going to yield any good data unless things are at least somewhat similar.

So I took the highest viewed number of games needed to bring everyone down to a similar percentage to what the B1G and AAC had. On all of them it is roughly the top half of their games, which is where the money is for tv contracts anyway.

Everyone else was brought to average around 46-50% of their total games (the highest I could go with the P5 due to limited # of B1G games) not including league title games for the P5 or anyone's bowls as each of those games carries some natural ratings expectation that isn't an even comparison.

SEC: 53 games, average of 4,300,461 viewers per game. 46.8% of 112 total games
B1G: 45 games, average of 2,766,214 viewers per game. 46.8% of 96 total games.
Big12: 37 games, average of 2,181,901 viewers per game. 49.33% of 75 total games.
ACC: 53 games, average of 1,992,091 viewers per game. 47.32% of 112 total games.
Pac12: 45 games, average of 1,974,786 viewers per game. 49.45% of 91 total games. (9 game league slate means 6 fewer than B1G but USC got an extra due to playing at Hawaii)
AAC: 38 games, 938,450 viewers per game, 47.5% of 80 total games.
MWC: Not enough data for a valid comparison.

I'll post school specific data later. I've got it organized where you can see it by each team's top 1, 2, 3, 4, etc all the way up to 11 for those who had that many. I just need to find a clear way to paste it here.

"The big issue I have with the data as the blog presented it is that it takes different percentages of each league's games."

This statement from your post raises the question:

so how are your adjustments above NOT taking different percentages of each leagues games?

You want to "fairly" adjust the regional games not included in the OP by factoring down how many viewers were in the televised footprint.

The Big 10 had 11 games (9 conference games) not counted in the OP.

The ACC had 8 games (6 conference games) not counted in the OP.

The Big 12 had only 3 games (2 conference games) not counted in the OP.

(07-24-2014 01:51 PM)bullet Wrote:  Just read some of HokieMark's post. What he is doing is much more valid.

(07-25-2014 07:40 AM)Topkat Wrote:  I hate to say it, but HokieMarks data is just ESPN. The free over the air ratings are what propel the SEC above the rest.

Pretty sure I included ALL networks, not just ESPN. At any rate, I agree - free TV (eg. CBS) puts SEC in front of ratings.

(07-25-2014 08:06 AM)Topkat Wrote:  

(07-24-2014 02:58 PM)1845 Bear Wrote:  I have gone and made two common sense adjustments and then ranked the leagues based on a similar percentage of games.

1- Split Coverage: If an ABC game had split coverage like the following 2 games I adjusted it based on % of tv households in the markets that got it on ABC (HH x ABC viewers), got it on the mirror station (% of mirror broadcast households X mirror viewers), and teams got no credit for the markets where the game was banished to ESPN3/gameplan like game #2 is here was in the Big 12 footprint. Obviously it's not perfect but it's a much closer estimation than the current data of how many viewers each game got. For instance game 1 between Ohio State and San Diego State had 42% of the nation's tv sets get it on ABC and 58% get it on ESPN2. So they got 42% of the ABC viewers and 58% of the ESPN2 Mirror Coverage Viewers.




2- Rain Delay Adjustment:

Games that had to be moved off of their normal spot for part of the broadcast due to delays show two ratings on the SMW data. I took the rating from the channel they were intended to run on and omitted the alternate channel data. For instance, like Mizzou-Toledo which was supposed to be ESPNU also had a rating for the hour it was on ESPNEWS. For this list only what they averaged on ESPNU is taken into consideration.


From there I compiled every game that a league participated in and sorted them by largest number of viewers to smallest.

The big issue I have with the data as the blog presented it is that it takes different percentages of each league's games.

SEC- 61.61%
B1G- 46.88%
Big 12- 73.33%
ACC- 49.11%
PAC12- 54.95%
AAC- 47.5%
MWC- 30.93%

So comparing very different percentages is not going to yield any good data unless things are at least somewhat similar.

So I took the highest viewed number of games needed to bring everyone down to a similar percentage to what the B1G and AAC had. On all of them it is roughly the top half of their games, which is where the money is for tv contracts anyway.

Everyone else was brought to average around 46-50% of their total games (the highest I could go with the P5 due to limited # of B1G games) not including league title games for the P5 or anyone's bowls as each of those games carries some natural ratings expectation that isn't an even comparison.

SEC: 53 games, average of 4,300,461 viewers per game. 46.8% of 112 total games
B1G: 45 games, average of 2,766,214 viewers per game. 46.8% of 96 total games.
Big12: 37 games, average of 2,181,901 viewers per game. 49.33% of 75 total games.
ACC: 53 games, average of 1,992,091 viewers per game. 47.32% of 112 total games.
Pac12: 45 games, average of 1,974,786 viewers per game. 49.45% of 91 total games. (9 game league slate means 6 fewer than B1G but USC got an extra due to playing at Hawaii)
AAC: 38 games, 938,450 viewers per game, 47.5% of 80 total games.
MWC: Not enough data for a valid comparison.

I'll post school specific data later. I've got it organized where you can see it by each team's top 1, 2, 3, 4, etc all the way up to 11 for those who had that many. I just need to find a clear way to paste it here.

"The big issue I have with the data as the blog presented it is that it takes different percentages of each league's games."

This statement from your post raises the question:

so how are your adjustments above NOT taking different percentages of each leagues games?

My adjustments are taking similar percentages of each league's total number of games played.

Quote:You want to "fairly" adjust the regional games not included in the OP by factoring down how many viewers were in the televised footprint.

I assume you are referring to the ABC Regional split coverage. On that it's more accurate to take a % of tv sets on the ABC broadcast and the mirror broadcast than to assume BOTH games of the split broadcast got ALL the viewers for the ABC network.

Quote:The Big 10 had 11 games (9 conference games) not counted in the OP.

The ACC had 8 games (6 conference games) not counted in the OP.

The Big 12 had only 3 games (2 conference games) not counted in the OP.

Two things on this part:

1- Where are you getting those numbers? That isn't remotely accurate. The Big 12 had 75 games overall and 19 did not make it to nationally rated networks. The B1G had 48 games on the BTN which didn't make it to nationally rated networks. The ACC had far more than your 8 total on each of espn3 or the raycom syndication.

2- Because of different TV deals that cause games to not get rated nationally (ACC syndicates games with Raycom or has ESPN3, BTN, P12N, Big12 tier 3 games) there are a lot of games with missing data. Typically these games are NOT the best matchups.

For instance in the ACC we saw games like New Mexico @ Pitt, ECU@NC State, Navy-Duke, ULM@Wake, ODU-Pitt, etc. You have some league games in there like Wake-Syracuse but it's hard to argue that game is going to draw as well as VT-UVA, Miami-GT, or others featuring a really big football brand. The games that did not reach national networks here are not the kind of games that are going to boost the ACC's numbers, frankly they are those that would tank them.

The Big 12 has a the games appearing on each team's T3 outlet (12 total. 1 for each team and the 2 games UT negotiated for with the league, FOX, and ESPN which ended up being home games vs Ole Miss and KU.) It also had 7 games relegated to FSN which included 5 league games. With the exception of UT-Ole Miss it was all the least valuable games.

Now the B1G and PAC have a little more argument as some decent picking orders were tied into their creation of their networks so they don't simply pick last. You had decent games like UCF-PSU, Nebraska-PSU, and others mixed in with a ton of the worst teams showing up. It's still mostly the lower stuff but it isn't quite as dramatic as with the ACC or Big 12 where the worst is what is not rated.


The PAC12 had some decent games on their network (UW-ASU, Stanford-Utah, BC-USC) but also a ton of the least valuable ones (9 FCS games, several G5 games, 10 games featuring horrific Colorado)


So with very rare exception it's the least desirable games that aren't making it to nationally rated networks. When you average the difference in viewership between the rare exceptions and the lowest nationally rated games over 38-55 games it isn't going to dramatically skew it anyway. The data isn't perfect because we don't have everything but it's a step in the right direction from the original data.

You know, it occurs to me that the most national games any conference is likely to get is 3 on Saturday plus one more during the week. Thus, maybe the best comparison (much hard to do, though) is to take the best game from each league in each time slot (if available). Otherwise you'll always be dealing with missing data.

(07-25-2014 09:08 AM)Hokie Mark Wrote:  

(07-24-2014 01:51 PM)bullet Wrote:  Just read some of HokieMark's post. What he is doing is much more valid.

(07-25-2014 07:40 AM)Topkat Wrote:  I hate to say it, but HokieMarks data is just ESPN. The free over the air ratings are what propel the SEC above the rest.

Pretty sure I included ALL networks, not just ESPN. At any rate, I agree - free TV (eg. CBS) puts SEC in front of ratings.

My bad and apologies, I tried to go off memory instead of re-reading your post.

(07-25-2014 09:27 AM)1845 Bear Wrote:  

(07-25-2014 08:06 AM)Topkat Wrote:  

(07-24-2014 02:58 PM)1845 Bear Wrote:  I have gone and made two common sense adjustments and then ranked the leagues based on a similar percentage of games.

1- Split Coverage: If an ABC game had split coverage like the following 2 games I adjusted it based on % of tv households in the markets that got it on ABC (HH x ABC viewers), got it on the mirror station (% of mirror broadcast households X mirror viewers), and teams got no credit for the markets where the game was banished to ESPN3/gameplan like game #2 is here was in the Big 12 footprint. Obviously it's not perfect but it's a much closer estimation than the current data of how many viewers each game got. For instance game 1 between Ohio State and San Diego State had 42% of the nation's tv sets get it on ABC and 58% get it on ESPN2. So they got 42% of the ABC viewers and 58% of the ESPN2 Mirror Coverage Viewers.




2- Rain Delay Adjustment:

Games that had to be moved off of their normal spot for part of the broadcast due to delays show two ratings on the SMW data. I took the rating from the channel they were intended to run on and omitted the alternate channel data. For instance, like Mizzou-Toledo which was supposed to be ESPNU also had a rating for the hour it was on ESPNEWS. For this list only what they averaged on ESPNU is taken into consideration.


From there I compiled every game that a league participated in and sorted them by largest number of viewers to smallest.

The big issue I have with the data as the blog presented it is that it takes different percentages of each league's games.

SEC- 61.61%
B1G- 46.88%
Big 12- 73.33%
ACC- 49.11%
PAC12- 54.95%
AAC- 47.5%
MWC- 30.93%

So comparing very different percentages is not going to yield any good data unless things are at least somewhat similar.

So I took the highest viewed number of games needed to bring everyone down to a similar percentage to what the B1G and AAC had. On all of them it is roughly the top half of their games, which is where the money is for tv contracts anyway.

Everyone else was brought to average around 46-50% of their total games (the highest I could go with the P5 due to limited # of B1G games) not including league title games for the P5 or anyone's bowls as each of those games carries some natural ratings expectation that isn't an even comparison.

SEC: 53 games, average of 4,300,461 viewers per game. 46.8% of 112 total games
B1G: 45 games, average of 2,766,214 viewers per game. 46.8% of 96 total games.
Big12: 37 games, average of 2,181,901 viewers per game. 49.33% of 75 total games.
ACC: 53 games, average of 1,992,091 viewers per game. 47.32% of 112 total games.
Pac12: 45 games, average of 1,974,786 viewers per game. 49.45% of 91 total games. (9 game league slate means 6 fewer than B1G but USC got an extra due to playing at Hawaii)
AAC: 38 games, 938,450 viewers per game, 47.5% of 80 total games.
MWC: Not enough data for a valid comparison.

I'll post school specific data later. I've got it organized where you can see it by each team's top 1, 2, 3, 4, etc all the way up to 11 for those who had that many. I just need to find a clear way to paste it here.

"The big issue I have with the data as the blog presented it is that it takes different percentages of each league's games."

This statement from your post raises the question:

so how are your adjustments above NOT taking different percentages of each leagues games?

My adjustments are taking similar percentages of each league's total number of games played.

Quote:You want to "fairly" adjust the regional games not included in the OP by factoring down how many viewers were in the televised footprint.

I assume you are referring to the ABC Regional split coverage. On that it's more accurate to take a % of tv sets on the ABC broadcast and the mirror broadcast than to assume BOTH games of the split broadcast got ALL the viewers for the ABC network.

Quote:The Big 10 had 11 games (9 conference games) not counted in the OP.

The ACC had 8 games (6 conference games) not counted in the OP.

The Big 12 had only 3 games (2 conference games) not counted in the OP.

Two things on this part:

1- Where are you getting those numbers? That isn't remotely accurate. The Big 12 had 75 games overall and 19 did not make it to nationally rated networks. The B1G had 48 games on the BTN which didn't make it to nationally rated networks. The ACC had far more than your 8 total on each of espn3 or the raycom syndication.

2- Because of different TV deals that cause games to not get rated nationally (ACC syndicates games with Raycom or has ESPN3, BTN, P12N, Big12 tier 3 games) there are a lot of games with missing data. Typically these games are NOT the best matchups.

For instance in the ACC we saw games like New Mexico @ Pitt, ECU@NC State, Navy-Duke, ULM@Wake, ODU-Pitt, etc. You have some league games in there like Wake-Syracuse but it's hard to argue that game is going to draw as well as VT-UVA, Miami-GT, or others featuring a really big football brand. The games that did not reach national networks here are not the kind of games that are going to boost the ACC's numbers, frankly they are those that would tank them.

The Big 12 has a the games appearing on each team's T3 outlet (12 total. 1 for each team and the 2 games UT negotiated for with the league, FOX, and ESPN which ended up being home games vs Ole Miss and KU.) It also had 7 games relegated to FSN which included 5 league games. With the exception of UT-Ole Miss it was all the least valuable games.

Now the B1G and PAC have a little more argument as some decent picking orders were tied into their creation of their networks so they don't simply pick last. You had decent games like UCF-PSU, Nebraska-PSU, and others mixed in with a ton of the worst teams showing up. It's still mostly the lower stuff but it isn't quite as dramatic as with the ACC or Big 12 where the worst is what is not rated.


The PAC12 had some decent games on their network (UW-ASU, Stanford-Utah, BC-USC) but also a ton of the least valuable ones (9 FCS games, several G5 games, 10 games featuring horrific Colorado)


So with very rare exception it's the least desirable games that aren't making it to nationally rated networks. When you average the difference in viewership between the rare exceptions and the lowest nationally rated games over 38-55 games it isn't going to dramatically skew it anyway. The data isn't perfect because we don't have everything but it's a step in the right direction from the original data.

From the same place the OP article says they used, Sports Media Watch.

I'm not sure why you are bringing Raycom, BTN and ESPN3 into the conversation at all since they are not used by the OP article.

Maybe re-read the article and explain why you want to intoduce more network sources.

The Big 12 had the highest percentage (10 team conference) on free tv (Fox, CBS, NBC, ABC) non-regional games. Why complain about who is televised when you get all those additional TV sets counted?

(07-25-2014 10:10 AM)Topkat Wrote:  From the same place the OP article says they used, Sports Media Watch.

That data was the starting point for my numbers as well.

Quote:I'm not sure why you are bringing Raycom, BTN and ESPN3 into the conversation at all since they are not used by the OP article.

I brought them in to point out that the number of games not counted was very different than what you were claiming. Also it's why some leagues have very different % of games played as chunks of inventory went to those outlets. In some cases that inventory is parallel to what other league's had lumped into the average numbers.

Quote:Maybe re-read the article and explain why you want to intoduce more network sources.

The Big 12 had the highest percentage (10 team conference) on free tv (Fox, CBS, NBC, ABC) non-regional games. Why complain about who is televised when you get all those additional TV sets counted?

Because the way people have ranked the leagues is in terms of AVERAGE viewers. When one league has 25% of it's lesser games tanking the average it invalidates the results. You won't be comparing similar samples.

I am not arguing to introduce more sources because they simply don't have published ratings information.

My argument is that in order to get any kind of simple league average that could be a valid comparison using this data you need to take a similar percentage of each league's inventory and then compare- which is what I did.

I don't care how the best 47% of the B1G's inventory compares to 62% of the SEC's... I care about apples to apples instead of apples to oranges.

It is my understanding that in 2014, when the AAC's new ESPN TV deal kicks in, the exposure (and subsequently the ratings) of that conference will substantially improve, especially after 2015 when Navy joins.

(07-25-2014 10:33 AM)1845 Bear Wrote:  

(07-25-2014 10:10 AM)Topkat Wrote:  From the same place the OP article says they used, Sports Media Watch.

That data was the starting point for my numbers as well.

Quote:I'm not sure why you are bringing Raycom, BTN and ESPN3 into the conversation at all since they are not used by the OP article.

I brought them in to point out that the number of games not counted was very different than what you were claiming. Also it's why some leagues have very different % of games played as chunks of inventory went to those outlets. In some cases that inventory is parallel to what other league's had lumped into the average numbers.

Quote:Maybe re-read the article and explain why you want to intoduce more network sources.

The Big 12 had the highest percentage (10 team conference) on free tv (Fox, CBS, NBC, ABC) non-regional games. Why complain about who is televised when you get all those additional TV sets counted?

Because the way people have ranked the leagues is in terms of AVERAGE viewers. When one league has 25% of it's lesser games tanking the average it invalidates the results. You won't be comparing similar samples.

I am not arguing to introduce more sources because they simply don't have published ratings information.

My argument is that in order to get any kind of simple league average that could be a valid comparison using this data you need to take a similar percentage of each league's inventory and then compare- which is what I did.

I don't care how the best 47% of the B1G's inventory compares to 62% of the SEC's... I care about apples to apples instead of apples to oranges.

Well, we are never going to get an apple to apple comparison.

The league that has 25% of its games tanking its average (your quote) also has the highest percentage of its members games televised on Free over-the-air NATIONAL TV (ABC, NBC, CBS, Fox).

It would seem you do not care for the inbalance of individual team selection for TV games, but have no problem with the inbalance in National vs. Regional tv sets to draw from if it suits your purpose.

A similar percentage of each leagues inventory seems fine if you can adjust a disproportionate amount of national REGIONAL games (3x-4x more) on Fox, ABC, NBC or ABC for a conference not called the B12, eh?

It's too bad you can't see you are doing exactly what you despise from the OP.