On Seeding, RPI, and Conference scheduling

[Bear11]

If you've read many of my posts, you'll see that I have strong negative feelings about the RPI. Many assume this is because of my particular rooting interest, but it is really because it is a truly terrible way to rank teams. I hope this helps explain some of that.

I don't begrudge the committee's rankings, they did their job according to specified criteria. My problem is that the criteria themselves are wrong. The RPI favors teams with certain types of schedules, in particular the conferences with 16 games instead of 18. It is possible to "game" the system, and I believe the ACC and SEC have.

Let's imagine 2 conferences, identical in every way, except one has 10 teams, with an 18 game double round robin schedule, and the other 15 teams, playing 16 games and only repeating 2 opponents. Assume each team plays exactly 30 games. The non-conference record for each of the 5 major conferences was between 71% and 78%, so we will assume each conference wins 75% of their non-conference games so the math is easier.

Here is how the 10 team conference will fare:
10 teams*12 non-conference games each equals 120. They win 90 of them. The conference schedule is 90 games. The record for the entire conference is therefore 180-120, a .600 winning percentage.

Here is the 15 team conference:
15 teams*14 non-conference games equals 210 total non-conference games. They will win about 158. The conference schedule is 120 games. The record for the whole conference is 278-172, a .6177 winning percentage.

Therefore, the 16 game conferences have a nearly 2% advantage in RPI before the season starts. .0177 may seem like a small amount, but remember that 75% of the RPI formula is based on your opponent's record and their opponent's record. The overall record of the conference is therefore tremendously important, as they make up the majority of your schedule. Games played within the conference are zero-sum propositions, since every team plays both the winner and the loser. They cancel each other out from the perspective of the rest of the conference. This makes non-conference games the key to raising conference RPI. Major conference teams tend to win most of their non-conference games, so playing more of them is essential to raising the overall record, and therefore the RPI.

This effect is magnified in the Big 12, where each opponent is played twice, doubling the importance of each non-conference game for the other teams in the conference. The margins are extremely thin in the RPI, especially in the middle. The difference between #50 and #70, according to realtimerpi, is .0189. An extra non-conference win for all of their conference mates may have made a big difference for Oklahoma State, West Virginia and TCU, who were all in that range. Perhaps it would've bumped them into the top 50, off the bubble and perhaps into the field. Sagarin actually has West Virginia in the top 25, but their RPI was hurt because they took so many losses to the teams above them in the Big 12. Take those games out, and replace them with a decent, but winnable non-conference game, and they likely make the tournament. In addition, if those 3 teams were in the top 50, it would've given Baylor 7 more top 50 wins, making their apparently weak schedule look a whole lot tougher.

Eight teams in the Big 12 took 4 losses to Texas and Baylor. In the ACC and SEC, most teams only play the other top teams once, so they avoid taking the extra loss. Since RPI only looks at record, and not the true quality of opponents, taking these extra losses is very harmful. I believe this is the reason that Sagarin and Massey each say that the Big 12 is underrated, because they dig deeper than just a superficial look at records. The SEC took a real beating this weekend, and every team that lost is rated higher in the RPI than Sagarin/Massey. I believe they are systematically overrated by the RPI. I don't know if that is the reason for the scheduling, but it seems to have worked out that way. This may not have been the Big 12's strongest year, but I don't think they got a fair shake. For reference, here is each major conference's non-conference winning percentage:

Big 12: .762
Pac 12: .719
ACC: .775
SEC: .754
Big 10: .713

Sorry for the long winded post, I spend way too much time thinking about these things.

 

[Plebe]

If you've read many of my posts, you'll see that I have strong negative feelings about the RPI. Many assume this is because of my particular rooting interest, but it is really because it is a truly terrible way to rank teams. I hope this helps explain some of that.

I don't begrudge the committee's rankings, they did their job according to specified criteria. My problem is that the criteria themselves are wrong. The RPI favors teams with certain types of schedules, in particular the conferences with 16 games instead of 18. It is possible to "game" the system, and I believe the ACC and SEC have.

Let's imagine 2 conferences, identical in every way, except one has 10 teams, with an 18 game double round robin schedule, and the other 15 teams, playing 16 games and only repeating 2 opponents. Assume each team plays exactly 30 games. The non-conference record for each of the 5 major conferences was between 71% and 78%, so we will assume each conference wins 75% of their non-conference games so the math is easier.

Here is how the 10 team conference will fare:
10 teams*12 non-conference games each equals 120. They win 90 of them. The conference schedule is 90 games. The record for the entire conference is therefore 180-120, a .600 winning percentage.

Here is the 15 team conference:
15 teams*14 non-conference games equals 210 total non-conference games. They will win about 158. The conference schedule is 120 games. The record for the whole conference is 278-172, a .6177 winning percentage.

Therefore, the 16 game conferences have a nearly 2% advantage in RPI before the season starts. .0177 may seem like a small amount, but remember that 75% of the RPI formula is based on your opponent's record and their opponent's record. The overall record of the conference is therefore tremendously important, as they make up the majority of your schedule. Games played within the conference are zero-sum propositions, since every team plays both the winner and the loser. They cancel each other out from the perspective of the rest of the conference. This makes non-conference games the key to raising conference RPI. Major conference teams tend to win most of their non-conference games, so playing more of them is essential to raising the overall record, and therefore the RPI.

This effect is magnified in the Big 12, where each opponent is played twice, doubling the importance of each non-conference game for the other teams in the conference. The margins are extremely thin in the RPI, especially in the middle. The difference between #50 and #70, according to realtimerpi, is .0189. An extra non-conference win for all of their conference mates may have made a big difference for Oklahoma State, West Virginia and TCU, who were all in that range. Perhaps it would've bumped them into the top 50, off the bubble and perhaps into the field. Sagarin actually has West Virginia in the top 25, but their RPI was hurt because they took so many losses to the teams above them in the Big 12. Take those games out, and replace them with a decent, but winnable non-conference game, and they likely make the tournament. In addition, if those 3 teams were in the top 50, it would've given Baylor 7 more top 50 wins, making their apparently weak schedule look a whole lot tougher.

Eight teams in the Big 12 took 4 losses to Texas and Baylor. In the ACC and SEC, most teams only play the other top teams once, so they avoid taking the extra loss. Since RPI only looks at record, and not the true quality of opponents, taking these extra losses is very harmful. I believe this is the reason that Sagarin and Massey each say that the Big 12 is underrated, because they dig deeper than just a superficial look at records. The SEC took a real beating this weekend, and every team that lost is rated higher in the RPI than Sagarin/Massey. I believe they are systematically overrated by the RPI. I don't know if that is the reason for the scheduling, but it seems to have worked out that way. This may not have been the Big 12's strongest year, but I don't think they got a fair shake. For reference, here is each major conference's non-conference winning percentage:

Big 12: .762
Pac 12: .719
ACC: .775
SEC: .754
Big 10: .713

Sorry for the long winded post, I spend way too much time thinking about these things.

The middle of the Big 12 got absolutely killed in the RPI this year, but it wasn't because of the 18-game schedule.

It was because 3 of the 4 teams in the second tier of the conference — Oklahoma State, West Virginia, and TCU — each played *horrible* nonconference schedules. Here are some of their illustrious opponents (RPI rank in parentheses):

• Louisiana-Monroe (338)
• Northwestern State *twice* (337)
• Incarnate Word (313)
• UT Rio Grande Valley (307)
• South Carolina Upstate (303)
• Coppin State (322)
• Central Connecticut (288)
• Morgan State (273)
• Sacramento State (272)
• Southwestern Louisiana (296)

You'll notice that Oklahoma's RPI didn't seem to be harmed by the 18-game conference schedule.

West Virginia seems to always schedule very light in the nonconference, and they count on making their hay of quality wins in the conference (which is exactly what they did last year). But this year TCU and Okie State did almost the same thing, and that caused the conference to lose the critical mass in collective strength of schedule that could any of those individual teams with a weak OOC SOS.

 

[IWearShoes]

If you've read many of my posts, you'll see that I have strong negative feelings about the RPI. Many assume this is because of my particular rooting interest, but it is really because it is a truly terrible way to rank teams. I hope this helps explain some of that.

I don't begrudge the committee's rankings, they did their job according to specified criteria. My problem is that the criteria themselves are wrong. The RPI favors teams with certain types of schedules, in particular the conferences with 16 games instead of 18. It is possible to "game" the system, and I believe the ACC and SEC have.

Let's imagine 2 conferences, identical in every way, except one has 10 teams, with an 18 game double round robin schedule, and the other 15 teams, playing 16 games and only repeating 2 opponents. Assume each team plays exactly 30 games. The non-conference record for each of the 5 major conferences was between 71% and 78%, so we will assume each conference wins 75% of their non-conference games so the math is easier.

Here is how the 10 team conference will fare:
10 teams*12 non-conference games each equals 120. They win 90 of them. The conference schedule is 90 games. The record for the entire conference is therefore 180-120, a .600 winning percentage.

Here is the 15 team conference:
15 teams*14 non-conference games equals 210 total non-conference games. They will win about 158. The conference schedule is 120 games. The record for the whole conference is 278-172, a .6177 winning percentage.

Therefore, the 16 game conferences have a nearly 2% advantage in RPI before the season starts. .0177 may seem like a small amount, but remember that 75% of the RPI formula is based on your opponent's record and their opponent's record. The overall record of the conference is therefore tremendously important, as they make up the majority of your schedule. Games played within the conference are zero-sum propositions, since every team plays both the winner and the loser. They cancel each other out from the perspective of the rest of the conference. This makes non-conference games the key to raising conference RPI. Major conference teams tend to win most of their non-conference games, so playing more of them is essential to raising the overall record, and therefore the RPI.

This effect is magnified in the Big 12, where each opponent is played twice, doubling the importance of each non-conference game for the other teams in the conference. The margins are extremely thin in the RPI, especially in the middle. The difference between #50 and #70, according to realtimerpi, is .0189. An extra non-conference win for all of their conference mates may have made a big difference for Oklahoma State, West Virginia and TCU, who were all in that range. Perhaps it would've bumped them into the top 50, off the bubble and perhaps into the field. Sagarin actually has West Virginia in the top 25, but their RPI was hurt because they took so many losses to the teams above them in the Big 12. Take those games out, and replace them with a decent, but winnable non-conference game, and they likely make the tournament. In addition, if those 3 teams were in the top 50, it would've given Baylor 7 more top 50 wins, making their apparently weak schedule look a whole lot tougher.

Eight teams in the Big 12 took 4 losses to Texas and Baylor. In the ACC and SEC, most teams only play the other top teams once, so they avoid taking the extra loss. Since RPI only looks at record, and not the true quality of opponents, taking these extra losses is very harmful. I believe this is the reason that Sagarin and Massey each say that the Big 12 is underrated, because they dig deeper than just a superficial look at records. The SEC took a real beating this weekend, and every team that lost is rated higher in the RPI than Sagarin/Massey. I believe they are systematically overrated by the RPI. I don't know if that is the reason for the scheduling, but it seems to have worked out that way. This may not have been the Big 12's strongest year, but I don't think they got a fair shake. For reference, here is each major conference's non-conference winning percentage:

Big 12: .762
Pac 12: .719
ACC: .775
SEC: .754
Big 10: .713

Sorry for the long winded post, I spend way too much time thinking about these things.

I'm going to be honest, I didn't read your entire post but have no doubt it was well researched and probably correct. I think MSU is clearly a Top 5 team and SC is likely a Top 12 team with a healthy A'ja, but after that the SEC was good but not great from TAMU to Bama and pretty weak below that.

MSU had a great season, but benefited from a somewhat weak at the top SEC and a schedule that laid out well for an undefeated run. Very few teams could have accomplished it, but it was a good setup with single games vs SC, UT, Mizzou, LSU and UGA (although I doubt UGA could beat us in 20 tries).

OK State was a tough out twice for us and for most teams and Baylor's 40 point win in Stillwater is extremely impressive to me. I thought OK State was more of a 5 or 6 seed than a 9 and Sagarin had them at 20 entering the MSU game so a 9 seed seems implausible.

 

[Bear11]

The middle of the Big 12 got absolutely killed in the RPI this year, but it wasn't because of the 18-game schedule.

It was because 3 of the 4 teams in the second tier of the conference — Oklahoma State, West Virginia, and TCU — each played *horrible* nonconference schedules. Here are some of their illustrious opponents (RPI rank in parentheses):

• Louisiana-Monroe (338)
• Northwestern State *twice* (337)
• Incarnate Word (313)
• UT Rio Grande Valley (307)
• South Carolina Upstate (303)
• Coppin State (322)
• Central Connecticut (288)
• Morgan State (273)
• Sacramento State (272)
• Southwestern Louisiana (296)

You'll notice that Oklahoma's RPI didn't seem to be harmed by the 18-game conference schedule.

West Virginia seems to always schedule very light in the nonconference, and they count on making their hay of quality wins in the conference (which is exactly what they did last year). But this year TCU and Okie State did almost the same thing, and that caused the conference to lose the critical mass in collective strength of schedule that could any of those individual teams with a weak OOC SOS.

Yes there were some weak games, but that isn't exactly my point. If all else is equal, the conference playing a 16 game schedule will always have a somewhat higher RPI than the one playing an 18 game schedule. Sagarin and Massey were not fooled by this, and had those 3 teams significantly higher rated. I think Oklahoma State proved themselves this weekend. It is true that according to the NCAA criteria they did not quite meet the bar, but perhaps those criteria should be changed. LSU, Mizzou, and Georgia didn't exactly play world beaters in their non-conference schedule, but they got in without debate. Playing an extra 2 non-conference games, and avoiding a rematch with Mississippi State and South Carolina probably helped their cause there in the eyes of the committee.

 

[Plebe]

Yes there were some weak games, but that isn't exactly my point. If all else is equal, the conference playing a 16 game schedule will always have a somewhat higher RPI than the one playing an 18 game schedule. Sagarin and Massey were not fooled by this, and had those 3 teams significantly higher rated. I think Oklahoma State proved themselves this weekend. It is true that according to the NCAA criteria they did not quite meet the bar, but perhaps those criteria should be changed. LSU, Mizzou, and Georgia didn't exactly play world beaters in their non-conference schedule, but they got in without debate. Playing an extra 2 non-conference games, and avoiding a rematch with Mississippi State and South Carolina probably helped their cause there in the eyes of the committee.

Your key phrase, "if all else is equal," glosses over a vast universe of other variables that come into play in team and conference RPI.

Are you saying that Oklahoma State was underseeded? It's easy to say that in retrospect. But then again, why did they lose to the likes of Kansas and Iowa State? They were probably in line for a 6 seed without those bad losses.

I don't think the committee is as easily fooled by the RPI as you think they are.

 

[Bear11]

Your key phrase, "if all else is equal," glosses over a vast universe of other variables that come into play in team and conference RPI.

Are you saying that Oklahoma State was underseeded? It's easy to say that in retrospect. But then again, why did they lose to the likes of Kansas and Iowa State? They were probably in line for a 6 seed without those bad losses.

I don't think the committee is as easily fooled by the RPI as you think they are.

Four SEC teams lost to lower seeded opponents, and every single one was ranked higher in RPI than Sagarin. I don't think this is a coincidence. The nature of the RPI favors the 16 game schedule. The committee clearly uses RPI heavily in its seedings, and the result was some rather predictable "upsets" for those who follow advanced metrics.

 

[willtalk]

You go into great detail to make a point. Still that isn't even necessary because the system is flawed at it's very foundations. I believe it was created to give more value to a teams won loss record. If a team plays a number of really low rated teams to begin with they have an up hill battle to get a decent RPI. They would have to play a considerably number of higher rated teams to balance out the low ones. Teams that have a much higher probability of giving them a loss. Thus they have nothing to gain and everything to lose.

Logically what difference is it to a team that is say top 10, 25, or 50 in respect to beating a team in the 200's or 300's. It logically should have no effect on their won loss record. They should all realistically be wins. Yet higher rated team that plays and beats a team rated in the 150 to 200's by a few points is rated far higher than a team that plays and blows out a team in the 250' and below. None of the teams played by either team has any chance to beat them so in reality those games have little impact on their won loss records. This negates the intent and purpose of the RPI. The better a team is the less impact teams lesser quality teams have on their won/loss records. The won loss record is only realistically effected by teams that have an actual shot at beating you. The better the team the higher the cut off for that actually is.

Some top teams use the early preseason as a combination of a practice and a boon to an ex assistant coach or local program. Now some teams with built in reps might use it to avoid playing out of conference teams to avoid an early loss which would bring down not only their rating but the conference as a whole. That whole problem could be solved by playing out most of the season and then making an adjustment by elimination any team a certain distance from them in the RPI evaluation. For example any team that had no legitimate shot at beating you is eliminated from the RPI calculations in respect to your team unless they beat you. What is left would constitute the new RPI used for calculating a teams actual strength of schedule. This would eliminate the drastic drop a team takes by playing a few really low rated cup cakes. Why should playing a lower rated cup cake hurt a teams RPI more than playing a higher rated cup cake?. All a team good team has to do to beat the RPI teams they should easily beat but not low enough to hurt them RPI wise. I say eliminate them all.

Teams can also raise their RPI by playing and getting blown out by a lot of really good teams. Play Uconn and another top team and your RPI goes way up despite the fact that you are blown out. This raises your RPI and also the RPI of every other team you play. Conversely a team effected by a lowered RPI based on playing some really low competition also effects the RPI of everyone they play as well. In what logical reality is it possible to raise your RPI from a blow out loss and lower it from a blow out win.

 

[KnightBridgeAZ]

All that many of you say is true, but I want to stress what one poster said - the selection committee is not solely focused on RPI, they are not "dumb". Often incorrect (or at least, some of the time) but not, I truly believe, being solely focused on rpi.

 

[UcMiami]

Interesting OP, but with a few flaws. Uconn for example has been incredible over the last decade and playing against Uconn for conferencemates has been a guaranteed loss most years, but has also been a guaranteed boost to their conferencemates RPI because of that 75% of RPI based on opponents and opponents records - Uconn's record is close to 1.00 (50% of the RPI for their opponent) each year and because they play a great OOC generally their opponents records are also very good as in .700-.800 (25% of the RPI for a Uconn opponent) in aggregate (OOC.) So the statement that not playing two games against similar teams (Louisville/ND) helps a conferencemates RPI is just not true - going 0-2 against a high win percentage team helps your RPI.

As others say - the RPI concept is just a bad idea, especially in a universe as vast and diverse as D1 women's basketball 349 teams that stretch from Evansville 1-27, 0-18 in the MV, to Uconn at 34-0. Common opponents or even common experience is non-existent. There are certainly ways to 'game' the RPI, but it has to do more with the scheduling of OOC games than the conference games. The SEC often benefits because their lower quality teams have scheduled really really bad teams and come into conference play with gaudy records - MsSt and Georgia and ... used to be 13-0 or 12-1 before conference play and then go 4-12 in conference, but everyone else in the conference benefited from that 90-100% win percentage OOC when they played them. The best OOC schedule for weak P5 teams for their conference opponents is to schedule the top 3 teams from the 4 weakest conferences - you go 100%, and your opponents end up going 85% in their conference seasons. The rest of the conference gets 50% of your record and 25% of your opponents record in their RPI.

 

[my2cents]

You can argue all you want about that stuff, but in the end you have to win 6 games in the Tournament to be declared National Champ.
There is no argument about who the Champion is.

 

[LoTrader]

All that many of you say is true, but I want to stress what one poster said - the selection committee is not solely focused on RPI, they are not "dumb". Often incorrect (or at least, some of the time) but not, I truly believe, being solely focused on rpi.

RPI puts up artificial guard rails for the selection committee. Also, RPI is what gets the "print" & is discussed ad nauseam for months. If all you read/hear is RPI, it has a way of becoming the truth.

 

[DEpup]

If you've read many of my posts, you'll see that I have strong negative feelings about the RPI. Many assume this is because of my particular rooting interest, but it is really because it is a truly terrible way to rank teams. I hope this helps explain some of that.

I don't begrudge the committee's rankings, they did their job according to specified criteria. My problem is that the criteria themselves are wrong. The RPI favors teams with certain types of schedules, in particular the conferences with 16 games instead of 18. It is possible to "game" the system, and I believe the ACC and SEC have.

Let's imagine 2 conferences, identical in every way, except one has 10 teams, with an 18 game double round robin schedule, and the other 15 teams, playing 16 games and only repeating 2 opponents. Assume each team plays exactly 30 games. The non-conference record for each of the 5 major conferences was between 71% and 78%, so we will assume each conference wins 75% of their non-conference games so the math is easier.

Here is how the 10 team conference will fare:
10 teams*12 non-conference games each equals 120. They win 90 of them. The conference schedule is 90 games. The record for the entire conference is therefore 180-120, a .600 winning percentage.

Here is the 15 team conference:
15 teams*14 non-conference games equals 210 total non-conference games. They will win about 158. The conference schedule is 120 games. The record for the whole conference is 278-172, a .6177 winning percentage.

Therefore, the 16 game conferences have a nearly 2% advantage in RPI before the season starts. .0177 may seem like a small amount, but remember that 75% of the RPI formula is based on your opponent's record and their opponent's record. The overall record of the conference is therefore tremendously important, as they make up the majority of your schedule. Games played within the conference are zero-sum propositions, since every team plays both the winner and the loser. They cancel each other out from the perspective of the rest of the conference. This makes non-conference games the key to raising conference RPI. Major conference teams tend to win most of their non-conference games, so playing more of them is essential to raising the overall record, and therefore the RPI.

This effect is magnified in the Big 12, where each opponent is played twice, doubling the importance of each non-conference game for the other teams in the conference. The margins are extremely thin in the RPI, especially in the middle. The difference between #50 and #70, according to realtimerpi, is .0189. An extra non-conference win for all of their conference mates may have made a big difference for Oklahoma State, West Virginia and TCU, who were all in that range. Perhaps it would've bumped them into the top 50, off the bubble and perhaps into the field. Sagarin actually has West Virginia in the top 25, but their RPI was hurt because they took so many losses to the teams above them in the Big 12. Take those games out, and replace them with a decent, but winnable non-conference game, and they likely make the tournament. In addition, if those 3 teams were in the top 50, it would've given Baylor 7 more top 50 wins, making their apparently weak schedule look a whole lot tougher.

Eight teams in the Big 12 took 4 losses to Texas and Baylor. In the ACC and SEC, most teams only play the other top teams once, so they avoid taking the extra loss. Since RPI only looks at record, and not the true quality of opponents, taking these extra losses is very harmful. I believe this is the reason that Sagarin and Massey each say that the Big 12 is underrated, because they dig deeper than just a superficial look at records. The SEC took a real beating this weekend, and every team that lost is rated higher in the RPI than Sagarin/Massey. I believe they are systematically overrated by the RPI. I don't know if that is the reason for the scheduling, but it seems to have worked out that way. This may not have been the Big 12's strongest year, but I don't think they got a fair shake. For reference, here is each major conference's non-conference winning percentage:

Big 12: .762
Pac 12: .719
ACC: .775
SEC: .754
Big 10: .713

Sorry for the long winded post, I spend way too much time thinking about these things.

Sorry but cry me a river! UConn has to deal with the AAC and still manages to get at the top of the RPI. Baylor just needs to drop the creme puffs early in the season. It's a simple solution.

 

[Bear11]

Sorry but cry me a river! UConn has to deal with the AAC and still manages to get at the top of the RPI. Baylor just needs to drop the creme puffs early in the season. It's a simple solution.

The point of the post is that the Big 12 is artificially suppressed in the RPI because of the way it is calculated. Massey and Sagarin agree. Baylor DOES have a good schedule, they just aren't given credit due to the committee's poor evaluation criteria. If the Big 12 cut back to a 16 game schedule, the record for the entire conference, and hence the RPI, would increase. I don't want this to happen, because the big 12 schedule is excellent. It just needs to be recognized as such.

This is not about my team affiliation. This is about the NCAA using a criteria which favors certain conferences for playing fewer conference games. The scenario drawn up in the OP demonstrates this.

 

[willtalk]

Sorry but cry me a river! UConn has to deal with the AAC and still manages to get at the top of the RPI. Baylor just needs to drop the creme puffs early in the season. It's a simple solution.

I think you make a valid point in respect to evaluating Baylors record when it comes to rating their record ultimately. But that should be done by people and not stats. Who could not see that they didn't play a tough out of conference schedule just by looking which should be considered when compared to teams that played more teams that were capable of beating them. The eyeball test on their schedule and common sense would be a far better measure of their record than using a flawed stat driven metric. Now they can say that they use other factors besides RPI but by looking at the final Bracketing it seems obvious that they use it as a major factor. That is just pure lazyness from people who do not have the knowledge or the desire to do enough research to get enough knowledge to trust their own judgement.

I don't think they use the RPI as much with the higher rated teams. It is with mid level and the bubble teams that it has it's greatest effect. For example Oklahoma getting in with a 16-14 record. That had to be if not RPI related if not justified in order to mitigate that record.