|
You're right in basically everything you said.
However, two points:
1) The main problem people have is that (to some degree now and definitely in the forseeable future) there's a severe ceiling effect for the #1 slots. If the game reaches say a million players in any given region (which is an extremely conservative estimate), and if we are to assume diamond is the top 5% then you have 50,000 people in diamond. If we're to assume 1%, then we will have 10,000 people in diamond. Now divide by a 100 and you have one or five hundred people wondering where they place relative to those other 100-500. For most everyone else, it's basically the same as a percentile, but it would be nice to know how many diamond leagues there are.
2) Without a lurking variable (maybe early playtime/qualifying time? Was that the first diamond league?) http://www.teamliquid.net/forum/viewmessage.php?topic_id=135691 the chances of those players all ending up in that bracket seem extremely low to me. It's possible they did some sort of "tier" within diamond. I'd also like to point out that the top 1 person in my league point wise would rank in at #9 in that league and that our #2 person wouldn't even make that top 12 list by a longshot (325 points)
|
Although I am not top-tier, personally I would like to know my specific ranking in all the players that play... I would rather be 10236 than 10237 / rank 5 in one league of 100s or something.. Anyway.. right now I am around rank 4 platinum and the ranking seems pretty accurate overall to me.. Although I was in diamond in phase 1 and I am way better now than I was back then so its confusing why I am not at least like rank 80 or something in diamond..
|
On July 13 2010 05:49 Ideas wrote:
its not as good as 1 big ladder. from what I understand, you're saying that over time someone's rank will correspond to their spot in the overall playing population such as rank 15 means you're in the 15th best percentile of the population. but if it was 1 big ladder that would mean that you can see what you rank is there and then divide to get a percentage right there. and you're not guestimating, it's your real rank.
The point of the system is to be easy to decipher at a glance, without doing math.
On July 13 2010 05:54 Roggay wrote:
I admit I haven't read it all, but I fail to understand why the division system need so much math and talk. In the end all the players care about is if they are pleased with being only 100 in their little ladder ranking.
In my opinion it isn't the division system that is the problem but the fact that there is only one ranking. The division are really good for low and average players because being ranked 10404th doesnt mean anything. But for the good players it IS important to have an overall ranking. Having two ranking, the actual one in the division and say, another like warcraft III's one, where you can check your overall ranking (maybe by league or by rating), would be ideal.
It needs math because it is math. The division system is based on a sample size of 100 of your league (which is a perfectly reasonable size for a sample) which allows a player to easily figure out how he's doing in his league in terms of percentile. I think a website will probably be set up with global/region-based xxxxx/yyyyyyy rankings but such a system is not what blizzard will implement in-game, I'd bet on it.
On July 13 2010 05:55 Andtwo wrote:You're right in basically everything you said. However, two points: 1) The main problem people have is that (to some degree now and definitely in the forseeable future) there's a severe ceiling effect for the #1 slots. If the game reaches say a million players in any given region (which is an extremely conservative estimate), and if we are to assume diamond is the top 5% then you have 50,000 people in diamond. If we're to assume 1%, then we will have 10,000 people in diamond. Now divide by a 100 and you have one or five hundred people wondering where they place relative to those other 100-500. For most everyone else, it's basically the same as a percentile, but it would be nice to know how many diamond leagues there are. 2) Without a lurking variable (maybe early playtime/qualifying time? Was that the first diamond league?) http://www.teamliquid.net/forum/viewmessage.php?topic_id=135691 the chances of those players all ending up in that bracket seem extremely low to me. It's possible they did some sort of "tier" within diamond. I'd also like to point out that the top 1 person in my league point wise would rank in at #9 in that league and that our #2 person wouldn't even make that top 12 list by a longshot (325 points)
Yeah this is an issue. I don't think blizzard would purposefully group up all the best players in one division (that would completely defeat the purpose of divisions in that it makes it impossible to determine percentile based rankings with such a pool of talent in one division) but I hope that they'll implement a system to guarantee something like that doesn't happen. To address the first point, that is an issue, yes, but for a vast, vast majority of players the system works very well. I hope they do make it so that the top 5 in diamond don't have to "guess" at their rank, but we'll have to see if that happens in the future.
|
On July 13 2010 05:41 Issorlol wrote:Show nested quote +On July 13 2010 05:40 FabledIntegral wrote:
I don't understand how "over time," means much if you don't move around in your division. If I'm placed in upsilon sigma tau or whatever, I'm stuck with those 100 people. Because 100 is a relatively low number relative to the population, it is likely many people are going to suffer from a variety of skill within their own division. Maybe if division size was ~1,500-2,000 you'd have a more valid point. But as it stands, players could get placed in a shitty division and make it to rank 1 relatively easily, while they might only be rank ~25 elsewhere in another division. That rank 1 person could not even compare to the other rank 1 people elsewhere in the beta. And the same situation can happen in reverse. The more divisions you have, the more this discrepancy will occur. Without any division system you have no discrepancy - you are where you should be.
At an extreme, imagine each division had only 2 players. If you were constantly ranked 2 in that division, would that be enough data to reliably say you're in the bottom 50%? Most likely not. Although 100 would seem like it's a large enough originally to separate people, given the total number of people in your league (especially at the silver/gold/plat levels), more and more outliers are going to happen. Yes, while the "general" population we could say would be roughly where they should be (still, 100 is NOT a very large indicator in this case), the amount of outliers would be enough that it wouldn't be an amazing indicator.
So I have no idea why you suggest that this division system is better than a straight ladder system - which I would very much prefer. I have no problem seeing I'm ranked 13,500 if I know that's still within the top 10% because it gives a total 13,500/150,000 for the Diamond league, etc. That gives you a much more defined rank on where you stand. 100 is a perfectly acceptable number to use as a sample size of a large population. you would find larger divisions only marginally more accurate.
If it's a really large population, then there will be even more outliers (yes I'm aware there may be the same percentage of outliers and thus you'd encounter them around the same frequency) but to have so many outliers has little to no advantage over a global system which lets you know exactly where you're at.
When the game is released we can expect the userbase to absolutely EXPLODE. I just feel 100 users is far too small of a division size if you want to accurately say you're around that percent. "over time" won't fix that your division skill could be different than anothers, it only fixes where you are within your own division.
|
Rank means nothing. Rating is everything. My current rating is like 140 Diamond. That puts me at rank 40 or so in my division. In my friend's division, that exact same rating puts me at rank 80. At various points in the beta, I could have been anywhere from lower 50s to top 10 with the exact same rating. What that essentially tells me is that ranking is worthless while rating is everything. Saying that you are #7 in Dropship Uncle means nothing because there is no reference point. You could have a rating of 400 to be #7, while in Omega Infestation, that rating puts you in the low 30s.
In fact, I'd argue that win: loss: rating ratio matters far more. Being rated 400 with a win:loss ratio of 2:1 is far more impressive than a 400 with a .9:1 ratio. I'd put rank only slightly above what league you placed into as a superficial determinant of skill.
|
All I know is that I'm tired of being 500+ rated in gold and going 70%w/l against non-stop 300+ diamond players. I would prefer if it was just like iccup or had some kind of global ranking. The problem with the ratings is that unless you are in diamond you have no point at which to base what your rating means.
|
100 is a perfectly acceptable number to use as a sample size of a large population. you would find larger divisions only marginally more accurate.
Acceptable doesn't mean optimal. You make some good points, but to be honest I don't understand why you are fighting so hard for this system. After the change from division numbers into names it became quite obvious that blizzard is intentionally trying to create ambiguity not for the average casual gamer, but to for the hardcore community who were putting a fair amount of effort into creating an accurate way to rank players among regions. Why defend something that goes out of its way to remain unreliable. Sure it might work out to be an ok reference, but it will never be an optimal reference.
|
On July 13 2010 05:21 Issorlol wrote:Show nested quote +On July 13 2010 05:10 Mindcrime wrote:
I don't see how rating isn't a better indicator of skill. More math please.
Did you actually read the post?
I certainly did. In order for this "#1 diamond = top 1% of diamond players" reasoning to work, there needs to be not only similarity between divisions, but near-parity. Everyone has to deal with the same rating system and matchmaking system, but some people will have the luxury of being compared against a division in which the majority of players are below the mean skill level in a league.
Sure, I'll accept that ranking may become a better indicator of skill than it is now, but rating will always be better.
|
On July 13 2010 05:59 FabledIntegral wrote:Show nested quote +On July 13 2010 05:41 Issorlol wrote:On July 13 2010 05:40 FabledIntegral wrote:
I don't understand how "over time," means much if you don't move around in your division. If I'm placed in upsilon sigma tau or whatever, I'm stuck with those 100 people. Because 100 is a relatively low number relative to the population, it is likely many people are going to suffer from a variety of skill within their own division. Maybe if division size was ~1,500-2,000 you'd have a more valid point. But as it stands, players could get placed in a shitty division and make it to rank 1 relatively easily, while they might only be rank ~25 elsewhere in another division. That rank 1 person could not even compare to the other rank 1 people elsewhere in the beta. And the same situation can happen in reverse. The more divisions you have, the more this discrepancy will occur. Without any division system you have no discrepancy - you are where you should be.
At an extreme, imagine each division had only 2 players. If you were constantly ranked 2 in that division, would that be enough data to reliably say you're in the bottom 50%? Most likely not. Although 100 would seem like it's a large enough originally to separate people, given the total number of people in your league (especially at the silver/gold/plat levels), more and more outliers are going to happen. Yes, while the "general" population we could say would be roughly where they should be (still, 100 is NOT a very large indicator in this case), the amount of outliers would be enough that it wouldn't be an amazing indicator.
So I have no idea why you suggest that this division system is better than a straight ladder system - which I would very much prefer. I have no problem seeing I'm ranked 13,500 if I know that's still within the top 10% because it gives a total 13,500/150,000 for the Diamond league, etc. That gives you a much more defined rank on where you stand. 100 is a perfectly acceptable number to use as a sample size of a large population. you would find larger divisions only marginally more accurate. If it's a really large population, then there will be even more outliers (yes I'm aware there may be the same percentage of outliers and thus you'd encounter them around the same frequency) but to have so many outliers has little to no advantage over a global system which lets you know exactly where you're at. When the game is released we can expect the userbase to absolutely EXPLODE. I just feel 100 users is far too small of a division size if you want to accurately say you're around that percent. "over time" won't fix that your division skill could be different than anothers, it only fixes where you are within your own division.
I understand what you're saying, and you are correct, but the difference in accuracy with the division system if the divisions were ten times larger than they are now is really not all that large. If the divisions were ten people, you would have a very valid point and my post wouldn't be defending the division system at all, but with 100 people you have a nice number (your rank) to determine your percentile (again, your rank). It has flaws, yeah, but they're not as large as you think they are.
On July 13 2010 06:00 Wr3k wrote:
All I know is that I'm tired of being 500+ rated in gold and going 70%w/l against non-stop 300+ diamond players. I would prefer if it was just like iccup or had some kind of global ranking. The problem with the ratings is that unless you are in diamond you have no point at which to base what your rating means.
If you never faced people out of your league, you would never be promoted.
On July 13 2010 06:00 0mar wrote:
Rank means nothing. Rating is everything. My current rating is like 140 Diamond. That puts me at rank 40 or so in my division. In my friend's division, that exact same rating puts me at rank 80. At various points in the beta, I could have been anywhere from lower 50s to top 10 with the exact same rating. What that essentially tells me is that ranking is worthless while rating is everything. Saying that you are #7 in Dropship Uncle means nothing because there is no reference point. You could have a rating of 400 to be #7, while in Omega Infestation, that rating puts you in the low 30s.
In fact, I'd argue that win: loss: rating ratio matters far more. Being rated 400 with a win:loss ratio of 2:1 is far more impressive than a 400 with a .9:1 ratio. I'd put rank only slightly above what league you placed into as a superficial determinant of skill.
Did you read my post? "Rank means nothing" is an absolutely incorrect statement. Rank means this: Your percentile-based rank in your league given a volume of games and a period of time lengthy enough for you to stop shuffling around leagues. This is not nothing. The ONLY TIME the division system is inaccurate in this regard under those conditions is in the very top of the diamond league.
On July 13 2010 06:01 Toids wrote:Show nested quote +
100 is a perfectly acceptable number to use as a sample size of a large population. you would find larger divisions only marginally more accurate.
Acceptable doesn't mean optimal. You make some good points, but to be honest I don't understand why you are fighting so hard for this system. After the change from division numbers into names it became quite obvious that blizzard is intentionally trying to create ambiguity not for the average casual gamer, but to for the hardcore community who were putting a fair amount of effort into creating an accurate way to rank players among regions. Why defend something that goes out of its way to remain unreliable. Sure it might work out to be an ok reference, but it will never be an optimal reference.
It's not that I'm "fighting" for the system as much as I'm trying to point out that it's not the worst thing in the world. Blizzard's intent behind the division name change, by the way, was the make it perfectly clear that divisions were NOT ranked based on their numbers.
On July 13 2010 06:01 Mindcrime wrote:Show nested quote +On July 13 2010 05:21 Issorlol wrote:On July 13 2010 05:10 Mindcrime wrote:
I don't see how rating isn't a better indicator of skill. More math please.
Did you actually read the post? I certainly did. In order for this "#1 diamond = top 1% of diamond players" reasoning to work, there needs to be not only similarity between divisions, but near- parity. Everyone has to deal with the same rating system and matchmaking system, but some people will have the luxury of being compared against a division in which the majority of players are below the mean skill level in a league. Sure, I'll accept that ranking may become a better indicator of skill than it is now, but rating will always be better.
I completely agree.
|
On July 13 2010 05:59 FabledIntegral wrote:Show nested quote +On July 13 2010 05:41 Issorlol wrote:On July 13 2010 05:40 FabledIntegral wrote:
I don't understand how "over time," means much if you don't move around in your division. If I'm placed in upsilon sigma tau or whatever, I'm stuck with those 100 people. Because 100 is a relatively low number relative to the population, it is likely many people are going to suffer from a variety of skill within their own division. Maybe if division size was ~1,500-2,000 you'd have a more valid point. But as it stands, players could get placed in a shitty division and make it to rank 1 relatively easily, while they might only be rank ~25 elsewhere in another division. That rank 1 person could not even compare to the other rank 1 people elsewhere in the beta. And the same situation can happen in reverse. The more divisions you have, the more this discrepancy will occur. Without any division system you have no discrepancy - you are where you should be.
At an extreme, imagine each division had only 2 players. If you were constantly ranked 2 in that division, would that be enough data to reliably say you're in the bottom 50%? Most likely not. Although 100 would seem like it's a large enough originally to separate people, given the total number of people in your league (especially at the silver/gold/plat levels), more and more outliers are going to happen. Yes, while the "general" population we could say would be roughly where they should be (still, 100 is NOT a very large indicator in this case), the amount of outliers would be enough that it wouldn't be an amazing indicator.
So I have no idea why you suggest that this division system is better than a straight ladder system - which I would very much prefer. I have no problem seeing I'm ranked 13,500 if I know that's still within the top 10% because it gives a total 13,500/150,000 for the Diamond league, etc. That gives you a much more defined rank on where you stand. 100 is a perfectly acceptable number to use as a sample size of a large population. you would find larger divisions only marginally more accurate. If it's a really large population, then there will be even more outliers (yes I'm aware there may be the same percentage of outliers and thus you'd encounter them around the same frequency) but to have so many outliers has little to no advantage over a global system which lets you know exactly where you're at. When the game is released we can expect the userbase to absolutely EXPLODE. I just feel 100 users is far too small of a division size if you want to accurately say you're around that percent. "over time" won't fix that your division skill could be different than anothers, it only fixes where you are within your own division.
100 sample size is actually pretty gigantic statistically speaking. What you'd be concerned about is more how people place because then the distribution is not really random. Once the population explosion has happened, people will quit and the original leagues will have the more practiced players which is going to skew things. I really don't think it matters except for diamond league though.
|
Technically, a single ladder where everyone is in the same boat and just using a straight elo rating is the best indicator of where you lie in terms of skill. Just calling out your phrasing really.
I like the league/division system, although I'd agree that a universal rating viewer would be a good asset to have, so players who care could see where they rate. Maybe if they implemented it for diamond only, as plat and below it wouldn't really matter anyway, as they're bound to be past the first 1,000 ranks
|
On July 13 2010 05:35 Issorlol wrote:Show nested quote +On July 13 2010 05:34 rastaban wrote:
What happens if for some reason say Idra and TLO are both placed in the same platinum division. They both should be rank 1, but obviously 1 will be higher than the other. so one of them now looks like he is in only the top 2%.
If I understand the OP right then in this case we should look at their points, wouldn't it be better if instead it moved TLO to a separate division so both are now listed as rank 1??
Yes, moving them to different divisions would be ideal, but in the current system rating is then the only reference we have to compare skill. Again, it's not perfect but it is certainly not the WORST IDEA EVER OMG as some seem to think.
Which is why your rating/points is still more important than your rank. You can be tenth in your diamond league and still be better/ a higher percentile than someone who's ranked number 1 in a different diamond league. You may just be in a stacked league, with more players who are better.
And PLEASE don't say "Oh it's simple math." It's not simple math. I know you're not saying it's perfect, but claiming that rank = percentile (even "in general") is completely wrong, which is why we keep asking for rating over rank on TeamLiquid when we're evaluating how good people are.
~An actual mathematician
(Most of your post was good though, so it was worth the read.)
|
On July 13 2010 06:09 Stropheum wrote:
Technically, a single ladder where everyone is in the same boat and just using a straight elo rating is the best indicator of where you lie in terms of skill. Just calling out your phrasing really.
I like the league/division system, although I'd agree that a universal rating viewer would be a good asset to have, so players who care could see where they rate. Maybe if they implemented it for diamond only, as plat and below it wouldn't really matter anyway, as they're bound to be past the first 1,000 ranks
Yeah my phrasing at points is pretty bad. Thanks for pointing out my flaws .______.
But yeah, thanks for taking the time to read the post and for the input :3
On July 13 2010 06:10 DarkPlasmaBall wrote:
And PLEASE don't say "Oh it's simple math." It's not simple math. I know you're not saying it's perfect, but claiming that rank = percentile ("in general") is completely wrong, which is why we keep asking for rating over rank on TeamLiquid.
~An actual mathematician
(Most of your post was good though, so it was worth the read.)
Not trying to be a smart-ass here, genuinely - can you explain this? The depth of my stastics knowledge is... AP stats. -_-
~An incoming freshman math major
|
From my observations on TL what the division system has done very well is convince a large number of pretty useless people that they are actually good at the game.
This is a great thing. People feeling good about themselves (I'm in DiamondYeah!) -> people playing a lot -> longer lasting and more popular game.
Unfortunately I played BW so I already KNOW that I suck. Reality sucks :p
|
On July 13 2010 06:13 Klive5ive wrote:
From my observations on TL what the division system has done very well is convince a large number of pretty useless people that they are actually good at the game.
This is a great thing. People feeling good about themselves (I'm in DiamondYeah!) -> people playing a lot -> longer lasting and more popular game.
Unfortunately I played BW so I already KNOW that I suck. Reality sucks :p
Sadly I kinda agree with that.
|
My friend is good enough to be #1 for the entire US server, how does being #1 in diamond even reflect his skill? He like 4-1'd whitera when EU servers were down
being in top 1% of diamond is not an achievement
|
Very simple here. There will be a pro-league that does not use division a month or two after launch as Blizzard has stated before. Pros/high-level players need not worry soon enough you will be able to see what you want and that is an accurate ranking amongst all your peers
|
On July 13 2010 06:11 Issorlol wrote:Show nested quote +On July 13 2010 06:10 DarkPlasmaBall wrote:
And PLEASE don't say "Oh it's simple math." It's not simple math. I know you're not saying it's perfect, but claiming that rank = percentile ("in general") is completely wrong, which is why we keep asking for rating over rank on TeamLiquid.
~An actual mathematician
(Most of your post was good though, so it was worth the read.) Not trying to be a smart-ass here, genuinely - can you explain this? The depth of my stastics knowledge is... AP stats. -_- ~An incoming freshman math major
Sure, no problem.
Since some divisions may be stacked (which was talked about), it's incorrect to assume that every division's third place is truly in the top 3% of their respective leagues. Therefore, you can't apply percentile rank unless every single player of a specific league (like diamond league) was on the same single ladder. Since there are so many different ladders with different players (and therefore different ability levels), you can't properly apply one sweeping percentile rank to apply to all of them.
When you apply percentile rank to something like a standardized test, it's because every single student in the country took the EXACT same test (not different "divisions" of "similar" tests). All the same questions (like all the same players), not different kinds of questions or groups of players.
|
On July 13 2010 06:16 iopq wrote:
My friend is good enough to be #1 for the entire US server, how does being #1 in diamond even reflect his skill? He like 4-1'd whitera when EU servers were down
being in top 1% of diamond is not an achievement
I take it you read some of the first post, none of the responses, and then posted this?
On July 13 2010 06:17 DarkPlasmaBall wrote:Show nested quote +On July 13 2010 06:11 Issorlol wrote:On July 13 2010 06:10 DarkPlasmaBall wrote:
And PLEASE don't say "Oh it's simple math." It's not simple math. I know you're not saying it's perfect, but claiming that rank = percentile ("in general") is completely wrong, which is why we keep asking for rating over rank on TeamLiquid.
~An actual mathematician
(Most of your post was good though, so it was worth the read.) Not trying to be a smart-ass here, genuinely - can you explain this? The depth of my stastics knowledge is... AP stats. -_- ~An incoming freshman math major Sure, no problem. Since some divisions may be stacked (which was talked about), it's incorrect to assume that every division's third place is truly in the top 3% of their respective leagues. Therefore, you can't apply percentile rank unless every single player of a specific league (like diamond league) was on the same single ladder. Since there are so many different ladders with different players (and therefore different ability levels), you can't properly apply one sweeping percentile rank to apply to all of them. When you apply percentile rank to something like a standardized test, it's because every single student in the country took the EXACT same test (not different "divisions" of "similar" tests).
Oh okay, yeah, thanks. I can see now I definitely could have talked about the percentiles and ranks in such in a much clearer (more correct :x) way. I think this will be less of an issue, though, if blizzard ensures that divisions aren't stacked in the future. I don't see why they won't do this, since otherwise there is literally no reason to use divisions and I doubt blizzard would invest time into creating divisions just to nullify them because of weird placement. It won't be exact, but it will be pretty close. I don't have much of an issue with that.
|
On July 13 2010 06:08 Andtwo wrote:Show nested quote +On July 13 2010 05:59 FabledIntegral wrote:On July 13 2010 05:41 Issorlol wrote:On July 13 2010 05:40 FabledIntegral wrote:
I don't understand how "over time," means much if you don't move around in your division. If I'm placed in upsilon sigma tau or whatever, I'm stuck with those 100 people. Because 100 is a relatively low number relative to the population, it is likely many people are going to suffer from a variety of skill within their own division. Maybe if division size was ~1,500-2,000 you'd have a more valid point. But as it stands, players could get placed in a shitty division and make it to rank 1 relatively easily, while they might only be rank ~25 elsewhere in another division. That rank 1 person could not even compare to the other rank 1 people elsewhere in the beta. And the same situation can happen in reverse. The more divisions you have, the more this discrepancy will occur. Without any division system you have no discrepancy - you are where you should be.
At an extreme, imagine each division had only 2 players. If you were constantly ranked 2 in that division, would that be enough data to reliably say you're in the bottom 50%? Most likely not. Although 100 would seem like it's a large enough originally to separate people, given the total number of people in your league (especially at the silver/gold/plat levels), more and more outliers are going to happen. Yes, while the "general" population we could say would be roughly where they should be (still, 100 is NOT a very large indicator in this case), the amount of outliers would be enough that it wouldn't be an amazing indicator.
So I have no idea why you suggest that this division system is better than a straight ladder system - which I would very much prefer. I have no problem seeing I'm ranked 13,500 if I know that's still within the top 10% because it gives a total 13,500/150,000 for the Diamond league, etc. That gives you a much more defined rank on where you stand. 100 is a perfectly acceptable number to use as a sample size of a large population. you would find larger divisions only marginally more accurate. If it's a really large population, then there will be even more outliers (yes I'm aware there may be the same percentage of outliers and thus you'd encounter them around the same frequency) but to have so many outliers has little to no advantage over a global system which lets you know exactly where you're at. When the game is released we can expect the userbase to absolutely EXPLODE. I just feel 100 users is far too small of a division size if you want to accurately say you're around that percent. "over time" won't fix that your division skill could be different than anothers, it only fixes where you are within your own division. 100 sample size is actually pretty gigantic statistically speaking. What you'd be concerned about is more how people place because then the distribution is not really random. Once the population explosion has happened, people will quit and the original leagues will have the more practiced players which is going to skew things. I really don't think it matters except for diamond league though.
100 is not gigantic whatsoever statistically speaking... almost no survey will use a sample as small as 100 ppl if you're trying to compare to the rest of the population.
|
|
|
|
|
|
EPT
