Talk:Don Bradman

Template:V0.5

B?

You can't be serious. What a kick in the guts for all those that have sweat blood and tears on this article. Shame on you. —Preceding unsigned comment added by Dyslexicbudgie (talk • contribs) 15:49, 13 October 2007 (UTC)

Bad statistics by Charles Davis

I stumbled upon Charles Davis' statistics in this article, and I am amazed. Davis must be a bad statistician. I am not saying that Bradman was a bad sportsplayer. I don't want to discuss if he was the best sportsplayer ever. I only want to discuss the numbers. Talking about Pele, it says that once in every 9300 players, a person of Pele's calibre will show up. This is obviously ridiculous. There are one million soccer players in the Netherlands alone. This would mean that, on everage, about 100 Dutch soccer players are of Pele's calibre in goalscoring. This is definitely not the case. Charles Davis obviously made a big mistake: assuming normal distribution for goal average/batting average/whatever.

Some other statistics: Eddy Merckx entered 1582 cycling races as a professional cyclist, of which he won 445. If we assume that there are only 10 cyclists in every race (there are many more), and that every cyclist in the race has equal probability of winning, then the binomial distribution tells us that such an achievement (or better) has a probability of only 0.0000(130 zeroes)002 %. This would imply that Eddy Merckx was a much better sportsman than Donald Bradman.

This is also bad statistics, because I compare things that should not be compared in this way. I looked up information about Charles Davis, and it is clear that he is a big cricket-fan. Nothing wrong with being a cricket-fan in general, but it is bad to let a cricket-fan decide which method will be used to determine the best sportsman of all time, and not frown upon the result that a cricket-player comes out best.

I understand that Davis wrote a book about it. I haven't. But I feel Davis should stick to cricket-statistics, and wikipedia should not use his other-sports-statistics, because he clearly (in my opinion) fails here. I hope that I made that clear. Although I understand statistics, I am not a statistician myself. There must be some real statistician that saw this strange result, and said something about it. Unfortunately, I don't know who did this and where. What is above here is only the opinion of one wikipedian (although according to me it is a fact ;)). I could just remove the flawed statistics, but I have the feeling that it would reappear soon. If somebody can find the statistician that explained that these statistics are wrong, put the info here. Then the bad statistics can disappear. --213.46.8.134 (talk) 10:54, 5 December 2007 (UTC)

According to Davis: A bradmanesque achievement isTennis: 15-20 Grand Slam titles in 10 years. Margaret Smith Court won 20 single Grand Slam titles from 1960 to 1970. Steffi Graf won 21 single Grand Slam titles between 1987 and 1996.

And statistically unbeatable: the ice skater Eric Heiden. In a professional career of about 4 years, he won 4 times the world sprint skating championship, 3 times allround (and 1 time second place), and 5 olympic golden medals (all medals available for ice skating). He won every race he could win, only coming second once in the world allround championship. Clearly these three sportspersons show better statistics than Bradman. I don't claim they were better sportspersons, I only claim their statistics are better. Margaret Smith Court and Steffi Graf show better statistics even according to Davis' own rules :S --213.46.8.134 (talk) 11:10, 5 December 2007 (UTC)

This is fascinating (I even understood some of the maths) but without RS your opinions, no matter how reasonably argued, are OR. --Dweller (talk) 12:11, 5 December 2007 (UTC)

I totally agree, that's why I did not change the article. There are two things I want to do: 1. Ask other wikipedians, when they see RS that disagrees with Davis, to put it here. 2. Make clear that Davis is not a reliable source here. Davis' list of best sportspeople fits well on a fan site, but not in an encyclopedia. I explicitly want to make clear that I do NOT want to include MSC, Steffi Graf nor Eric Heiden on this list; I say we should remove the list altogether, because the source is not reliable in this context.--213.46.8.134 (talk) 14:11, 5 December 2007 (UTC)

Looks to me that a lot of people don't understand statistics and particularly standard deviations. Let me clear up a simpler misunderstanding. "Once in every 9300 players, a person of Pele's calibre will show up" is correct. This does not of course mean 9300 casual players, it means 9300 international footballers who are a rather rarer beast. Davis claims Bradman comes out ahead in his sport and in the other sports he looked at in terms of standard deviations. The performance of cricket batsmen follows a more or less perfect bell curve - a normal distribution with a low variance - with very few over 50 and just 4 at an average of 60 plus, 1 at an average above 61, Bradman, out of the thousands that have batted in tests. These 60 plus guys are already a couple of standard deviations away from the norm. It is beacause of the lack of variance over so many players that Bradmans singular average is so many SD's from the mean. Compare this to the number of tennis players bandied about as Bradman beaters. Well the fact we can name Court, King, Graff, Federer, Sampras and Borg as exceptional players suggests that the variance in tennis is much higher than that in cricket, the very fact that there are so many extremely successful tennis players means that they are not as far in terms of SD's from the mean as each one adds to variance. The ice skating guy doesnt even come close, a 2 second scan of the history of champs in his events show guys that have won more often than him! As for the supposed maths about cycling that simply proves that the sport measures a skill and that the cyclist is skilled in that skill, rather than simply random results. Or to put it simply, it just says taht cycle races have an element of skill, they are not simply random events. Try to bear in mind that Davis is a professional statistician and as such if your thinking disagrees with his, most likely you are wrong. --LiamE (talk) 15:45, 5 December 2007 (UTC)

I disagree on the "ice skating guy"; some guys have won more, but in a longer career. Davis is a professional statistician, but it looks like his love for cricket disturbed his statistics. Cricket is a good sports for statistics. Other sports are not. If you want to compare tennis players with cricket players, what do you compare? Points won? Games won? Sets won? Matches won? Tournaments won? Any statistician can tell you that the result depends on the rule you choose. Davis choose a set of rules. A set of rules that might look ok on first sight, but should be examined more carefully. But Davis is a statistician, so I will assume that he did this well. But there is the simple fact that according to Davis' own rules, Court and Graf are better than Bradman. Now you can choose to make a new set of rules (not looking at standard deviations anymore, but looking at how many exceptional players we can name for a sport), but then you make the set of rules after the 'experiment', and that is pseudo-science. Bradman was good at cricket. Graf was good at tennis. Who was better? Uncomparable. --Pie.er (talk) 08:52, 6 December 2007 (UTC)

I can assure you that they are comparable in terms of standard deviations from the mean in their own sports. If you don't understand that you simply have no place even reading what a statistician has to say let alone question it. --LiamE (talk) 10:26, 6 December 2007 (UTC)

Comparable, of course. No one doubted that. Of course there is very large correlation between persons that are "several standard deviations better" in their sport, and persons that are the only exceptional player in their sport. But, as you of course understand, they are not the same. Davis chose to use the standard deviation criterion, (which can be argued to be a good choice), and in that criterion Graf is better than Bradman. And in real science, you shouldn't change the method only because you don't like the result... --Pie.er (talk) 14:00, 6 December 2007 (UTC)

It is better to remain silent and appear stupid than to open your mouth and remove all doubt. Throwing in terms like real science still won't deflect from the fact that you know jack about statistics. Where are your results for Graf? You are talking rubbish, you have neither calculated the variance not the SD of tennis grand slam title winners nor shown that Graff is more than 4.4 standard deviations from mean for a pro tennis player so how can you say she is better than Bradman "in that criterion"? The standard deviation is not a criterion, it is the result used for comparison. Davis did not change criteria to suit, he used the best available for each sport he looked at. The criterion for tennis was grand slam title wins, for cricket the batsman's average, for baseball batting average etc... in each case a solid choice for each sport in question. The only one that I noticed that could be questioned was using goals per game for footballers as that is not a commonly used stat. Please stop trying to critique what you clearly do not understand. "Of course there is very large correlation between persons that are "several standard deviations better" in their sport, and persons that are the only exceptional player in their sport" What the hell is that meant to mean? What correlates between them? If you can find a good cite to say the Davis' work is flawed then cite it otherwise stop trying to flog the dead horse of your bogus OR here thanks very much. --LiamE (talk) 15:19, 6 December 2007 (UTC)

I think you miss the point. Check the archives of this page, to find the source of the section in the article. You will see that it was Charles Davis who said that a tennis player achieves a bradmanesque achievement when he/she conquers 15/20 grand slam titles in 10 years. Not me. Charles Davis did the maths. Charles Davis is a statistician. There is no need to go into deep statistical analysis here, because Charles Davis already did this. And his result was that Bradmans achievement is at the same level as a tennis player that wins 15/20 grand slam titles in 10 years. All I do is point out the fact that Steffi Graf did this. Nothing more, nothing less. We don't have to do statistics ourselves (which is good because you even seem to be unaware of correlation), it's just simple logic. (btw: Davis' work is not peer-reviewed.) But check out Phanto's comment below, this might solve some things.--Pie.er (talk) 15:57, 6 December 2007 (UTC)

You are missing the point: Davis is attempting to measure the gap between Bradman's batting average and the next best batsmen, then comparing it to other sports and selecting the best statistically performed player and measuring the gap to the second best. It's that simple; I haven't read the book so I don't know the methodology but that is the basis of the comparison, not "Bradman is better than Steffi Graf/Ty Cobb/Tiger Woods/that bloke what won 78 world championships in some sport most people have to have spelled to them to know what it is". The reason the table is in the biog is that Bradman is the only sporting figure about which it is claimed he is the best sports person ever. Of course it is a matter of opinion and can never be resolved; but the fact that the claim is made is notable. Look up Michael Jordan's article, is this claim made? no. Pele, uh-uh, no. So all Davis is trying to do is apply a statistical basis to a matter of opinion and help contextualise Bradman's dominance of his sport, particularly for the many people unfamiliar with cricket. Phanto282 (talk) 10:00, 6 December 2007 (UTC)

Okay, I can agree with this. But then the article is not clear... --Pie.er (talk) 14:00, 6 December 2007 (UTC)