Jethro Tull’s "Thick As a Brick" #1 Prog?

bluetailfly wrote:   This is getting tedious, oh possessed one

10 total posts and you're already getting tired. That doesn't bode well for your future o little troll. Unless of course you're really a nickname for someone else on the forum. In which case you should get out of the closet and reveal the erudite scribbler you reeeealy are.

richardh wrote: The ranking sytem is quite clearly useless: 1. Thick As A Brick - 38 reviews ,84% 5 star reviews, thats about 33 'masterpeice' ratings 2. Selling England By The Pound - 113 reviews ,82% 5 star reviews,thats about 100 'masterpeice' ratings 3. Close To The Edge - 142 reviews,87% 5 star reviews,that's about 125 'masterpeice' ratings Even down at No 48. you've got ELP 'Brain Salad Surgery' with 68 entries and 62% 5 star reviews, that's more 5 star reviews than the No1 album! I would be happy if Close To The Edge was NO1.I'm slightly baffled as to why it isn't.

I believe that there is a problem with the current algorithm. It appears that the original algorithm was ditched recently, firstly in order to take everyone's ratings into account - not just the Collaborators reviews - and secondly to calculate differently the overall rating value of an album. I have sent a PM to M@X with an explanation of why the current algorithm is wrong plus a proposal for a new algorithm that would appear to be better than the current one and also better than the original algorithm.

BebieM wrote: Fitzcarraldo wrote: BebieM wrote: no, i think that simply taking the amount of 5star reviews to create a list would be the best way. I can't think of any case that this wouldnt fit to. Not necessarily. Consider for example the case where Album A has, say, ten 5-star ratings (only) and Album B has, say, one hundred 4-star ratings (only). It could be argued that Album B is more popular than Album A.   oh, gotcha now. Yes, you are right. But does this case exist on the archives? since reviewers tend to give a lot of 5 star-ratings ....

I bet there are similar instances in the Archives, albeit perhaps not as extreme as my example, and I'm sure this type of scenario could occur when Album A is from one of the obscure bands. Just taking 5-star reviews into account would not be the best way to assess popularity. Statisticians who perform market research surveys would not discard ratings less than 5-stars if they were using a rate-from-1-to-5 survey form for different brands of biscuits (say), would they?

Possessed wrote: bluetailfly wrote:   This is getting tedious, oh possessed one  10 total posts and you're already getting tired. That doesn't bode well for your future o little troll. Unless of course you're really a nickname for someone else on the forum. In which case you should get out of the closet and reveal the erudite scribbler you reeeealy are.

C'mon guys.

Take a hit.

Fitzcarraldo wrote: BebieM wrote: Fitzcarraldo wrote: BebieM wrote: no, i think that simply taking the amount of 5star reviews to create a list would be the best way. I can't think of any case that this wouldnt fit to. Not necessarily. Consider for example the case where Album A has, say, ten 5-star ratings (only) and Album B has, say, one hundred 4-star ratings (only). It could be argued that Album B is more popular than Album A.   oh, gotcha now. Yes, you are right. But does this case exist on the archives? since reviewers tend to give a lot of 5 star-ratings ....  I bet there are similar instances in the Archives, albeit perhaps not as extreme as my example, and I'm sure this type of scenario could occur when Album A is from one of the obscure bands. Just taking 5-star reviews into account would not be the best way to assess popularity. Statisticians who perform market research surveys would not discard ratings less than 5-stars if they were using a rate-from-1-to-5 survey form for different brands of biscuits (say), would they?

no, they wouldnt.

ok, so why not a system like

5 points for a 5star-rating, 3 points for a 4star-rating, 1point for a 3star-rating, 1 point off for a 2star, 3 points off for a 1star, 5 points off for a 0-star? .... or the same thing with (3, 2, 1, -1, -2, -3)

???

Id give the album 5 stars..

heck its one of my favourite of all time, i think if CTTE and foxtrot were there its reasonable that thick as a brick is.

ivan_2068 wrote: Thick as a Brick is not my choice for N° 1, but I don't feel surprised, it has excellent music, intelligent lyrics, imaginative concept and superb interpretations, a 5 stars album. There is another fact, Jethro Tull is a band that almost doesn't has haters, everybody likes Ian Anderson and feels he's a very talented musician, many people hate Yes, Genesis, ELP, etc, but very few hate Jethro Tull. Why can't people choose Thick as a Brick for N° 1? If someone told me Invisible Touch., Big Generator or Love Beach is N° 1, then I would feel surprised and maybe angry. Richardh wrote: The ranking sytem is quite clearly useless: 1. Thick As A Brick - 38 reviews ,84% 5 star reviews, thats about 33 'masterpeice' ratings 2. Selling England By The Pound - 113 reviews ,82% 5 star reviews,thats about 100 'masterpeice' ratings 3. Close To The Edge - 142 reviews,87% 5 star reviews,that's about 125 'masterpeice' ratings Even down at No 48. you've got ELP  with 68 entries and 62% 5 star reviews, that's more 5 star reviews than the No1 album! I would be happy if Close To The Edge was NO1.I'm slightly baffled as to why it isn't. True, but on the other hand: Thick As A Brick: 84% of 5 stars and 16% of 4 stars, no 3, 2, 1 or 0 stars rating. Selling England By The Pound: 82% 5 stars and 18% 4 stars no 3, 2, 1 or 0 stars rating. Close To The Edge: 87% 5 stars, 8% 1 star but 1% 3 stars ratings so the average is lower. Brain Salad Surgery: 62% 5 stars, 26% 4 stars, but 4% 3 stars and even 6% 2 stars rating Don't be simplistic, the average is obtained from all the ratings not only the 5 stars, if you get an album with 50% 5 stars and 50% 0 stars would be even lower. Iván

But I believe that negative reviews are often prejudicial and 'dishonest'.I would rather know what albums inspire people.Clearly Selling England By The Pound and Close To The Edge are well ahead in this respect.The one person who gave Close To The Edge a ridiculous 1 star rating has messed up the whole thing.I'm seriously tempted to give Thick As A Brick 1 star just to redress the balance but that goes against the grain.Even though I'm not impressed by DSOTM I still gave it 3 stars.It's not an album I can dismiss despite my own feelings.Other people don't follow the guidelines in my view.

As a suggestion then ..why don't we ignore,say, the top five per cent and bottom five per cent ratings for the sake of calculating a meaningfull average score? Wouldn't this then give a more representative ranking? We should the 'hysterical' ratings mess up the whole thing? 

The ProgArchives star ratings are ordinal data, and averaging is normally not a good (or correct) way to evaluate popularity when using ordinal data, although it is sometimes used if the ordinal data is converted to numerical data AND it follows approximately a Normal (a.k.a. Gaussian) Distribution. However, averaging is definitely NOT the way to evaluate the data if it does not approximate a Gaussian Distribution and is instead e.g. a bimodal distribution, or a uniform distribution, or a skewed distribution etc. Furthermore, averaging ordinal data when the sample size per album (let alone the people comprising the sample) is not the same is also fraught with problems. Sometimes the median is used instead of the mean (average) for ordinal data, but for reasons I won't go into now that would not be appropriate here either.

To illustrate how the use of averages in this type of 'survey' can lead to incorrect rankings, consider the case where Album A has fifty 5-star ratings (only) and Album B has fifty 5-star ratings and fifty 4-star ratings (only). If the ordinal data is considered to be numerical data, the average rating for Album A is 5 stars and the average rating for Album B is 4.5 stars. Album B not only has more ratings than Album A, it also has the same number of 5-star ratings and all of the people who rated the album liked it (and twice as many people liked it). Yet using the average rating as a measure, Album B is ranked below Album A. This seems completely wrong to me (and hopefully to anyone else!). So I say again that the use of the average rating to rank the two albums appears inappropriate to me. This is why market researchers, medical researchers etc. who evaluate surveys generally do not use averaging to rank the different products/services rated using rate-from-X-to-Y type survey forms.

As I mentioned earlier, I have proposed a better algorithm to M@X which does not use averages - so let's see how he receives it. The use of averages for the purpose of producing the Top 50 must be dropped in my opinion, as it is a poor way of ranking data of this particular type.

Bah. If this was a perfect world RED would be no. 1