User
Soundoffs 4 Album Ratings 142 Objectivity 62%
Last Active 09-25-20 9:08 pm Joined 11-12-11
Review Comments 2,122
| Staff Metrics
I analyzed data from a few staff members. I looked to see whether their review ratings coincided with the average ratings of the sputnik users. My analysis says nothing of the causality. Are some staff members more influential than others, more similar to the Sputnik community, more contrarian, better looking? Those aren’t things this answers. Sorry. Also listed are their mean number and 68% confidence interval of the mean of the amount of ratings the albums they reviewed had, their bias relative to the average user rating (positive means that they rated higher than the users) and the Bayes factor of the comparison (values above one mean that there is evidence that staffs ratings are significantly biased, values below one mean evidence against the bias hypothesis), the polynomial statistical model that best fit the data (as assessed by Bayes information criterion with 8 as the highest order) and what the model predicts as the average score of an album that has a 1.0, 3.0, and 5.0 review from each staff member (essentially, what does a staff member’s 1.0, 3.0, and 5.0 review generally mean for the average rating). Values in each model were weighted by the log10 of the number of ratings that each album reviewed had. I chose to do this to reduce the influence of albums that had few ratings (which would, by small n, have more noisy estimates) while still not granting a ton of weight to the most rated albums.
Albums listed represent the most outlying review of the staff member based on marginal medians calculated projection distance of review scores to the average rating score. This isn’t the biggest difference between the staff member’s review and the average Sputnik score (in fact, Omaha’s review score of Cypress Hill is a 2.5 and the actual score was a 2.2, he/she had other ratings that had above 0.8 rating units difference), but rather the review that deviated most from their joined trend. | 1 | | Justin Bieber My World 2.0
Klap
Rating #s:
170.0 +- 16.1 Ratings
Bias:
+0.027 rating bias, Bayes Factor = 0.114 (strong evidence that there is no bias)
Model :
-0.5 + 2.6x-0.6x^2 + 0.1x^3 , R^2 = 0.63 (rank 2)
1.0 Review = 1.6 Average Sputnik Rating
3.0 Review = 3.3 Average Sputnik Rating
5.0 Review = 4.0 Average Sputnik Rating
Notes:
A third order polynomial best fit Klap’s rating. It may be partially be related to the fact that Klap has a lot of unique ratings (he has a review with almost every possible number that staff are allowed to rate with like 2.3 and 4.1). More points may make it more likely for a curve rather than a straight line to best fit a set of points (like a circle being more apparent when you’ve demonstrated it with 100 points rather than 3). At the same time, no albums reach the average rating of 1.0 or 5.0 even though individual reviewers will give that rating. Models of other reviewers show this compression; however, for klap, his/her tails rise faster than the center of his/her model. | 2 | | The Crash Motive Consequence
SowingSeason
Rating #s:
394.6 +- 45 Ratings
Bias:
+0.178 rating bias, Bayes Factor = 2.7 (weak evidence that there is bias)
Model :
1.8 + 0.5x, R^2 = 0.57 (rank 3)
1.0 Review = 2.3 Average Sputnik Rating
3.0 Review = 3.2 Average Sputnik Rating
5.0 Review = 4.1 Average Sputnik Rating
Notes:
By sheer luck (or a sign of larger trend) the next three reviewers have almost the same model (note that I rounded the models values). Sowing is not a fan of his/hers non-.5-or-.0-decimal-which-makes-more-continuous-and-reliable-models. We could all benefit from a little nuance. | 3 | | The Summer Set Legendary
Atari
Rating #s:
336 +- 50 Ratings
Bias:
+0.27 rating bias, Bayes Factor = 125 (very strong evidence that there is bias)
Model:
1.8 + 0.5x, R^2 = 0.65 (rank 1)
1.0 Review = 2.28 Average Sputnik Rating
3.0 Review = 3.25 Average Sputnik Rating
5.0 Review = 4.2 Average Sputnik Rating
Notes:
Our first probably biased reviewer. One discrepancy I have to admit, however, is that though I weighed the linear model to add importance to the more often rated albums, I was not able to do that in order to properly calculate a bias with those weights and their corresponding Bayes Factor. So, all Bayes Factors assume equal weight, which should make you a little more skeptical of their interpretation. Also, of these 8 staff members, Atari had the highest R^2 indicating that his/her model fits its trend really well. | 4 | | Jessica Simpson Do You Know
Willie
Rating #s:
136.7 +- 14.9 Ratings
Bias:
+0.020 rating bias, Bayes Factor = 0.10 (strong evidence that there is no bias)
Model:
1.8 + 0.5x, R^2 = 0.48 (rank 6)
1.0 Review = 2.28 Average Sputnik Rating
3.0 Review = 3.25 Average Sputnik Rating
5.0 Review = 4.2 Average Sputnik Rating | 5 | | Cypress Hill x Rusko Cypress Hill x Rusko
Omaha
Rating #s:
358.9 +- 45 Ratings
Bias:
+0.09 rating bias, Bayes Factor = 0.40 (weak evidence that there is no bias)
Model:
2.6 + 0.3x, 0.297 (rank 10)
1.0 Review = 2.9 Average Sputnik Rating
3.0 Review = 3.48 Average Sputnik Rating
5.0 Review = 4.05 Average Sputnik Rating
Notes:
Omaha has our most compressed range (1.0 to 5.0 review for him/her translates to a 2.9 to 4.05 range). His/Her model is also the least related to the average user rating (as assessed by his R^2). A positive spin of this information would be that his/her ratings are unique and unaffected by the masses (or nobody likes his/her opinion and his/her review value will be unrelated to your enjoyment of said album). | 6 | | Jimmy Eat World Stay on My Side Tonight
DaveyBoy
Rating #s:
279 +- 27.3 Ratings
Bias:
+ 0.021 rating bias, Bayes Factor = 0.098 (strong evidence that there is no bias)
Model:
0.9 + 1x -0.1x^2, R^2 = 0.538 (rank 4)
1.0 Review = 1.84 Average Sputnik Rating
3.0 Review = 3.19 Average Sputnik Rating
5.0 Review = 3.9 Average Sputnik Rating | 7 | | Chroma Key Dead Air For Radios
Voivod
Rating #s:
61.2 +- 8.9 Ratings
Bias:
+0.22 ratings bias, Bayes Factor = 131.9 (strong evidence that there is bias)
Model:
2.1 + 0.4x, R^2 = 0.306 (rank 9)
1.0 Review = 2.49 Average Sputnik Rating
3.0 Review = 3.32 Average Sputnik Rating
5.0 Review = 4.14 Average Sputnik Rating
Notes:
Voivod reviews some real underground shit as evidenced by the fact that the average number of ratings of said albums is unable to collect social security. Given the small standard error of the number of ratings, a weighting that favored the more rated albums would probably not change his bias value much, so his Bayes Factor may be accurate he may actually be positively biased. | 8 | | Fran Healy Wreckorder
Irving
Rating #s:
142 +- 30.2 Ratings
Bias:
-0.036 ratings bias, Bayes Factor = 0.151 (strong evidence that there is no bias)
Model:
-1.7 + 4.4x -1.3x^2 + 0.1x^3, R^2 = 0.443 (rank 8)
1.0 Review = 1.5 Average Sputnik Rating
3.0 Review = 3.28 Average Sputnik Rating
5.0 Review = 4.27 Average Sputnik Rating
Notes:
The only bias that’s negative (though probably not significantly so) and our widest range for the user rating. Irving’s model fit also looks a lot like klap’s (thank you unique review values). | 9 | | Megadeth Super Collider
AtomicWaste
Rating #s:
239 +- 36
Bias:
0.165 ratings bias, Bayes Factor = 0.916 (very little evidence in favor of the no bias hypothesis)
Model:
2.26 + 0.36x, R^2 = 0.531 (rank 5)
1.0 Review = 2.5 Average Sputnik Rating
3.0 Review = 3.35 Average Sputnik Rating
5.0 Review = 4.08 Average Sputnik Rating | 10 | | Skrillex Recess
Brostep
Rating #s:
134 +- 50.9 ratings
Bias:
0.111 ratings bias, Bayes Factor = 0.27 (weak evidence that there is no bias)
Model:
1.9 + 0.4x, R^2 = 0.452 (rank 7)
1.0 Review = 2.3 Average Sputnik Rating
3.0 Review = 3.1 Average Sputnik Rating
5.0 Review = 3.9 Average Sputnik Rating | |
macman76
03.02.16 | I want to scrape more data from staff/contribs/emeritus... if you guys could list as many staff user names as you could that would be great. Also if anyone wants the data from their reviews, just ask and I'll post it on your shoutbox (or somehow get it to you).
Edit: somehow I missed that the list of staff/etc. is at the bottom of the site. | Relinquished
03.02.16 | oh my god for the love of odin delete this list | Apollo
03.02.16 | "I analyzed a data from a few staff members"
Stopped reading here. If you can't use proper grammar in one sentence........ | macman76
03.02.16 | *their | Hep Kat
03.02.16 | list is shit lol butts | ChoccyPhilly
03.02.16 | wow this has depth to it | Ashen
03.02.16 | 😂😂😂 | hal1ax
03.02.16 | you might need a new hobby | Conmaniac
03.02.16 | well shit | Piglet
03.02.16 | fuckin got turing up in here searching around in staff's arses | macman76
03.02.16 | I also play guitar, hal1ax...not as thrilling as for loops | granitenotebook
03.02.16 | this is interesting. if you get real bored you could try with contributors or, if you want a lot of attention, particularly controversial users | Asdfp277
03.02.16 | i love polynomials | MrSirLordGentleman
03.02.16 | I'm a humanities guy, what the hell are all of these numbers | Asdfp277
03.02.16 | basically, everyone disagrees with klap | klap
03.02.16 | amazing | user
03.02.16 | props for the effort but can't say I read more than 1/10th of what you wrote. | RogueNine
03.02.16 | With the time this must have taken, please tell me you did this for school or something. | Asdfp277
03.02.16 | being educated is embarrassing agreed | DoofusWainwright
03.02.16 | I think I get it | DoofusWainwright
03.02.16 | Actually, no I don't | RadicalEd
03.02.16 | What is this numbers. | RogueNine
03.02.16 | Yeah you're spot on. | Archelirion
03.02.16 | Interesting stuff. Love a good set of statistics :] | Irving
03.02.16 | This is a phenomenal piece of work. Wow. | Hep Kat
03.02.16 | wat | Pangea
03.02.16 | pretty interesting | AtomicWaste
03.02.16 | If you're interested, the full list of staff is here:
http://www.sputnikmusic.com/staff.php
I'm interested to see how much bias I have based on these calculations haha | Calc
03.02.16 | i think i get it...i really think i get it.. | macman76
03.02.16 | wow, this got featured.
I'll post your info after work, AtomicWaste.
I did write a script to do all of this, minus the write up. An example of the data that i scraped is:
"","Band.Album","score","date","nratings","avscore","replies","staff"
"1","Deep Sea Diver Secrets",3.5,"2016-02-07",5,3.4,9,"Irving"
"2","Narc Twain Narc Twain",3.8,"2015-11-22",16,3.1,23,"Irving"
"3","Iron Maiden The Book of Souls",4,"2015-09-08",582,3.9,338,"Irving"
"4","Disturbed Immortalized",3,"2015-08-16",206,2.5,176,"Irving"
"5","Elbow Lost Worker Bee",2.7,"2015-07-26",12,2.7,14,"Irving"
| Asdfp277
03.02.16 | can u make an algorithm to make my life less shitty | ZackSh33
03.02.16 | can you share the code you used to get this on git or something? this is pretty cool but I'm having a hard time understanding the conclusions of your work. | Sniff
03.02.16 | What even. Mathematical statistics is the most boring thing ever tbh. | Feather
03.02.16 | This could have been presented far more eloquently. Hate to admit it, but this actually looks fun to me. - Industrial Engineer | Alastor
03.02.16 | best list | hal1ax
03.02.16 | this makes my penis soft | Ashen
03.02.16 | makes my angular gyrus erect tho | Mystletainn
03.02.16 | Wtf man, WHO does this? | UpwardSpiral
03.02.16 | tldr | Asdfp277
03.02.16 | what's industrial engineering? | Brostep
03.03.16 | yo! quick question: are there any tutorials on how to write scripts that scrape data like this? I have some ideas for projects like this re: Sput but dunno how to program well enough in any language that isn't Java haha
this is super damn cool and v much appreciate it. do me too! | Aftertheascension
03.03.16 | really good and well done list, shit thread | Ashen
03.03.16 | @brostep
i don't have any links ready but most ANOVA software should give you what you're looking for | theacademy
03.03.16 | i am actually the staff's primary tastemaker (you can see how many of my 5's make it to best new music) | Jots
03.03.16 | _| ̄| ⍨⃝ | Kman418
03.03.16 | https://twitter.com/bikinibabelover/status/618497469890977792 | Lord(e)Po)))ts
03.03.16 | this is quite possibly the stupidest fucking list ive ever seen on this site | theacademy
03.03.16 | dont insult me i have made way dumber lists | theacademy
03.03.16 | http://www.sputnikmusic.com/list.php?memberid=428955&listid=37295 | theacademy
03.03.16 | wait that wasnt me | Titan
03.03.16 | still wondering how they gauge the difference between a 3.8 and a 3.7
idiots [2] | someguest
03.03.16 | Like we needed metrics to know they're all idiots. | someguest
03.03.16 | Like we needed metrics to know they're all idiots. | Jots
03.03.16 | kinda like how they gauge the difference between a 3.5 and a 4 | someguest
03.03.16 | Like we needed metrics to know they're all idiots. | macman76
03.03.16 | @brostep
I scraped all of the data with R, specifically the 'XML' package, there are plenty of tutorials for reading html tables with that package... working on putting your data up now | macman76
03.03.16 | anybody want to learn R, this book is free and short... https://leanpub.com/rprogramming | macman76
03.03.16 | Two new staff members up | Ashen
03.03.16 | ignore pots. it may be stupid, but it's also pointless. | Brostep
03.03.16 | @macman thanks a ton! I'm learning R for a statistics/machine learning class atm but am only four weeks into the semester and as such am not great yet, but it seems like pretty simple coding thus far (esp since I'm four semesters into my CS major track). will give it a shot! | macman76
03.03.16 | I majored in psychology and learned my first computer language a year and a half ago (matlab), so I'm sure you'll get it faster than i did. | Brostep
03.03.16 | haha hopefully! do you know of any good tutorials thus far? I've googled around and can find a slideshow from UC Berkeley that looks decent but nothing else yet | macman76
03.03.16 | That book I linked to is really good. | Brostep
03.03.16 | Oh wow I already have that book for my class. Will look through it thanks so much! | StreetlightRock
03.03.16 | This is awesome. | AtomicWaste
03.03.16 | I do a lot of analysis in Spotfire which features some R programming capability I've not yet tapped, so I may have to check out this book too. Thanks for putting up my stats! It's interesting to see how my tastes equate to the site averages. | Aftertheascension
03.03.16 | Lordepots showing his stupid ass once again | NordicMindset
03.04.16 | sabersputnik | silentstar
03.05.16 | this is actually really cool and one of the more interesting things to exist on this website compared to the 99.9% shit tier lists good job man
| p4p
03.05.16 | u should work for the site | Irving
03.06.16 | Honestly I always thought that I tended to rate albums a lot higher than they should be scored. The fact that I have actual, empirical evidence suggesting otherwise is pretty cool. | ShadowRemains
03.06.16 | mj has returned | macman76
12.28.16 | this is quite possibly the stupidest fucking list ive ever seen on this site
Bump |
|