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 Part 2: The Rest of the Staff
Description for the analysis is in the previous list. Short explanation: a Bayesian t test is calculated to assess whether staff average review ratings are higher than the average rating, the strongest linear model among various polynomial fits (with 8 as the highest order) was selected via Bayes information criterion (values were weighted by the log10 of the number of ratings that the album had to reduce the influence of albums that had few ratings), the coefficients of the models are presented, as well as what average user rating the model predicts would correspond to a 1.0, 3.0, and 5.0 review. | 1 | | The Stooges The Weirdness
Dave de Sylvia
Rating #s:
317.8 +- 48.5 Ratings
Bias:
-0.055 rating bias, Bayes Factor = 0.147 (moderate evidence that there is no bias)
Model:
2.21 + 0.36x, R^2 = 0.394
1.0 Review = 2.57 Average Sputnik Rating
3.0 Review = 3.3 Average Sputnik Rating
5.0 Review = 4.02 Average Sputnik Rating | 2 | | David Guetta Nothing but the Beat
Deviant.
Rating #s:
207.8 +- 32.8 Ratings
Bias:
-0.01 rating bias, Bayes Factor = 0.128 (moderately strong evidence that there is no bias)
Model:
1.61 + 0.68x + -0.04x^2, R^2 = 0.752
1.0 Review = 2.25 Average Sputnik Rating
3.0 Review = 3.31 Average Sputnik Rating
5.0 Review = 4.06 Average Sputnik Rating | 3 | | Logic Under Pressure
GnarlyShillelagh
Rating #s:
269.1 +- 54.7 Ratings
Bias:
-0.0025 ratings bias, Bayes Factor = 0.232 (moderate evidence that there is no bias)
Model:
2.58 + 0.28x, R^2 = 0.277
1.0 Review = 2.86 Average Sputnik Rating
3.0 Review = 3.42 Average Sputnik Rating
5.0 Review = 3.98 Average Sputnik Rating | 4 | | Squarepusher Music for Robots
Hyperion1001
Rating #s:
59.8 +- 9.9 Ratings
Bias:
+0.148 ratings bias, Bayes Factor = 0.348 (weak evidence that there is no bias)
2.42 + 0.33x, R^2 = 0.429
1.0 Review = 2.76 Average Sputnik Rating
3.0 Review = 3.43 Average Sputnik Rating
5.0 Review = 4.09 Average Sputnik Rating | 5 | | Good Charlotte Cardiology
insomniac15
Rating #s:
80.4 +- 15.8 Ratings
Bias:
+0.204 rating bias, Bayes Factor = 367.7 (very strong evidence that there is bias)
Model:
0.84 + 0.71x, R^2 = 0.503
1.0 Review = 1.55 Average Sputnik Rating
3.0 Review = 2.96 Average Sputnik Rating
5.0 Review = 4.38 Average Sputnik Rating | 6 | | Kerridge Always Offended Never Ashamed
Jacquibim
Rating #s:
179.1 +- 29.0 Ratings
Bias:
-0.0096 rating bias, Bayes Factor = 0.159 (moderately strong evidence that there is no bias)
Model:
2.21 + 0.38x, R^2 = 0.586
1.0 Review = 2.59 Average Sputnik Rating
3.0 Review = 3.35 Average Sputnik Rating
5.0 Review = 4.11 Average Sputnik Rating | 7 | | Questions Happiness January
JohnnyOnTheSpot
Rating #s:
43.9 +- 8.3 Ratings
Bias:
-0.065 rating bias, Bayes Factor = 0.232 (moderate evidence that there is no bias)
Model:
-1.39 + 2.26x + -0.24x^2, R^2 = 0.570
1.0 Review = 0.64 Average Sputnik Rating
3.0 Review = 3.28 Average Sputnik Rating
5.0 Review = 4.04 Average Sputnik Rating | 8 | | Dope Stars Inc. Terapunk
Metalstyles
Rating #s:
101.5 +- 18.0 Ratings
Bias:
+0.186 rating bias, Bayes Factor = 3.473 (weak evidence that there is bias)
Model:
2.06 + 0.37x, R^2 = 0.407
1.0 Review = 2.44 Average Sputnik Rating
3.0 Review = 3.18 Average Sputnik Rating
5.0 Review = 3.92 Average Sputnik Rating | 9 | | Omar Rodriguez-Lopez Despair
SgtPepper
Rating #s:
291.6 +- 68.3 Ratings
+0.276 rating bias, Bayes Factor = 331.1 (very strong evidence that there is bias)
Model:
1.44 + 0.58x, R^2 = 0.592
1.0 Review = 2.01 Average Sputnik Rating
3.0 Review = 3.16 Average Sputnik Rating
5.0 Review = 4.31 Average Sputnik Rating | 10 | | Micol Cazzell Broken Things
StrangerofSorts
Rating #s:
136.2 +- 30.4 Ratings
Bias:
+0.143 rating bias, Bayes Factor = 0.825 (very weak evidence that there is no bias)
Model:
2.42 + 0.34x, R^2 = 0.536
1.0 Review = 2.76 Average Sputnik Rating
3.0 Review = 3.45 Average Sputnik Rating
5.0 Review = 4.13 Average Sputnik Rating | 11 | | dredg Chuckles & Mr. Squeezy
VheissuCrisis
Rating #s:
185.5 +- 27.9 Ratings
Bias:
+0.002 rating bias, Bayes Factor = 0.186 (moderate evidence that there is no bias)
Model:
2.07 + 0.42x, R^2 = 0.415
1.0 Review = 2.49 Average Sputnik Rating
3.0 Review = 3.33 Average Sputnik Rating
5.0 Review = 4.16 Average Sputnik Rating | 12 | | Iwrestledabearonce It's All Dubstep
Xenophanes
Rating #s:
299.6 +- 32.0 Ratings
Bias:
-0.076 rating bias, Bayes Factor = 0.243 (weak evidence that there is no bias)
Model:
2.45 + 0.35x, R^2 = 0.453
1.0 Review = 2.80 Average Sputnik Rating
3.0 Review = 3.49 Average Sputnik Rating
5.0 Review = 4.18 Average Sputnik Rating
Notes: Xenophanes has a review for an album that technically does not exist in the sputnik database (http://www.sputnikmusic.com/review/39564/Iniciativa-Del-Cambio-Abril/). That page kept crashing my code. | |
macman76
03.10.16 | the numbers | macman76
03.10.16 | request for the sputnik community: does anyone know if there is an announcements/info page for when users are designated staff/contrib/emeritus? | klap
03.10.16 | usually comes up in the news post i believe. or staff blog | Feather
03.10.16 | Okay I understand the idea you're going for here, but when it comes down to it I feel like all this is showing is how much a staff writers ratings match up with the general sputnik community. For example, here you showed that SgtPepper's review ratings generally match the average user ratings for the album. I get that you are trying to show that the average user ratings match the staff reviewers rating, but I feel like this is the other way around unfortunately. | macman76
03.10.16 | I'm not claiming any directionality in regards to which causes which | Pon
03.10.16 | Well I can sleep easy now being moderately certain that I am not biased, ty | Feather
03.10.16 | @macman okay fair. What would be interesting is if we had data on what the average rating is prior to a staff review and then months post staff review and see if they begin to allign themselves with the staff reviewers rating or not. But normally the staff reviews new albums very quick so there wouldnt be enough data for before the staff review for this to work. | macman76
03.10.16 | true, feather, there wouldn't be enough data for the staff members that review albums with very few ratings (like JohnnyOntheSpot). If you have any ideas for what I should look into next, I'm all ears. Currently, I'm thinking of building a model to predict year end list albums and using ratings an text analysis of words in reviews to make similarity scores between users. | Feather
03.10.16 | @macman do you just do this in your free time? What is your job/course of study? What software are you using? | macman76
03.10.16 | This is free time/learning excercise for me, I work as a research assistant, and I do most of this in R and am also trying to learn Python. | Feather
03.10.16 | No way haha. I just downloaded R and Python today. I get bored at work and I figured learning the two of those would be productive. | macman76
03.10.16 | They're both free! For Python, download anaconda, it comes with a ton of libraries and an ide. For r, download rstudio. Here's also a free book on Python (http://greenteapress.com/wp/think-python/) and r (https://leanpub.com/rprogramming) | Feather
03.10.16 | thanks man! I really appreciate it! Im an intern in continuous improvement and we have really hit a valley as far as work I am able to do on projects so I figured I may as well do something productive. | Spacesh1p
03.10.16 | Macman, thanks for another cool list. Feather, I like your point about staff reviews, that would be really interesting.
I would love to see something in the same vein as Feather's comment that might get the juices going on this site, but my mind doesn't function this way so I wouldn't know how to go about setting it all up. If there's sufficient data, would be cool to track ratings pre/post a major reviewer like Pitchfork or Fantano. | macman76
03.10.16 | I think it's doable though would require a ton of work, also interesting to see whether outside reviews increase the amount of ratings | Feather
03.14.16 | Fantano is definitely making an impact on peoples opinions | Asdfp277
03.14.16 | spiderty fartano | VheissuCrisis
03.14.16 | "moderate evidence that there is no bias." Phew. | Feather
03.14.16 | @Macman thanks for the links to the textbooks! I just downloaded them and I am excited to learn a bit of it! | macman76
03.15.16 | no probs feather, if you ever need a stats book rec or something hit me up
@vheissu the jury is still out, don't get too comfortable |
|