If you have watched ‘Soaked in Bleach’, you can easily understand why it is making Courtney Love and her myrmidons nervous, very nervous. With a long series of facts and evidence, the movie very clearly questions the suicide theory, and points to police negligence while building the case for a reexamination.
A few days ago, Courtney Love sent her lawyers to intimidate every theater that was showing the movie with cease and desist letters, and since she did not succeed very well (it only worked in one theater in Hollywood), she is trying something else…. A Reddit user, who has experience with surveys and softwares dealing with preference aggregation, has noticed something even more insidious on IMDb.
On the IMDb page for ‘Soaked in Bleach’, users can grade it like for any other movie, and Ben Statler’s drama-documentary has currently received a 3.9 out of 10 by 4,021 voters. However, Reddit user ParrhesiaJoe was surprised by the look of the statistics — you just have to click on the number of users who have voted to get the details. If you have done a bit of of statistics, you know that a normal distribution looks like a bell curve with a pick around the average. Movie voting can lead to very divided opinions and you will not get a perfectly normal distribution because of extreme opinions, but as ParrhesiaJoe notes, you should get a ‘bulge around the mean with spikes at the extremes’. He then goes into a full demonstration taking a few examples of other movies, proving he is absolutely right. Whatever you consider a very popular movie or a very controversial one, such as ‘Bowling for Columbine’, ‘Fahrenheit 9/11’, or ‘Wal-Mart: The High Cost of Low Price’, their stats profiles always show a bulge at the mean and some spikes at the extremes. I checked myself for a few others, ‘The Last Temptation of Christ’, not such dividing opinions after all, with a bulge at 8 (the mean is 7.6 ) and some spiking at 1… totally fitting the profile. ‘Antichrist’? This one has to be hated by a lot of people, but you still see a bulge around 7 (the mean is 6.6) and some spiking at 1, the lowest score you can give on IMDb. In any case you can see this bell curve around the mean….
However, when you look at ‘Soaked in Bleach’ stats profile as ParrhesiaJoe notes, you don’t get this at all! There is simply no bell curve around the mean 3.9, just huge spiking at the extremes, 1 and 10… so it is as if almost all the 4,021 IMDb users had voted at the extremes, which is statistically not very credible. There is something in statistics called standard deviation, which measures the amount of dispersion of a set of data around the mean, and something sounds fishy here.
‘This doesn’t even resemble a real movie. From a first look, you can see that about 1000 of the votes follow a normal curve centering at 10. The other 2500 votes are distributed between one and two’ says ParrhesiaJoe…. ‘Anything less than 100 for the 3 votes is HIGHLY suspicious (demonstrably 95%+)’, and the 3 vote only gets 0.2% or 7 votes out of the 4,021 in total!
‘Someone decided to kill this movie, and they didn’t have a statistician look at it, or the guy would have said, You need WAAAAAY more 2’s and 3’s. There should be a BULGE at the weighted average of 3.9. A total of 4 people out of 3,600 gave the average vote. That screams fraud so loudly that I had to write this post.’
The numbers have slighty changed since he looked at them, but it is basically the same, 8 out of 4,021 people have voted for 4, the closest grade to the average (3.9) and it looks highly suspicious.
According to the user ratings report, the attack comes from the US — US users gave an average of 1.9 for this movie, whereas non-US users gave an average of 7.5. It is clear there was a manipulation of the data because someone wants to kill this movie!
Now has IMDb a strong impact? I don’t know but it is the first page with statistics that pops up when you do a research on Google. Rotten Tomatoes is another site which allows people to vote, and if the official critics are not favorable to the movie (29% positive, what a surprise!), the audience score is 95% positive? What does it say one more time about the IMDb users statistics? They are wrong!