Script analysis & statistical evaluation

Zensteve

Zensteve

Prolly old news to a bunch of ya'; maybe not for others. Figured it worth a link, anyway.

http://www.nytimes.com/2013/05/06/b...ilm-script-with-data.html?pagewanted=all&_r=0

For as much as $20,000 per script, Mr. Bruzzese and a team of analysts compare the story structure and genre of a draft script with those of released movies, looking for clues to box-office success. His company, Worldwide Motion Picture Group, also digs into an extensive database of focus group results for similar films and surveys 1,500 potential moviegoers. What do you like? What should be changed?
Click to expand...

.
 
Makes me think of this, about how poorly hedge fund managers do compared to basic index funds:

Tips From Wall Street Hedge Fund Gurus Fail to Reward Faithful

And yet they keep raking in the money for themselves because people keep paying them. I wonder what "Worldwide Motion Picture Group"'s track record is?

I actually suspect their role is more CYA than any material impact on the films performance. If an executive greenlights a film and it does poorly they can always point to the 'research' and say "hey, we did everything they said we needed to do to make it a success, so the fact that it failed must have been something outside of our control".
 
I actually have a statistics and math background.

Just like any profession, there are plenty of bad statisticians. Or plenty of audiences and consumers who don't know the difference between math and insight?

One of the biggest maxims that I'm sure some of you may remember from your stats class (if you ever took one) is this:

Correlation doesn't imply causation.

With enough data, you can find correlations in the most random things. For example, if you discovered that people choose candy over popcorn at a movie theater concession 80% of the time when the sky is blue - does that mean that the sky being blue influences their choice at the concession? (or even more ridiculous - that choosing candy over popcorn causes the sky to turn blue). If you flipped a coin and 8 times out of 10 it is tails, does that mean there's a 80% chance of landing tails with all coins? And if you tested the coin to ensure that it wasn't weighted (so that the true probability is 50%), does that mean that your chances of getting heads on the next toss will be greater than 50% because you already got tails 8 of 10 times previously?

Sports is often the worst for this in coming up with meaningless correlations. For example, until a few days ago, there was a stat that the Golden State Warriors had never beaten the Spurs at home since 1997. Think about how meaningless that stat is in predicting the likelihood of winning the next home game, or the relevance in the entire series (hint: what does the 1997-98 lineup for both those teams have anything to do with THIS current series? Or 2001-02, or any year for that matter other than THIS YEAR where you can match up comparable lineups).

Stats has its place in sports, but can often be abused by people who don't understand what is meaningless correlation and true causation drives correlation.

Stats can be meaningful if its backed up by old fashioned CRITICAL THINKING.

What this guy is peddling to dumb studio executives is a series of correlations. Again, any one with a basic stats background can do a regression analysis. But what this guy CANNOT do is explain what causes these correlations, or what separates random junk from meaningful insight beyond the numbers.

And all of this assumes that he took a rigorous approach to handling the data. Junk data = junk results.
 
Last edited:
Zensteve said:
...and a response by the Incredible Hulk...
Click to expand...
I've looked into opening a coffee shop. For less than $20k you could get one of these:

TN-2.jpg


Mmm... coffee.
 
You must log in or register to reply here.
Back
Top