Skip to content

Look. At. The. Data. (Hollywood action movies example)

Kaiser Fung shares an amusing story of how you can be misled by analyzing data that you haven’t fully digested. Kaiser writes, “It pains me to think how many people have analyzed this dataset, and used these keywords to build models.”


  1. numeric says:

    In actual fact, the homogenization of movies can be traced to this one book:

    though of course there were precursors. Movie execs like to say that if they made all the movies they rejected and rejected all the ones they made they would still be in the same place. I suppose the analogue for a statistician is that if they proved all the theorems they couldn’t and didn’t prove all the ones they could they would still be in the same place.

    • Andrew says:


      I don’t like that analogy: movie execs make movies, and statisticians . . . prove theorems? Can we instead say that statisticians work on the design and analysis of quantitative studies?

      • numeric says:

        Mathematicians would fit better than statisticians, I agree, but I can’t find a way to use your formulation either, since it would be something like “If I analyzed correctly the studies I analyzed incorrectly and analyzed incorrectly the ones I analyzed correctly, I’d be in the same place.” Sounds more like a social psychologist.

        Incidentally, while I’m commenting, I missed your “The Martian” discussion (I still say you want to be a fiction writer–though maybe now it’s a screenwriter, but whatever), the movie was simply unbelievable since the crew had to launch due to a sandstorm threatening to topple the ship but the Ares IV rocket was expected to sit there for x number of years–gee, wouldn’t a sandstorm topple it at some point also–certainly the probably would be high enough that you wouldn’t leave it on the Mars surface expecting it to be there untoppled when you needed it (and I’m well aware that the wind isn’t strong enough to topple a rocket on Mars, but that’s not my point–its the contradiction of having to leave after a few months due to a sandstorm threatening to topple the rocket with stashing a rocket for years that even a suspension of disbelief can’t make me stomach). If they had just had the hero fall down a gully right before launching to hit a window and the suit indicating he was dead due to instrument malfunction and the crew having to launch at a certain point or not make it back to Earth and giving him up for dead, then he could have winched himself back up out of the gully and done all of the things he did do to survive.

        • Andrew says:


          I don’t think I could be a good fiction writer because I can’t remember, or write, dialogue. I can only get conversation down right if I write it down, word for word, immediately when it happens.

          • numeric says:

            This is what is known in the trade as a non-denial denial. The fact that you wouldn’t be good doesn’t negate intent–given the amount of column inches (html code?) you devote to literature and now movies (as opposed to say, regulatory economics or human starvation) who seem to indicate more than a voyeur’s interest.

            Anyway, for your first foray into the popular arts, I would suggest a murder mystery based on the statistics profession. It would follow the plot of The Warriors, where a profession-wide meeting has been called to settle the dispute between frequentists and Bayesians once and for all. But sadly, a prominent frequentist is found murdered at this conference and the Bayesians are blamed and must run for their lives, their flight pattern being gamed out by Bayesian game theory precepts. The frequentists are hot on their trail, using Fisherian methods to analyze the path and make predictions, hoping to intercept the Bayesians and bring them to justice. Let the best statistical paradigm win. There would be a video game tie-in and on-line multi-player games which could be used to teach players statistical concepts and allow implementation of the same. Use of this game would replace academic offerings in statistics as undergraduates flock to it, negating the need for introductory statistics classes and increasing statistical literacy throughout academia.

            • Andrew says:


              You can add your movie idea to this list.

            • Andrew says:

              Also, here’s another movie idea:

              The Psychological Science Justice League

              The peaceful country of Psychologia is being terrorized by rampaging methodologists from northern Europe who, going beyond scientific argument and counterargument, threaten the land’s traditions of civility and mutual tolerance. It seems that nobody can stop these terrorists—until a team of superheroes, led by the Chair of the APS Publications board and anchored by the past president of the American Psychological Association, Eastern Psychological Association, International Association for Cognitive Education and Psychology, decide to step in. The heroes enlist various specialists, Mission Impossible-style, including the Jedi-like author of “Controlling other people: The impact of power on stereotyping,” and also the mysterious Arina K. Bones. The battle rages on, with the Justice League inexplicably having trouble getting the other hand, until it is revealed that Bones has actually tried a pre-registered replication of her own work: by questioning her own claims, Bones has revealed that she’s switched over to the Dark Side! At this point, the League gets their act together, banishes Bones to the Netherlands, and re-establishes full control of the Association for Psychological Science slate of journals. The triumphant conclusion appears in a Ted talk, where the leaders of Psychologia announce, “Consumers of popular culture cannot get enough of our field.”

              Great movie idea, also room for a sequel when it turns out that, all along, there’s been a mole within the Ted organization . . .

              • numeric says:

                Hadn’t seen your earlier movie post, but it’s clear how formulaic making movies is and how any reasonably intelligent individual can come up with a plot following the usual conventions. And I now have additional evidence that you really do want to become a screenwriter and have updated my prior accordingly.

                Incidentally, it’s not just you. There is allegedly a video shot by a camera crew walking down Hollywood Boulevard asking random goer-by’s how their screenplay is coming (first question) and no non-plus responses, just comments such as “Nearly done”, “Still working on it”, etc. Can’t find it on youtube so perhaps it is apocryphal (though I first heard about it 30 years ago so maybe it does exist and it wasn’t ever posted–any readers out there have any knowledge?)

          • Realist Writer says:

            Three Ways of Handling Dialogue:

            1) Don’t write out word by word but instead write out what is being said. “The university professor slammed his fist onto the table, cursing the grant committee. He then turned to his lab assistants and asked them to analyze the dataset and find a statistically significant result to appease his superiors before dinner time.” This dialogue isn’t exactly appealing but it *works* (conveying information as to the professor’s mental state, moving the plot along, etc.), and that’s probably good enough for you. You can’t really follow this approach for (spoken) film, but for written fiction, why not?

            2) Find dialogue from public domain works and edit them to match your needs. Make sure to cite where you’re grabbing the public domain works from…to avoid charges of literary plagiarism. Basically, I’m envisioning a more ethical version of that spy novel that you blogged about before. Instead of claiming to be an original author, make it clear that you are merely adapting what other people have written…and that since you’re copying and pasting from public domain works, you’re not engaging in copyright infringement.

            3) Treat the fictional story as a research paper, only one where you’re not constrained by petty things like facts. The fictional work can have footnotes referring to citations to non-existent works, pretty visualizations of fake data, anecdotes and just-so stories about purported events, etc. Eventually, you may get the courage to come up with a fake person and attribute fake quotes to that fake person…and there’s your dialogue.

  2. Ted says:

    Unlike on Kaiser Fung’s blog, I’m glad to see the folks commenting on here aren’t on the fainting couch worried about upsetting Jeff Bezos for scraping IMDB.

  3. kjetil b halvorsen says:

    Reminds me of where somebody found that “… increasing the diversity of the most homogenous country in the sample (Bolivia) by 1 percentage point would raise its income per capita in the year 2000 CE by 41 percent … “. Doesn’t get more silly than that.

Leave a Reply