The intersection of entertainment and social science is an exciting space to work. Not only can we measure the effects of big Hollywood blockbusters, but we can also apply some of the latest research methods to study phenomena that would have been unmeasurable just ten years ago.

Social media has emerged as a key source of data to drive this research. Recently, filmmakers and marketers have employed social media to drive attention and visibility. Our study of Bully’s successful Change.org petition provides one potential model for the spread of information. On the other end of the filmmaking spectrum, we are interested in how social media strategy and engagement can affect giant Hollywood franchises. Some of our findings support a surprising new model of influence on social networks.

To set up an initial case study, we looked for films that shared similar characteristics. After some thought, we decided to compare The Hunger Games and The Girl with the Dragon Tattoo. Our motive for the comparison was based on a number of strong similarities. Both films were adapted from hugely successful literary franchises, both featured strong female lead characters, and both have dark storylines. Moving beyond these initial assumptions, other data points began to align as well. Both films garnered critical success with almost identical Rotten Tomatoes scores in the high 80s. The size of their budgets was comparable—The Girl with the Dragon Tattoo was slightly more expensive at a reported cost of $90 million than the $78 million of The Hunger Games.

Of course, outside of a laboratory setting, it’s almost impossible to engineer identical control and treatment categories, and there were important differences between the films. The less graphic Hunger Games received a PG-13 rating, allowing it to tap into younger teen markets, while The Girl with the Dragon Tattoo’s R may have limited the film’s reach. The timing of the release and number of screens may also have played a role—The Hunger Games opened on about 1,000 more screens than The Girl with the Dragon Tattoo.

Emerging research suggests that primarily targeting influencers may not be an effective marketing strategy.

Nonetheless the two films are similar enough that the difference in their box office performance is striking, especially when considering the magnitude of the difference. The Hunger Games was a runaway success, with a domestic gross of over $400 million. The Girl With the Dragon Tattoo, on the other hand, posted a more modest domestic gross of $100 million. When marketing costs were taken into account, the film actually lost money for its studio. Accounting for this $300 million difference is difficult to do with a great deal of precision, but there were some indications that these films had important differences in marketing strategies.

One factor that came up again and again in press accounts was The Hunger Games’ innovative social marketing campaign. Their “social media machine” gave fans access to exclusive content and helped create an immersive environment. This led us to believe that social media engagement may have played a role in the huge box office gap between the two films. Beyond journalistic descriptions, we wondered if there was data to support this hypothesis.

Although The Hunger Games used a variety of social media platforms including Tumblr and Facebook we focused on Twitter to compare online engagement for the two films. On Twitter alone, the differences were dramatic:

Moving beyond the surface, we wondered whether there was also a difference in the type of people that each film’s marketers were engaged with. This approach was inspired by the work of NYU network scientist Sinan Aral, whose latest article in Science explores the role of influence and susceptibility in social networks.

Popular figures like Malcolm Gladwell have argued that marketers should court a small group of highly influential figures to drive adoption. Aral’s experimental work challenges this approach, suggesting that marketers should primarily seek out susceptible users who are more likely to adopt a product or viewpoint. A key conclusion from his work is that, on average, it’s harder to influence a highly influential person and easier to influence a less influential person.

Following this insight, we looked for ways to measure influence and susceptibility in the social media campaigns of our two films. Because we didn’t have the same resolution of data as Aral, we thought about meaningful indicators of influence. Of these, Klout scores, a proprietary measure of online influence, provides a proxy. While Klout is far from perfect, its normalized 1-100 scale provides a standardized measure to work from.

In order to measure interactions in these two campaigns, we queried the tweets of their official accounts— @TheHunger Games and @DragonTattoo—looking for all interactions with other users. We then used the Klout API to return the scores of all these users, giving us a sense of the types of users each campaign targeted. The results reveal yet another significant difference between the two films:

*Here, “users” means those twitter handles that showed up in the feeds of the movies’ official accounts.

The graph and table above show the distribution of Klout scores of the unique users mentioned by each Twitter handle. What we see is that, in general, @TheHungerGames interacted with less influential, and perhaps more susceptible users than @DragonTattoo.  Strikingly, @TheHungerGames also directed a large part their attention towards users who fall below the mean Klout Score of 20, while @DragonTattoo concentrated on users with scores well above this mark. While we cannot be certain that these differences played a role in the film’s varying levels of success, the data suggests that each film pursued highly divergent social media marketing strategies.

Of course, targeting highly influential, visible users remains an important driver of attention, and one can see that The Hunger Games still reached out to users with high Klout scores. However, what emerging research suggests is that targeting only influencers may not be an effective strategy. Duncan Watts work, for instance, has tested “portfolio” approaches of targeting many less influential users rather than a few highly influential ones.

We recognize that it is difficult to point to one single cause for the astounding success of The Hunger Games. From a methodological point of view, cherry-picking successful examples and working backwards isn’t as effective as a randomized study. Nevertheless, we believe that this study provides evidence that supports recent work in the computational social sciences. It also shows that working with publicly available data in clever ways can produce interesting results. Combining innovative research frameworks with a deeper knowledge of the entertainment domain is a way to produce knowledge and meaning from rapidly growing mountains of data.

Images: Lionsgate, Columbia TriStar Marketing Group