By Marshall Flores
Behind the veneer of glitz and progressive utopia of Hollywood, there are a few hard, sobering truths. One of these is that inevitably, every actress encounters the stereotype Goldie Hawn once famously described in The First Wives Club: “There are only three ages for women in Hollywood – babe, District Attorney, and Driving Miss Daisy.”
The primary reason for this is the realities of the film business – studios are primarily in the business of targeting specific audiences with their films. And although females now comprise the majority of the film-going population, males (especially younger males) still account for the bulk (55%) of theatrical attendance. Speaking as a relatively enlightened but beleaguered member of that subset of humanity, it’s a truth universally acknowledged that men like pretty young things. As such, the casting of female roles is generally centered on maximizing the relatability/fantasy factor with men. Or, as the late actress/writer Adrienne Shelly once bluntly recounted during a meeting with an agent: “What is important is that they think you are fuckable.”
Are the Academy Awards, which are held up not only as symbolic of the best cinema has to offer, but also the best Hollywood can offer, affected by the Tinseltown predisposition towards younger actresses? And if so, to what degree? Statistics can provide a means in investigating this subject.
There are many potential facets to explore regarding age and the Oscars – today’s episode of Statsgasm will focus on a specific aspect: age and Best Actress. Specifically, we’ll look at three things: age distributions of Best Actress nominees, how the average age of Best Actress nominees/winners has trended over time, and if age is a useful variable in predicting Best Actress winners
To start, let’s visualize the ages of every woman nominated for Best Actress (421 nominees in total) in the entire 85-year history of the Oscars (and the ages we’ll be using will the the ages of the nominees at the time of their respective Oscar ceremony). The best way to do this? A histogram, of course!
Not surprisingly, the distribution is right-skewed – it appears most Best Actress nominees are younger than 40. The positive skewness holds if we decide to divide the distribution into non-winners and winners (non-winners will be the left histogram, winners the right).
So okay, the data confirms what we already know: the male-dominated AMPAS in general prefers to nominate and award younger actresses than older ones. Now let’s see if we can analyze the degree of this preference by tracking the average age of Best Actress nominees, winners, and non-winners over time.
Is this line chart what you expected? We seem to have every possible outcome in this chart: moments in Oscar history where the average Best Actress winner is roughly the same age as the non-winners, eras where the winners were clearly younger, and a couple points in time where the winners were definitely older. The 80’s, which featured three of the oldest Best Actress winners in history (Katherine Hepburn, Geraldine Page, and Jessica Tandy), seemed to be the heyday for veteran actresses over 40 to have their moments of Oscar glory. Meanwhile, it appears AMPAS voters were most willing to award young in the 1930’s and 1940’s.
There are a few tests one can run to verify if the age difference between winners and non-winners is (statistically) significant, e.g. a student T-test. Over the entire 85 year history of Oscar, the tests indicate that there isn’t a significant difference between the two groups. But over certain periods like the ones mentioned in the preceding paragraph and post-1990? Yes, the difference is significant.
But let’s take a closer look at the chart. If we exclude the fist couple of decades of Oscar history and the 80’s, the average age of a Best Actress winner hovers around 36 – in fact, the average age of all 85 Best Actress winners is 36. Meanwhile, the average age of a Best Actress non-winner (and nominee in general) has steadily risen, from 34 in the first decade of the Oscars to its current value of 40+. This appears to be an indication that AMPAS voters have gradually become more willing to nominate older actresses over time. And it appears this trend will continue this year; barring a surprise nomination from Adele Exarchopoulos or Brie Larson, the six consensus contenders for the 5 Best Actress slots (Amy Adams, Cate Blanchett, Sandra Bullock, Judi Dench, Meryl Streep, Emma Thompson) are all at least 39 years old.
Finally, can we use age as a predictor in determining who will win Best Actress? As I’ve indicated earlier, if we look at the entire history of Oscar, there isn’t a statistically significant difference in the average ages of Best Actress winners and non-winners. If we decided to throw age into a logistic regression model anyways using the full history, we would obtain the following:
The band indicates the 95% confidence interval of predicted Best Actress win probability given a nominee’s age. It is downward sloping, but it is also hourglass-shaped (i.e. wider at the ends than the middle), and the band is widest once we get past age 70. This indicates a muddled relationship between age and win probability. Meanwhile, the dark blue line in the middle of the band indicates the predicted average probability of a win – at no point does this line exceed 0.5, so age alone is not enough for a logistic regression model to be able to predict winners.
Yet, I did indicate that the age gap between winners and non-winners is significant from 1990-onwards. Lets re-run the regression using just data from that period and see what comes up:
The band is still downward sloping, but it’s no longer hourglass-shaped. There’s a wide variance in predicted win probability from 10-20 years old, but the width of the predictive band does stay fairly uniform from 30-up. Again, though, the mean predicted win probability never exceeds 0.5, so even though the negative relationship between age and Best Actress win probability is far clearer post-1990, age is *still* a weak predictor.
As far as Awards Daily’s prediction model for Best Actress is concerned, age is indeed one of four predictors; each additional year of age is estimated to decrease the likelihood of winning by about 5 percent. But this effect is absolutely dwarfed by the predictive influences of the other the three variables in the model: the Globe, the SAG, and the BAFTA. And yes, I did just reveal AD’s Best Actress model in it’s entirety. 🙂
That concludes today’s exploration into age and the Oscars. With the home stretch of this year’s Oscar season about to begin, I will definitely be preoccupied with running AD’s prediction models. But I will certainly revisit this provocative topic at a later date! Signing off from the Nostromo for now.