2026-10-20

Topic - Age of Acting Oscar Winners

The Academy Awards, hereafter known by its colloquial name “The Oscars”, are a ceremony held annually by the Academy of Motion Picture Arts and Sciences where awards are given for achievements in film from the previous year. Some of the most popular and most talked about awards are those for the best lead performances in a film: Best Actor and Best Actress. These awards have been given out every year since the first ceremony in 1929. I was curious to see if I could find any patterns regarding the ages of the winners.

Methods and Process

My goal for this assignment was to see if there are significant trends in the ages of best leading performance Oscar winners. I focused on two things: how the ages of winners have changed over time, and how the ages of winners differ between actors and actresses.

The data I used is from OpenIntro (https://www.openintro.org/data/index.php?data=oscars), and includes the winners of the Best Actor and Best Actress Oscars every year between 1929 and 2018.

My process was a pretty straightforward analysis, involving graphs, key statistics, linear regressions, and a t-test.

The Data

## Number of observations: 184
## # A tibble: 2 × 4
##   award            n meanage medianage
##   <fct>        <int>   <dbl>     <dbl>
## 1 Best actor      92    43.8        42
## 2 Best actress    92    36.2        33

There are 184 observations (92 of each gender); each observation represents an actor or actress who won the award. “oscar_yr” is the year of the ceremony. “age” is the age of the winner at the time of the ceremony.

It is worth noting that ties occurred twice: in 1933 for Best Actor, a tie between Fredric March and Wallace Beery, and in 1969 for Best Actress, a tie between Katharine Hepburn and Barbra Streisand.

Age Over Time

Each dot represents a winner. The fitted lines show estimated trends. The shaded bands around the fitted lines show 95% confidence intervals for the mean. Just from looking at the graph, it seems the age of actresses has slightly increased over time, and the age of actors has stayed the same.

Finding and Testing a Trend

I used a linear regression to describe the relationship between the year of ceremony and the ages of winners. The format of a linear regression equation is as follows:

\[ \text{Age}_i = \beta_0 + \beta_1 \text{Year}_i + \varepsilon_i \]

\(\beta_0\) is the y intercept, \(\beta_1\) is the slope, and \(\varepsilon_i\) is the random error.

When looking at key statistics for our regression model, we have to keep in mind our null hypothesis and our alternative hypothesis:

\[ H_0: \beta_1 = 0 \quad \text{(There is no linear trend)} \] \[ H_A: \beta_1 \ne 0 \quad \text{(A linear trend exists)} \]

I used a significance level of \(\alpha = 0.05\). If \(p < 0.05\), there is evidence from the data of a linear trend. If \(p \geq 0.05\), there is not enough evidence to conclude that there is a linear trend.

Regression Results and Interpretation

## # A tibble: 2 × 6
##   award            n correlation slope pvalue rsquared
##   <fct>        <int>       <dbl> <dbl>  <dbl>    <dbl>
## 1 Best actor      92       0.089 0.029 0.4       0.008
## 2 Best actress    92       0.241 0.109 0.0209    0.058

Best Actor

  • Correlation: 0.089 (very very weak positive)
  • p value: 0.4 (insignificant)
  • Slope: \(b_1\) = 0.0294
  • R squared = 0.00788 (less than 1% of variation accounted for by linear trend)

The estimated trend is an increase in age of 0.29 years per decade. But, since our p-value is greater than 0.05, I don’t reject \(H_0\). There is not enough evidence to suggest a linear trend in the ages of Best Actor winners.

Best Actress

  • Correlation: 0.241 (weak positive)
  • p value: 0.0209 (significant)
  • Slope: \(b_1\) = 0.109
  • R squared = 0.0579 (around 5% of variation accounted for by linear trend)

The estimated trend is an increase in age of 1.09 years per decade. Since our p-value is less than 0.05, I reject \(H_0\). There is statistically significant evidence that a positive linear trend exists in the ages of Best Actress winners.

Confidence Intervals for Linear Models

##                     2.5 %       97.5 %
## (Intercept) -150.78694242 122.26302356
## oscar_yr      -0.03972676   0.09861522
##                   2.5 %    97.5 %
## (Intercept) -360.703337 2.9180903
## oscar_yr       0.016894 0.2010883

Ages of Winners by Category

Best Actor

  • Median age: 42
  • Mean age: 43.8
  • Middle 50%: 38 - 48.2
  • IQR: 10.2

Best Actress

  • Median age: 33
  • Mean age: 36.2
  • Middle 50%: 28 - 41
  • IQR: 13

Outliers

## # A tibble: 8 × 5
## # Groups:   award [2]
##   award        name              movie              oscar_yr   age
##   <fct>        <chr>             <chr>                 <int> <int>
## 1 Best actress Jessica Tandy     Driving Miss Daisy     1990    80
## 2 Best actor   Henry Fonda       On Golden Pond         1982    76
## 3 Best actress Katharine Hepburn On Golden Pond         1982    74
## 4 Best actress Marie Dressler    Min and Bill           1932    63
## 5 Best actress Meryl Streep      The Iron Lady          2012    62
## 6 Best actress Katharine Hepburn The Lion in Winter     1969    61
## 7 Best actress Geraldine Page    Trip to Bountiful      1986    61
## 8 Best actress Helen Mirren      The Queen              2007    61

The upper fence for Best Actress is 60.5. The upper fence for Best Actor is 63.5.

Comparing and Testing the Means

\(\mu_A\) is the population mean age for Best Actor winners.

\(\mu_B\) is the population mean age for Best Actress winners.

\[ H_0: \mu_A - \mu_B = 0 \]

\[ H_A: \mu_A - \mu_B \ne 0 \]

The estimated difference in population means is

\[ \widehat{\Delta} = \bar{x}_A - \bar{x}_B \]

I will use a t-test to evaluate the difference. Because I can’t be sure if the two populations have similar variances, I will use a Welch’s two sample t-test.

Mean Comparison T-test Results and Interpretation

## Mean age difference (actor - actress): 7.6 years
## 95% confidence interval: 4.55 to 10.65 years
## p-value: 2.072e-06

Running a Welch’s t-test in R tells me that the difference in mean winning age is 7.6 years and the p value is 0.000002072 (statistically significant). So, I can reject the null hypothesis of equal population means. 95% confidence interval of 4.55 - 10.65 means I can be 95% confident that the the difference in estimated population means is between 4.55 and 10.65 years.

Interactive Plot

Hover over a point for information about the winner it represents.

R Code

Here is some of the code I used to look at trends:

actor <- filter(actingoscar, award == "Best actor")
actress <- filter(actingoscar, award == "Best actress")

actormodel <- lm(age ~ oscar_yr, data = actor)
actressmodel <- lm(age ~ oscar_yr, data = actress)

summary(actor)
summary(actress)
confint(actormodel)
confint(actressmodel)

t.test(age ~ award, data = actingoscar)

Conclusion

I found that there is statistically significant evidence that could indicate the following trends:

  1. There is a positive linear trend in the ages of Best Actress winners over time, with the age of the winners increasing as time goes on.

  2. On average, the winners for Best Actor are older than the winners for Best Actress.

It is important to note that this is not the whole story - just because there is statistically significant evidence pointing to certain trends does not necessarily mean I can make accurate predictions. Only a small portion of the variation in the ages of the winners can be explained by my simple analysis.