Abstract:

This data includes an entry for every member of Congress who served between January 1947 and February 2014. Entries include individuals’ personal identifying information such as their name, age, bio guide, and birthday. For each individual, there are also details on their party, chamber, term start, the state represented, and which Congress they served on. We used this data to understand trends and distrubtions of congressperson’s ages, by party, incumbency status and gender. Our results indicate that congress is, on average, older (average 49-57 years, depending on congress) than the US general population (average age 38 years), regardless of party, likely because the majority of congress is incumbents. We show that women in the house tend to be older than their male counterparts, but across the whole of congress after 1970s, there is little difference between the ages of men and women. We hope future work can address the new and different factors which may explain who is elected to congress, and why they are relatively old compared to the general US population.

Topic

The Relationship of Age, Gender, Party, and Incumbency Status within Congress

Background:

The data was used for an article researching the age of Congress members. The article, “Both Republicans And Democrats Have an Age Problem” can be found at https://fivethirtyeight.com/features/both-republicans-and-democrats-have-an-age-problem/; the data source is the fivethirtyeight R package. The authors were inspired to write this article because of a New York Times article that cited little to no statistical evidence regarding how the 2008 election didn’t encourage young people to run.

Congress Age Data (Congress_age)

Variable	Description (class)
congress	The number of Congress from least to greatest (integer)
chamber	The chamber of Congress the person is in (house; sentate) (character)
Bioguide	A unique identifying number (character)
firstname	Person’s first name (character)
middle name	Person’s middle name (of available) (character)
lastname	Person’s last name (character)
suffix	Person’s suffix (if available) (character)
birthday	Person’s date of birth (date)
state	The state represented (character)
party	Person’s political party (character)
incumbent	Person’s incumbent status (“TRUE” = incumbent (first term already served), “FALSE” = challenger (in first term)) (logical)
termstart	Start date of the term (date)
age	Age at the time of term start (number)

---

Hypothesis:

• The age of congress members is increasing over time.
• Incumbents are the majority and are the oldest members.
• The female and male members of Congress will have different average ages.

---

Graphs:

Congressperson Age: Incumbency and Party

This graph displays a summary of the main points made in the “Both Republicans And Democrats Have an Age Problem” article. The blue lines represent Democrat members while the red represent Republican members. The solid lines give the ages for incumbent members, while the dashed lines give ages for the non-incumbent (newly elected) members. Over time, Congress is getting older regardless of party or incumbency status. Starting in 1947, congresspeople start with an average age of 52 years, then average ages dip around 1981 to 49 years, then increase into the most modern period with an average around 57 years for 2013.

However, there is a visible difference in age between the incumbents and challengers. The age of incumbents trends much higher throughout all recorded election years. This can be partially explained by the fact that senate terms are 6 years, and house terms are 2 years - meaning that any incumbent in the senate would be at least 6 years older than their initial, non-incumbent run; and at least 2 years older for the house.

Focusing on the 113th Congress, we noticed the number of incumbents is much higher. This was not very surprising because we would expect that the experienced runners are already well-credited and have garnered much recognition as compared to new runners. As well, there are slightly more Republican incumbents and slightly more new Democratic runners, but the differences are minimal.

Again, we looked at the 113th Congress to understand the most current data we have. The bell curve can be explained by the pattern we see in the rest of our data: congress is dominated by incumbents, therefore they are the older and more experienced members. They are the peak of our bell curve because congress is made up of mostly experienced members. Thus, our younger members are new and not so experienced. Fewer people vote for the challenger, so we see fewer young members. By the time members get up into their mid-60s we see fewer of them running again (possibily because they plan to retire or persue something different). Since we are looking at one election period, we infer which that the youngest members are likely non-incumbents - it appears that Democrats and Republicans both have new runners around age 40.

Gender’s Effect on Age of Congresspeople

We use the gender package, which will allow us to infer the gender of congress people using the U.S. Social Security Administration, the U.S. Census Bureau (via IPUMS USA), and the North Atlantic Population Project data sets from 1932-2012. The package takes the first name and returns the proportion of people in these data sets who are male and female. From this, it infers if the person is male or female. Note that this package is, of course, limited since it uses a state-standardized male/female binary. We are using this data set to understand how the demographics of Congress have changed over time. More information is available on the documentation, https://www.rdocumentation.org/packages/gender/versions/0.5.4.

First, we need to add a column to our data frame which includes the inferred gender. We infer the gender in gender_tib, and then join the data frames together. Finally, we remove repeated rows (since the gender package only uses first names, it will add extra rows to congress_age for each repeated name). The added data in this case includes:

Variable	Description (class)
proportion_male	The proprotion of individuals in the genderdata data with the name who are labeled male, in the given year range (year_min to year_max) (double)
proportion_female	The proprotion of individuals in the genderdata data with the name who are labeled female, in the given year range (year_min to year_max) (double)
gender	The inferred gender (chatacter)
year_min	The lower bound of the year range to search name/gender combinations (default is 1932) (double)
year_max	The upper bound of the year range to search name/gender combinations (default is 2012) (double)

Note: For our purposes, we remove any congressperson whose gender cannot be infered.

First, it is helpful to consider how the proporitons of women and men in congress have changed over time, by party. We can see that historically, there were very few women in congress. Over time, representation of women has increased; between 1991 and 2013, the proportion of Democrats who are women increases more quickly compared to Republicans, though gender parity is far from achieved. This small number of women in congress, especially historically, could lead to interesting historical trends in congresswomen age (since relatively few women are in congress, a drift like event occurs where the small number of successful women lead to trends outside normal or expected distributions). We will address this more later.

Here, we use comparative boxplots to understand how chamber and gender affect the age of congresspeople. We can see that in the House, women tend to be slightly older than men, in terms of each quartile and the median. In the Senate, however, women and men are approximately the same ages during their service. Since many more men have served in congress, their ages span a much wider range and include more outliers. This pattern holds for the data subset containing only the 113th congress.

When we consider the interaction between gender and party, we see an interesting pattern. For the most recent congresses (which have more women relative to past congresses), we see that women and men within a party have similar ages. Historically, we see a different pattern. In our earliest data information (from approx. 1947-1973), the gender/party groups have distinctly different ages. In these data subsets (approx. 1947-1973), female Democrats tend to be younger than their male counterparts, while female Republicans tend to be older. Men, in contrast, have similar ages regardless of party. This could be explained by the low number of women in congress before the 1970s since as more women join congress, the average age and age structure becomes more similar to that of men. As more women enter Congress, their ages become more similar to men - indicating that gender is not a salient factor in congress’s (combined senate and house) age structure.