Not So Representative Congress

For a long time, I’ve been working on a project analyzing the make up of the U.S House of Representatives.  I started with a basic assumption that most states use gerrymandering to skew the delegation that goes to D.C.  For example, the 2016 election in Indiana saw Republicans earn 54.28% of the votes and Democrats had 39.61% with various 3^rd party candidates making up the remaining 6.11%.  However, the delegation that went to D.C. was 7 Republicans (77.78%) and 2 Democrats (22.22%).  Clearly Indiana drew districts to maximize Republican voters while limiting Democratic voters.

If Indiana is skewed, what other states are and how badly?  What’s the net impact on the entire Congress?  Would it change the majority party?  How many states are skewed?  Is it just a handful or do the majority of states draw lines to favor one particular party?

If you want to jumps straight to the visualizations, visit my Tableau Public viz.

Before I get into the results, a few notes on methodology and other concerns.  The general rule was to find the correct mix of delegates that most closely matched the election results.  The basic formula: (Majority actual – Majority Ideal) + (Minority Ideal – Majority Actual) = Net gap.  In some cases the 3^rd party gap was also included.  The goal was to make the Net gap as small as possible.  I did make an exception for states with only 2 delegates since 25% seemed too low a threshold to meet.  For those states the bar was set at 37.50% (In years where a 2 rep state managed to split the vote, roughly 40% voted for the minority party)  

There were also some issues obtaining accurate voting results.  Most of the data is taken from officials results as reported by the FEC.  However, four states (Arkansas, Florida, Louisiana, Oklahoma) do not require totals to be reported when a candidate is running unopposed.  Where possible I dug into precinct level results to determine precincts that made up the congressional district and then used the total votes of either the Presidential or Gubernatorial race as the figure for that race.  In some cases, precinct results were unavailable or required more time/money than I had available.

The final issue is my treatment of 3^rd party candidates.  Since the focus was on the two major parties, I made the decision to lump all the other candidates into one pool.  Obviously, that misses the nuances of various third parties being generally conservative or liberal.  It also introduces the idea that enough 3^rd party candidates could get votes to impact the ideal mix without accounting for ideological differences.

On to the results.  One of the biggest things that came out of the data was the impact of Barack Obama.  In 2008, Democrats held 14 more seats than ideal, while Republicans were 10 seats short of ideal (3^rd party candidates made up the other 4 ideal seats).  Two year later, 3^rd party candidates should have held 8 seats, Republicans were +12, and Democrats were -4.  To be fair, the President’s party typically loses seats at the mid-term, but the flip is significant.  That was also the same year as the decennial census.  Republicans took major advantage of the census and reapportionment to lock in their gains. 

Prior to 2010, the number of states with delegations matching their ideal mix waffled between 22-24 states depending on the year.  In fact, only 10 states had a Republican tilt in 2008.  Two years later that number more than doubled to 21 states with a Republican tilt and it’s never dropped below 20 since then.  States that are neutral dropped to 14-16 each year.  The jump in states that are skewed Republican was so great that it effectively offset the single greatest skew from the data (California 2018 +11 Democrats).

Here in Indiana, the evidence is similarly clear.  In 2008, we sent 5 Dem and 4 Rep to Congress, matching the ideal mix.  In 2010 Republicans were +1 and then +2 in every election since.  The 2020 election saw the race in Indiana’s 5^th district become a little tighter for Republicans than they really liked.  So what did the legislature do after the census?  They redrew the districts, shoving Democratic leaning precincts into the same district as the urban core of Indianapolis and adding a few more Republican rural precincts to the 5^th district.  That way, Republicans didn’t have to worry about losing a seat in 2022 and beyond.

A few other interesting results from the data.  In both 2014 and 2016, more people voted for Democratic candidates in Michigan but the actual Congressional delegation had 9 Republicans and 5 Democrats.  California had the largest single skew as Democrats were +11 in 2018.  Texas had the largest Republican skew at +5 in 2018.  Republicans in 2016 own the largest gain from gerrymandering as they were +23 during the first two years of the Trump Administration.  Republicans also own the 2^nd largest gain of +20 in 2012.  Democrats had their biggest gain in 2008 at +14.

I also have to respect the home of gerrymandering, Massachusetts, for failing to elect a single Republican to Congress in more than 2 decades, despite roughly 15% of the state voting Republican each year.  I’m no fan of gerrymandering and would like to see it removed from practice, but Massachusetts has been at this game for a very long time.

I hope to have another post up soon looking at changing the balance in Congress.  The data on hand suggests that we do not have a good representative democracy.  So, how do we get better balance?  How can we get more states to move back to the neutral column and reduce the number of states that are skewed towards a particular party.

Take a look at the visualizations at the link above and let me know what you think.  Have a question you want answered or a visualization you want to see, let me know in the comments.