I set out recently to solve what seems to be a fairly simple question that seems relevant to many current discussions of polarization, ideology, and electoral strategy: what is the geographic distribution of the farthest-left and farthest-right members of congress? Where do the most progressive and most conservative legislators in America tend to come from?
Methodology (Boring Stuff)
In 2004, UCLA researchers published a dataset of all representatives, Senators, and Presidents who served from the years of 1937–2002, along with each of their states and their DW-NOMINATE scores. DW-NOMINATE (which stands for “Dynamic Weighted NOMINAl Three-step Estimation”) is the most common measurement of legislator ideology used by political scientists, ranking a congressperson somewhere between 1 (the farthest right) and -1 (the farthest left). This gives us a dataset of 3,320 national American politicians over the last century to analyze.
Of these, I excluded Presidents and then chose the 200 legislators that were closest to each extreme , giving us a sample of the 400 most progressive and most conservative members of congress to serve in that period. I’ll refer to them as “hyper-progressives” and “arch-conservatives,” respectively, for the rest of the article. Looking only at the absolute extremes of the measure may distort my findings (for example, perhaps one state has few legislators who made it in the top 200 but quite a few more who made it in the next 200; ideologically extreme, but not the most extreme), but it will work well enough for my purposes. Additionally, the time period our data is from excludes the developments of the last 16 years, meaning our results won’t be exactly accurate for America today.
While discussing potential problems with my analysis, I should also note right from the start that the measure I am using has some limitations: DW-NOMINATE’s advantage is in “its capacity to order an enormous amount of information without doing undue damage to the way politicians actually behave,” but even though it greatly succeeds in compiling a large and complex set of data into an easy-to-digest format, its simplifications can lead to inaccuracies as well. Importantly, the score works in such a way so that its measures of conservatism and liberalism are relative and “have no absolute, fixed meaning.” In other words, how “conservative” a legislator is according to DW-NOMINATE is based on how their legislative record compares to other legislators serving at the same time as them, rather than any permanent set of conservative principles that each vote is compared to. This, combined with the ways that its tracking of legislative behavior may further distort ideological differences due to the increasing polarization of routine parliamentary procedures, means that DW-NOMINATE is going to get some individual cases wrong, and that our data may be a bit rough. Nonetheless, the fact that we’re operating with a sample of 400 legislators across a relatively long stretch of time means that we’ll hopefully capture enough data that the trends we uncover are useful even in spite of the limitations.
States Which Produce The Most Progressives and Conservatives
I began by simply counting the 200 most progressive and 200 most conservative legislators by the state that they were elected to serve from, and then mapping out the results. First, here is the map for hyper-progressives, with darker colors representing a higher number of legislators from the sample being elected from that state.
Much of the results here are expected, though some are more surpising. The most progressive members of congress are clustered in the Northeast, the Great Lakes, and the West Coast, all of which are still stronghold regions for the Democratic Party today. States with the largest populations tend to have more hyper-progressives, all else equal: Georgia has more hyper-progressives than Vermont even though Vermont is much more to the left, simply because Georgia’s size means it has had far more members of congress generally.
It may surprise some readers to see a number of traditionally red states in the Midwest and West represented on this map. This is a residual effect of a significant political trend of the late 19th and early 20th centuries which had long-lasting impacts, but has since faded. Left-wing populism used to have a fair amount of success in the American West and Midwest, where people elected progressive legislators serving in both parties and supported other movements such as the People’s Party and the Nonpartisan League.
Let’s move on to conservatives:
The geographic distribution of arch-conservatives is much more dispersed than the distribution of the hyper-progressives, for several reasons. First, large states have more room for intra-state political heterogeneity, allowing many conservatives to emerge there even as the states have a left-ward lean on aggregate. California is a classic example of this: though the state as a whole is very liberal, there have always been large pockets of conservatism in certain wealthier areas; combined with the state’s raw size, this allows for it to host one of the largest numbers of arch-conservative legislators of any state. Just as importantly, the map illustrates the shifting nature of American conservatism over the time period. Some states like Texas have always spawned adherents to hard-right politics, while others had unique moments: Arizona’s sunbelt conservative movement following Barry Goldwater’s meteoric rise in the 1960’s, the Midwest’s embrace of social conservatism in the 1980’s, and the unique maneuverings of Southern Dixiecrats shifting from the Democratic Party to the Republican Party over the second half of the 20th century.
An Aside About The South
The fact that South has had a relatively small number of arch-conservatives in our data despite widely being considered a conservative stronghold today deserves an explanation. Southern Democrats dominated Southern politics up until the mid-to-late 1960’s, when the civil rights movement would start a political realignment that would leave it more Republican than Democratic by the 1980’s. Political scientists who analyzed Southern Democrats from 1933–1950 found that their views were relatively close to non-Southern Democrats’ views on most issues, except that they were substantially less friendly to labor unions and radically less supportive of civil rights. DW-NOMINATE explicitly takes voting patterns on both economic and social dimensions into account, but this mixed record on issues often meant that even the most infamously racist and traditionalist Southern Democrats appeared as moderates on the measure for most of the century. Over time, Southern voters abandoned the traditional Southern Democrats for the more consistently conservative Republicans, at which point the only remaining Democrats in the South were consistently liberal.
Additionally, because DW-NOMINATE measures ideology by comparing the voting record of each individual legislator against that of others serving at the same time, Southern Democrats would appear more conservative in legislative sessions where civil rights and labor issues were more prevalent. This further distorts our data. For example, Sean Trende explains that the apparent right-ward shift among Southern Democrats from the 1930’s to the 1960’s may not entirely be an actual shift in ideology, just a shift in topics being debated and voted upon:
…we should consider that the House was populated with many more relatively liberal members from 1929 to 1939, and that the issue agenda changed from a conservative Republican one, which the Southern Democrats opposed, to a progressive Democratic one, which the Southern Democrats also opposed. Going from mostly opposing conservative Republican legislation to mostly opposing the legislation put forward by their progressive brethren would create an illusion of movement on the part of the Southern Democrats in terms of their absolute conservatism, when in reality they stayed the same. DW-NOMINATE simply gives us no particularly useful tools for discriminating between these competing explanations.
In summary, the tradition of Southern politicians who took right-wing stances on civil rights and labor unions but left-wing stances on many other issues explains why the South is relatively absent from our analysis: for the majority of our period of measurement, Southern politicians were not consistently right-wing across legislative issues, and thus DW-NOMINATE did not capture them as right-wing even when they espoused extremely far-right views on racial matters. Doing this analysis again with data from the last 16 years, thus capturing more modern Southern Republicans, would likely identify the South as the source of far more arch-conservatives, thus altering our entire analysis.
Comparison and Analysis
As I mentioned earlier, the results above could be skewed by any number of factors: DW-NOMINATE’s issues, my choice to choose a sample size on the tails of the ideological distribution, the timespan of our sample, and larger states being better represented simply because they have had more legislators total. For that reason, I decided to do a bit of a deeper dive into my results. First, I plotted out each state’s number of hyper-progressives and arch-conservatives against each other to look for trends.
This allows for a number of interesting observations. The majority of states (29 of 50) fall within the range of 5 or less legislators from each group, with only the most populous and the most ideologically-solid states breaking out. All states above the diagonal line have more arch-conservatives than hyper-progressives, while the opposite is true for those below the line. Among the red states, Arizona, Idaho, Indiana, Missouri, Ohio, and Texas stand out as having substantially more arch-conservatives. The story among blue states is even more interesting however. Michigan, Minnesota, and Pennsylvania make appearances near the top, but nearly half (99 of 200) of the left-wing legislators from our sample come from just three states: New York, California, and Illinois. Compare this to the arch-conservatives, where getting roughly that same percentage would involve aggregating their top eight states. This goes to show how much more geographically concentrated hyper-progressives are in heavily-populated states, and especially in major cities.
Illinois is a surprising outlier here because of how close it is to the even-split line despite having so many legislators. It has been home to the second most arch-conservatives (16) and third most hyper-progressives (18) of any state. The prevalence of the most progressive congressmen here makes enough sense: it’s host to a large, racially-diverse city with a Democratic political machine nationally famous for its strength. However, it appears that other areas of Illinois are a host to large numbers of conservatives as well. Indeed, the state’s abundance of strong ideologues on both sides has shown up in its mixed history in presidential elections: the state has only been solidly Democratic since 1992, but voted for Eisenhower twice, Nixon twice, Reagan twice, and George H.W. Bush once. It’s likely that the growth of the Chicago metro area outpacing the growth of the rest of the state is responsible for this shift.
I then decided to look at this another way: which states produced the most members of congress of each group on net? I mapped out each state’s net result by subtracting its number of arch-conservatives from its number of hyper-progressives. I figured this might help control for population somewhat, and serve as one way of showing which states each group actually dominates most.
According to this measure, the states which produce the most net hyper-progressives in congress are New York, California, Pennsylvania, and Minnesota, in that order. The states which produce the most net arch-conservatives in congress are Texas, Arizona, Indiana, Missouri, and Idaho, in that order. Interestingly, this analysis illustrates that there were only three states which never elected anyone on the outer boundaries of American political ideology during this time period: Nevada, New Mexico, and North Dakota.
However, this map has problems as well. Smaller states are still not properly represented, as a state who elected only one or two people in our sample cannot possibly break out of the +/- 1–3 categories here, regardless of how many hyper-progressives or arch-conservatives they produce proportional to their size. For these states with limited data, any visualization is probably going to misrepresent them by the very nature of having limited data. However, I tried to get around this by trying two more methods. First, here is the same chart as above with all states that had less than 5 total legislators in our full sample removed.
This is a useful way of identifying the states which, for one reason or another, have a definite tendency in one direction or the other towards electing people on the far wings of American political ideology.
Finally, I decided to try an entirely alternative method of visualizing the data, one which will overrepresent smaller states rather than underrepresent them. To do this, I looked at what percentage of each state’s total legislators in our sample that were hyper-progressives. Thus, a state which sent only one hyper-progressive would have a score of 100%, a state which sent one hyper-progressive but three arch-conservatives would have a score of 25%, and so on. Here’s that map:
Again, this over-represents the hyper-progressive and arch-conservative natures of smaller states, as a state which elected just one arch-conservative (like Kentucky) appears more conservative than a state which elected nine arch-conservatives and one hyper-progressive (like Idaho). It also makes some Republican states appear to be progressive (Montana and South Dakota) and some Democratic states appear to be conservative (Maryland and Delaware) because of a the election of just a few legislators several decades ago. But the point here is that low sample-size contributes to ideological strength in this map, unlike the previous map where it takes away from it. As such, this map is also inaccurate for smaller states, but somewhat useful in its own way.
I should start this conclusion with a reminder of all the limitations to the research here. This analysis relies on a measurement which can sometimes ideologically miscategorize legislators, our sample is limited to the most extreme ends of American ideology and doesn’t capture changes to American politics since 2002, and limited data from many smaller states makes them harder to properly analyze.
These problems are especially apparent in the South. Determining the proper orientation of the South here poses unique problems due to our limited time frame and the nature of how DW-NOMINATE captures ideology, and thus we may have additional problems measuring its actual placement. It seems likely that the South today is in fact a stronghold of American arch-conservatism, and that this analysis simply doesn’t capture that.
Even with all of this in mind, however, a general answer to our original question seems to emerge from this analysis. It appears that, from 1937–2002, New York, California, Pennsylvania, and Minnesota had the strongest tendencies towards electing highly progressive legislators, while Texas, Arizona, Indiana, Missouri, and Idaho had the strongest tendencies towards electing highly conservative legislators. Illinois holds the unique distinction of producing very high numbers of both. As far as electoral politics go, these results suggest the most left-wing progressives often emerged from the Northeast, Great Lakes, and West Coast, and are especially concentrated in large states and major cities. Though the issues discussed above mean that the modern South may be excluded, this analysis suggests that the most right-wing conservatives emerged from various parts of the American West, Southwest, and Midwest.