Votes, Voters, and Populations, Part 2: What is Normal Voter Registration? (art. 18)
Is there an 'expected' and 'proper' level of voter registration for Nevada?
On 18 March 2024, lawyers representing the Republican National Committee (RNC), the Nevada Republican Party, and Nevada resident Scott Johnston filed a complaint in U.S. District Court against the Nevada Secretary of State and five Nevada County Clerks for violations of the National Voting Rights Act (NVRA) by failing to maintain proper voting rolls (Case 2:24-cv-00518-CDS-MDC). This submittal was the completion of the notice of filing documents submitted last December. This filing (available wity responses from democracydocket.com) provides a clearer picture of the data and the arguments used.
That also means I can play a home version of the lawsuit here at “The Rurals of Nevada.”
Let me state again that I think there are some significant problems with Nevada’s voter registration practices and even more with its transparency on the matter. However, I am not sure this lawsuit will succeed beyond just highlighting the issue. While this might have indeed been the goal, I feel the lawsuit is an expensive means of publicity, especially considering some of the poor statistical arguments being made, which could potentially undermine the real issues being raised.
These statistical issues come in two forms. The first deals with trying to establish the actual percentage of registered voters in any given county—or a state, for that matter. I spent the last article looking at the various Census Bureau datasets that could provide some baseline measurements. The best dataset, the American Community Survey 5-Year Population Estimates (ACS 5-Year), does cover both the states and the counties with a large enough sample size, but lags real population significantly and is an estimate. These factors make it almost impossible to accurately estimate the number of potential voters, which is rather vital in the lawsuit.
The second issue, and in my view, the main deficiency, comes from trying to establish some ‘natural’ voter registration level. I understand the argument that a variance from some normative level can indicate a problem. The question is, what is that normative level supposed to be? The March complaint draws on the Current Population Survey (CPS) Voting Supplements to establish national and Nevada average levels. I explored some of the complaints about the CPS itself in my last article.
In this article, I want to explore the second question a bit more as well as begin examining the other major dataset in play, the Voter Registration Statistics maintained by the Nevada Secretary of State (NVSOS) based on information provided by the county clerks. It is this data which is technically being challenged as inaccurate. Ironically, it is the only dataset which is timely to the lawsuit. As many of critics correctly point out, the current registration numbers are being compared to datasets more than an year out of date. But can the voter registration numbers still tell us something?
One final caveat before we begin: I am not a lawyer. I am interested in the statistical and demographic arguments advanced rather than the legal arguments. Much of the current back and forth between the parties concerns precedents, standing, and other legal issues. I am just interested in what the demographic points raised may be telling us rather than the legal arguments.
Voters Active and Inactive
One of the first articles I did in this series was a few weeks after the 2022 Midterm Elections. The impetus for that article was a then-current argument that the rural counties encountered reduced turnout, which undermined the predicted “red wave.” In fact, Republican turnout was very good for a midterm election. It was the increasing number of non-partisan voters registered through the DMV who likely would not have voted who—surprise—did not cast a ballot.
It is the question of which voters would cast a ballot which is underlying the whole argument over voter registration numbers. Higher registration numbers indicate more potential votes. For those suspicious of election outcomes, the high registration numbers—spurred in part by the introduction of automatic voter registration in 2018—creates a huge potential for future fraud with other changes such as universal mail ballots. Recent kerfluffles such as the erroneous recording of mail ballots as ‘voted’ in the Presidential Preference Primary for people who did not vote just adds greater controversy and, for some, ‘proof’ the fix is in.
One of the major issues enflaming the whole debate about voter registration is the distinction between “active” and “inactive” voters.
I would argue that one of the major issues enflaming the whole debate about voter registration is the distinction between “active” and “inactive” voters. The terms are often used to distinguish between voters who regularly participate in elections and those who registration is in question due to lack of participation or unverified addresses. Although these terms are used commonly in discussions on issues such as the NVRA, there is no consistent, national definition of what they mean. They are defined at the state level—and the definitions can change over time.
In Nevada, discussion of inactive voters is in NRS 293.530, originally written in 1960 but updated numerous times since, with the latest iteration taking effect on 1 January 2024. As currently stated, the main trigger between the two statuses is the updating of voter registration information. Generally, if election notices (including mail ballots) are returned or a voter has not voted or updated information, county clerks can move them to the inactive voter roles for investigation. At that point, it may take up to two more federal election cycles (4 years in total) for the voter to be removed from the voter rolls unless they show up as having moved or are deceased. Bear in mind voters can still vote during this four-year ‘inactive’ period, which is one of the issues.
The confusion for election integrity debates comes from a commonplace idea that 'inactive voters’ should be immediately removed from voter rolls because they are ‘inactive'. But that is not how the law is written. The time before removal has resulted in somewhat enlarged voter rolls. And there is also the fact that automatic voter registration has effectively shift large number of voters into the active column—effectively ‘resetting the clock’ on voter registration.
When ALL registered voters in Nevada—both active and inactive—are counted, any semblance of rational statistics go out the window. Based on the 2022 ACS 5-Year Estimates, Nevada has a total registered voter percentage of 107.3% of the estimated population of citizens over 18. More surprising, 10 of the 17 counties have voter registration over 100%. Only three—Pershing (69.9%), Esmeralda (77.4%), and White Pine (82.4%)—have registration levels below 90%. Obviously, the inactive voter rolls are bloated enough to throw any rational analysis out the window.
But how bloated? Overall, 84% of all registered voters in Nevada are classified as active, as of February 2024; only 16% of voters are inactive. However, the variation between counties is astounding. Clark County, for instance, has 15.7% of its voters listed as inactive—slightly less than the state average. Esmeralda has 2.4% inactive—astounding low, but this is an older county with a shrinking population. Yet Nye (30.8%) and Mineral (30.5%) have incredibly high levels of inactive voters, and 5 other counties have levels over 23%.
But in addition to Clark, what about the other counties specified in the March complaint? That is where things get interesting. Douglas, Lyon, and Storey have inactive voter levels below 8.3%; only Esmeralda is lower. Carson City is at 10.5%, still lower than the state average. It appears these counties are doing a very good job cleaning the inactive voters from their rolls.
Why then are they the focus of the lawsuit? It appears counterintuitive, but there are two issues going on. The first, and most obvious, is that the comparison with the total population figures indicates the problem with the active voter numbers. The second I think is the fact that the high number of active voters itself is an issue. The argument would go like this: a low number of inactive voters indicates the clerks are not taking the obligation to move voters from active to inactive seriously, thereby forestalling any removal process.
There is also a minor discrepancy between the total number of voters and the individual totals of active plus inactive voters. Across the state, the difference is just over 6100 more voters listed in the total voters than the sum of active plus inactive, about 0.26% (that is, a quarter of a percent). Again, wide variation occurs between counties. Only 4 counties have the two numbers match—including Douglas and Storey, named in the lawsuit. Only one county—Elko—has more voters from the active and inactive tally than in the total voters (and admittedly by a rather large amount). These problems might be irritating but I think more the result of some bookkeeping errors which need to be addressed than malfeasance.
The problem is establishing how many voters Nevada should have, and whether they are active or not.
Anyone Know How Many Voters Nevada Has?
The short answer is no. That is the whole problem.
For those of you still reading after that headline, we can start getting into the weeds.
We can start by looking at the specific counties mentioned in the lawsuit. The complaint targets both the state of Nevada as a whole and five specific counties—Carson City (I’ll consider it a county for these purposes), Clark, Douglas, Lyon, and Storey—for having “suspiciously high rates of active voter registration.” The “most up-to-date count of registered active voters” was apparently the February 2024 active voter registration numbers from the Nevada Secretary of State, which were released on 1 March 2024. The voting-eligible population was taken from the 2022 American Community Survey (ACS), presumably the 5-Year Estimates given in Table DP05. At least, that combination of sources matches the numbers given in the March complaint.
Table 1, below, covers the numbers listed in the complaint, as well as a similar number for the United States with the percent eligible registered taken from the CPS Voting Supplement of November 2022. It is the closest parallel for the NVSOS voter registration numbers, with the obvious caveat it is about 15 months out of date.
I have also added two additional columns that I believe add some key context to be discussed a little later. First is the Median Age, which as a reminder is the point with half the population older and half the population younger. The second is the percent of the total population that are citizens over 18 (and therefore generally eligible to vote). I added these two columns because I suspect they contribute to the question of registration levels. Both of these statistics are taken from the 2022 ACS 5-Year as well (Table DP05—learn to love it!).
Remember that the ACS 5-Year Estimates have a lag; they are composites of the previous five years. So the 2022 ACS 5-Year includes the two years prior to the COVID pandemic as well as the more recent migration into Nevada since the end of the pandemic—which has been pronounced. Moreover, the 2022 Estimates are now over a year out of date compared to the Active Voter Registration list, which is updated monthly. It is this relationship which is likely to cause the lawsuit to fail.
But how would updated numbers—if they were available—impact the results? We can attempt to partially answer this question. The ACS also includes 1-Year Estimates, which can give a more accurate count for a given year without the lag of the 5-Year Estimates. The problem is that the 1-Year is only available for geographies with more than 65,000 people.1 For Nevada, that includes only the state as a whole and Washoe and Clark counties.
According to the 1-Year Estimates, Nevada has almost 77,000 more citizens over 18 than show up in the 5-Year Estimates—3.6% more in percentage terms. Clark County has over 60,000 more (3.9%), and Washoe has about 9700 more (2.7%). When these numbers are compared to the Active Voter Registration list, the percent of voters registered drops to below 90% for Nevada and Clark County. In the terms used in the March complaint, that means they no longer have “abnormally high” levels of voter registration
While we do not have precise numbers for the other counties in the complaint, we can use Washoe as a proxy here, in part because it is the major urban area to which the three other counties and Carson City are connected. If we assume a similar level of difference between the 5-Year and the 1-Year of 2.7%, Carson City also drops below 90% as well. We can plausibly eliminate Carson City and Clark County as likely having high registration levels—but not “suspiciously” high. Maybe.
However, Douglas, Lyon, and Storey counties still remain over 100% even when the 2022 ACS 1-Year estimates are taken into account. Even with the margin-of-error added, these three counties have extremely high registration levels. Explaining these counties’ high registration rates should be the priority.
Douglas, Lyon, and Storey Counties as well as Carson City share an interesting set of characteristics around age which can explain their high registration rates.
What is interesting is that the three counties share some important parallels. Intriguingly, Carson City shares the majority of the same characteristics. The four most significant similarities in the 2023 ACS 5-Year Estimates data in my opinion are:
They are all areas that have median ages well above the state average. Storey County at 55.7 years has the highest median age in Nevada while Douglas County at 53.5 years is the third highest. Carson City is older than the state median age as well.
All three have three-quarters or more their population who are Citizens over 18. Carson City is not far behind, with 73.7%. Nevada’s population is 69.3%, slightly lower than the national level of 71.9%.
According to Nevada’s Elders Count 2023 from the Nevada Aging and Disability Services Division (ADSD), Storey County has 33% of its population 65 or older, Douglas is close at 29%, and Lyon and Carson City over 20%. Nevada’s total 65+ population is 15%, while the United States’ is 17%.
These three counties continue to be retirement destination counties, where people voluntarily choose to retire either from other states or perhaps fleeing the increasing population in Washoe. I discussed the retirement destination issue in an earlier article.
In other words, Douglas, Lyon, and Storey counties (and Carson City) are united by age and a growing aged population—and that likely explains their higher registration levels.
Revenge of the Aging Rurals: Age and Voting
Long-term readers will be aware of a central theme of this series, the myth of the Aging Rurals. While there are many rural areas in Nevada which are not aging, there are also many which do match the myth—that is, with significantly older populations than the national average and an increasingly large percentage at or above senior citizen status.
The question of age plays an important role in this discussion because of the long-known relationship between age and propensity to vote. In 2017, the Census Bureau put together an infographic that illustrates this phenomenon, based on the CPS Voting Supplements going back to 1980.
The above chart clearly illustrates that Americans tend to vote more the older they are. Each age band has a higher percentage of its population voting consistently in elections than the next younger age band.2 Moreover, while this chart was made to argue that Millenials were the only group to increase voting numbers from 2012 to 2016, note that the 65+ age group is the only segment voting in higher numbers in 2022 than in 1980.
But our question is about registration, not voting. Does the same pattern hold, of people at increasing age groups being more likely to register? Well, the CPS Voting Supplements can provide this data at the national level. However, the citizenship data is only consistent after 2002—but this provides 20 years of data for our purposes. The graph below presents this information based on the same age groups as above.
Two items about this graph stand out. First, there does appear to be a clear correlation between age group and the percentage of that group registered to vote. Like voting, registration is closely correlated with age. Naturally, a large part of this is the result of once people register to vote they tend to maintain that registration. Over time, older people are likely to have registered.
The second item is more intriguing. While each age group shows fluctuation in that registrations go up during presidential years, note how each successive age group is flatter—that is, smaller variations from election year to election year. For instance, the 18–24 age group has a difference of 17.6% between the lowest year (2016, with 42.2% registered) and the highest year (2020, with 59.8% registered). For those citizens 65 and older eligible to vote, the difference between the lowest year (2010, with 74.9%) and the highest year (2012, with 79.4%) is only 4.5%.
The correlation between age and the tendency towards registration is strong and consistent.
To me, this pattern indicates that the correlation between age and the tendency to register is both strong and consistent. As people age, the number of instances they might be inclined to vote increase. Over time, the percentages creep higher. Of course, there also numerous non-political factors that likely contribute to this pattern as well. For example, people as they get older tend to move less, making it is easier to maintain constant registration. They might also become more active in various local campaigns or community organizations, tending to encourage registration as well.
Regardless of the reasons, the conclusion is that older counties should have higher registration levels. Moreover, since the correlation between age and voting is well-established, it would also follow that most older registered voters are active.
Of course, this relationship is not absolute. For instance, the two counties with low levels of active voters—Mineral and Nye—are among the oldest in Nevada. I suspect Mineral’s low level of active voters is the result of the large Native American population in the county, who have only consistently been participating in elections since 2016. Nye I do not have a good explanation for, other than the population in its most dynamic region—Pahrump—has long exhibited a strong independent attitude, which might extend to reluctance to register. But there might be other dynamics at play that are not apparent to me at the moment. Please treat that as speculation.
So it follows that generally speaking the older a county’s population is, the more likely that a large percentage of the population will be registered, will maintain that voter registration, and will vote—and hence large percentage of active voters.
There is one other contingent issue which also can shed light on the relationship between age and voting registration: how people register by parties. The Nevada Secretary of State’s office maintains a list of voter registration by individual parties and non-partisan (no party affiliation chosen), both by county and by age. This is vital information for the organization of primaries. The NVSOS data does not break the age and registration down at the county level, but for my purposes the state level data is intriguing.
The chart below shows the registration of voters in Nevada by age group and by party registration for February 2024, at the same time as the March complaint numbers. For my purposes, I have grouped all the registered third parties—the Independent American Party, the Liberal Party, and others—together into a single category. The result is a rough measure of where Nevada voters stand in terms of party registration by age.
The chief item of interest is the large number of Non-Partisan voters, shown in gray at the top of the chart. As of the February 2024 registration numbers shown here, Non-Partisan active voters were 33.5% of the electorate, making Non-Partisans the largest registration group in Nevada. This fact has generated a lot of recent media fascination on the rise of Non-Partisan voters as a political choice against the party system (see, for instance, this article from The Nevada Independent on the issue from last year). It is also one of the drivers of the Ballot Question # 3 on ranked-choice voting and open primaries on the ballot this November.
But remember that Non-Partisan is also the category into which people who are automatically registered to vote but do not choose a party are placed. Note that Non-Partisan registration numbers of the different age groups is the only registration that decreases consistently. While almost 55% of voters in the 18-24 age group are Non-Partisan, it is just over 25% for the 65+ age group.
So is it that younger voters are choosing to vote Non-Partisan—or that automatic voter registration merely capturing large numbers of people who would not normally register to vote and placing them in the Non-Partisan category? To answer this, I have added a black line to the chart based on the average voter registration percentages over the last 20 years (2002-2022) from the CPS Voting Supplements. The age groups do not match exactly, so I placed the registration numbers at mid-points of age groups.
What strikes me is how closely the curve of the national voter registration percentages by age follows the curve of the party registrations (excluding Non-Partisan) by age. I think this indicates the tendencies of people to maintain registration and to vote the older they get. I do not want to claim this is a ‘natural’ phenomenon, but it seems to follow a lot of other data about how people register and vote.
But it raises a real issue for the question of Non-Partisan registration as a political choice. Note that the average registration curves cuts through the Non-Partisan section. While these are not exact numbers, my suspicion is that the thin sliver of gray beneath that registration curve represents true Independent voters—that is, those who consciously choose to not register with a party. Also note that group remains relatively constant except for a slight bump in the 25-34 age range. These true independents I think are a relatively stable part of the electorate.
The gray above the average registration curve, however, I believe approximates voters who have been automatically registered since 2019, mainly through the DMV. In other words, these are voters who would not appear in voting rolls except for automatic voter registration. Remember the discussion at the beginning about active and inactive voters? I think that this is the real issue with the high registration numbers across Nevada: we are including a number of people who just are unlikely to vote historically who now because of voter registration are on the voter rolls.
And, in the next few years as these voters start migrating from the active to inactive voter lists, we are going to see another round of controversies over voter roll maintenance.
Conclusion
I wended my way through this confusing array of statistics on active voter numbers, populations, and age groups to get to one key point: the best predictor of registration levels is not comparison with other areas, but the age of the area concerned. Older areas will tend to have higher voter registration levels. Assuming that all populations—whether it be age groups, states, counties, or other groups—should have roughly equal registration and voter turnout is unrealistic. Basing lawsuits on some ideal of equal results—from both the political right and the political left—is one of the factors generating much of the election controversy.
The Census Bureau has argued something similar in a story about the November 2022 election released about three weeks ago. One of their conclusions is that voters over 65 are “overrepresented” (their term) in the November 2022 election (along with Non-Hispanic Whites with college degrees). The use of the term “representation” (both over- and under-) is problematic in my view because it assumes that voting is somehow not a choice but should always be equal and the goal of policy is to make it so. Is the fact that older people vote in larger numbers than younger voters merely the result of external restrictions on the young or preferences of the old—and how can policy account for that preference? I do not discount that voting barriers still exist and need to be fixed, but I am not convinced that we will ever get to equal representation given that people have a choice to vote or not.
I am actually not surprised that Douglas, Lyon, and Storey counties have a higher percentage of their population registered as active voters. And it might even be significantly higher than the state average. These three counties have populations that are of the correct age to be registered, active voters in much larger numbers than elsewhere. And they are experiencing significant growth since 2020 which is not captured in the underlying data fully due to statistical constraints, growth which is likely to be of this same active age group. I would add that Carson City can be included here as well, although the age and growth dynamic is more muted.
Clark County is much harder to explain. It is both younger than the state and growing rapidly, but has a large number of active voters. I do not think it is any secret the real focus of the March complaint is Clark County, and the other counties (all strongly Republican, by the way) are ‘collateral damage',’ for lack of a better term. I suspect the problem is that automatic voter registration is embracing large number of voters who might not have registered on their own. And in a few years, as these voters fail to maintain their registration, Clark County might have an explosion in inactive voters.
The surprising fact is that Nevada might actually know the number of active voters it has and is maintaining the list properly. The issue is that the number is effectively disconnected from any other measure that can be used to assess the accuracy of the list. The lack of clarity of active-versus-inactive voters caused by poor tracking and the recent introduction of automatic DMV registration is creating an appearance of problems that are statistical rather than actual. But there is also the issue of what the proper level of registration should be. I do not think that is as easy as it sounds.
There is also a ACS 1-Year Supplemental Estimate dataset which covers geographies down to 20,000 people. This would have covered all the counties in the lawsuit except for Storey County. The problem is that the tables in data.census.gov do not provide Citizenship status of the 18-and-Up population, which means that it would not be completely comparable. The data may be available through an API interface for microdata. Statistics are frustrating.
I suspect that the lower voting levels for the 65+ age cohort before about 1992 is the result of reduced life expectancy and more importantly old age mobility issues. Life expectancy now is almost 6 years more than in 1980 (see this chart from Statista.com). In either case, it is a minor variation for my main argument.
I think the Real Issue is the composition of the Non Partisans who registered thru DMV. Take Washoe for example if you were a CA Democrat, moved to NV & were registered as NP when you got your NV DL, would you feel like reregistering as a Dem until you got your Local Bearings? You would still vote for Dem Candidates like Biden & Rosen though - maybe against Amodei also.