North Dakota’s Hard-to-Count Populations and Where We Find Them

The Census 2020 State Task Force has designated seven groups as “Unique Populations”

Census 2020 is a mere two months away and getting an accurate count is paramount for the next decade, as the count will impact communities’ finances, political power and be a host for future statistics on everything from poverty to calculating the number of births.

The task is simple: count every individual, once, in the right place. The problem is,it is not nearly as easy to do as one may think. In fact, it has proven difficult. It probably never has been, nor will it ever truly get any easier than it is now due to digital access.

The decennial census is something that has been done every 10 years. As such, a lot has been learned over time as to why some individuals are difficult to count and who they are.

The Census Bureau has identified four broad categories as theories as to why individuals have been identified as “Hard-to-Count” (HTC). These include:

Hard to Locate — Individuals without a typical residence such as those who move from place to place.
Hard to Contact — Once located, populations can be difficult to physically access (e.g., gated communities or populations experiencing homelessness).
Hard to Persuade — They may feel their information is none of our business. Or they may be breaking some rule, such as a limit on the number of individuals residing in the rented home. They may be in the country illegally.
Hard to Interview — These individuals may not share a common language with those conducting the count and have an inability to communicate or simply do not answer the door.
The characteristics of “who” is hard to count has a considerable overlap. A child under age five may be living in a household that moves often and has lower education achievement and lower household income. A person who does not trust government may be a non-English speaker or live in non-traditional housing and so on.

The Census Bureau has been studying the population that is hard t -count almost continually for decades in order to find better ways to count these populations. Studying how well various population were counted in past decennial censuses helps identify strategies that may help in Census 2020 and future censuses. 

In conjunction with Census 2010, demographic data from other government sources indicates 4.6 percent of children were missed. In North Dakota, this data indicates there were 846 households where children were identified as present in other government data sources (e.g. personal income tax returns) that were not counted in the census.

The Census Bureau also identified that 4.9 percent of American Indians on reservations were missed in Census 2010. 

Using the estimate of $19,100 lost per person missed over a 10-year period in calculating lost funding due to an undercount in 2010 means residents of North Dakota missed out on roughly $33.5 million in federal funds after Census 2010. American Indians on reservations likely received only 95.1 percent of the funding per person that has been allocated this decade from federal funding sources such as the Bureau of Indian Affairs.
Knowing What Areas are Harder to Count
After determining the “why” and “who” are HTC, the next step is where. The Census Bureau has gone to great lengths using the characteristic make-up of a given area to map expected locations that are going to be more difficult to count. By calculating a low response score (LRS), a calculated percentage of the population that is expected to not respond to the decennial census questionnaire, the Census Bureau can inform community leaders their areas of concern and better target their resources.

The bureau also has identified these individuals and grouped them by their characteristics. (This list is not exhaustive).
  • Young children under age five
  • Highly mobile persons
  • Racial and ethnic minorities  
  • Non-English speakers
  • Low-income persons
  • Those with a GED or less
  • Persons experiencing homelessness
  • Undocumented immigrants
  • Persons who distrust the government
  • Persons with mental or physical disabilities
  • Non-traditional housing
North Dakota’s Unique Populations
In preparing for Census 2020, the state worked to identify populations that were felt to be at significant risk of being undercounted or counted elsewhere.  

Seven populations were identified:
American Indians
New Americans
Frontier counties 
Bakken region oilfield workers
Retired snowbirds
Federal military
College students 
Essentially, the list breaks down into two groups. The first group includes populations posing a greater risk of being missed in the census count. The second group is comprised of those who will likely be counted, but may end up being counted in another state, when they should be counted in North Dakota.

The first group consists of American Indians, recent immigrants and individuals living in frontier counties (those with rural populations of six persons per square mile or less). These individuals, for a wide variety of reasons, have had lower response rates to the censuses in the past, and that's why they're HTC.

When individuals do not self-respond, they are more likely to be missed completely in the census. 
They also are more expensive to count, as often they must be counted in person by a Census Bureau hired enumerator at their residence.
The second group consists of retired snowbirds – those with more than one residence who may reside at another residence in a warmer climate on April 1, as well as federal military personnel and oilfield workers who are legal residents outside the state. 

This also can include students on a college campus. This includes foreign exchange students, who may only be in the United States to attend college.

Although they may have addresses in other states or countries, if they are in North Dakota six months and a day, they are counted here.

American Indians constitute the largest minority in the state. Knowing that nationally 4.9 percent of American Indians on reservations were not counted, this population is of concern to the state’s Complete Count Task Force.

In the past few years, the state has seen a significant increase in international in-migration, individuals arriving from outside the county. 

The most recent estimate in the American Community Survey indicates nearly 20,000 non-U.S. citizens are residents in the state. An additional 16,000 are naturalized residents. The decennial census counts these individuals regardless of citizenship. As such, recent immigrants are on our list of unique populations for which we are concerned.
The chart above shows from left to right the 52 counties in the state where individuals had the opportunity to self-respond in Census 2010. The smallest counties are on the left, and the largest are to the right. Rural county residents tend to respond to the Census at a lower rate than more populated counties, although the correlation of these factors are far from perfect. 

As an example, Pierce County in north-central North Dakota had the highest response rate of any county in the state in the 2010 census, much higher than Williams, which has a much higher population.  

That said, the less populated counties are over-represented in the lowest response rates category, while larger counties tend to be over-represented at the high end of response rates in Census 2010.

Similar reasoning was used for designating the state’s unique populations for Census 2020. Federal military, college students and Bakken oilfield workers all have large appropriations of legal residents outside the state, and therefore they are all on the list. 
The decennial census counts individuals where they spend most of their time, not where they vote or where the license plate is from.We know most individuals stationed at Air Force bases in North Dakota originate from outside the state. The same is true with many of the workers in the Bakken oilfield.

North Dakota has been a workplace destination for a long time. Data from the Census Bureau Local Employment Dynamics program indicates that as of 2017, 44,000 residents, or 14 percent of workers, in North Dakota came from other states.

Many of these individuals and their households only recently became residents of North Dakota. They also may not consider themselves residents here, but meet the census definition of "residents," as they are here a majority of the year. It is important that we capture as many of these populations as possible.

Counting every resident is paramount to a complete count for our state. This is the result the local complete count committees and state task force is striving to ensure occurs on Census Day.
Mapping Locations of High HTC Populations

Statistics on response rate are available for each area. For the 2000 and 2010 censuses, this means the rate for every census tract, city, county and state are readily available on the Census Bureau website in an electronic, downloadable format. In 2012, the Census Bureau put out a challenge as to who could come up with a predictive tool based upon housing, demographic and socioeconomic variables derived from the 2010 Census and five years of American Community Survey estimates. The winner was to develop a model that could predict the response rate of each block group using the characteristics of that area. The result of this competition was an online tool titled “Response Outreach Area Mapper” or ROAM that is available for everyone to use.
North Dakota in the Census Bureau’s Response Outreach Mapper
ROAM shows those geographical areas expected to be more difficult to count in Census 2020. The darker the areas, the more difficult the count is expected to be, as the characteristics of those areas indicate a lower self-response rate. These are shown in the darker blue colors. For North Dakota, two unfortunate aspects of ROAM are: 1) the areas not defined using tribal geography; and 2) areas not given a self-response opportunity in Census 2010 do not receive an expected non-response score at all. Sioux County, the northern portion of Standing Rock Reservation, receives no score whatsoever, nor do all of Turtle Mountain Reservation and most of Fort Berthold Reservation.

Census tracts that cover Spirit Lake and the portion of Lake Travers that is in North Dakota did receive a score. Based upon response rates dating prior to Census 2010, the Census Bureau chose to go directly to “in-person” door-to-door enumeration. As a result, self-response rates cannot be calculated as these populations never had the same chance to self-respond as other households did in the state or a majority of the nation. Secondly, the census tracts displayed are based upon county boundaries and are not contiguous with the reservation as a whole.

As a result, the portion of Fort Berthold that is in Dunn County is scored with Dunn County as a whole. This is because Dunn County has only one census tract. The same is true for the portion of Spirit Lake Reservation that is in Eddy County. This tract is scored with Eddy County rather than with the rest of Spirit Lake Reservation.

The result was that for a majority of tribal lands in North Dakota, a non-response score was not calculated. Any tribal areas that overlap into adjacent counties are obscured by the larger population of non-American Indians in that area. Looking at the self-response rates of tribal areas nationwide, we can tell that these areas will likely have a higher percentage of individuals who do not respond. Of the 141 tribal areas that were given the chance to self-respond in Census 2010, the typical response rate was about 52 percent.

The two census tracts that are entirely on the Spirit Lake Reservation do receive scores in ROAM. The portions of Spirit Lake that are in Benson County have low response scores of 33.2 and 30.1. These rates are about double that of the two census tracts in Benson County that are not part of the Spirit Lake Reservation. The two scores provided are the highest of any area in the state. These implications should be of concern to us, as it implies this key population could end up being undercounted yet again. As we stated earlier, 
nationally 4.9 percent of American Indians on reservations were not counted.
The LRS tool in ROAM displays a 32 percent expected non-response rate to Census 2020 in this area given the area’s characteristics (displayed below). Low personal income and younger median age among other factors are likely contributors to an area’s higher non-response score.
The map above shows the Census Tracts in the Grand Forks area. Note the darker spots where UND is located. Unfortunately for those areas, residents did not have the opportunity to self-respond in Census 2010 so no score has been calculated. Those areas are shown in gray. 
Upcoming Events:

March 1, 2020 - North Dakota Day Picnic - Mesa, Red Mountain Park
March 1-7, 2020 - Valley City Winter Show - Valley City Event Center
April 14-15, 2020 - League of Cities Spring Workshop, Bismarck, Ramkota
(Sources: Census 2020, The Low Response Score (LRS) a Metric to Locate, Predict and manage Hard-to-Survey Populations: North Dakota Census 2020 Outreach Plan, August 2018, Author’s notes)
Be Legendary – Be Counted!
Become A Community Champion!
Copyright © 2020 North Dakota Department of Commerce, All rights reserved.

Want to change how you receive these emails?
You can update your preferences or unsubscribe from this list.