Every decennial census changes how data are collected, measured, or defined. This isn’t surprising given the technological and social advancements that can occur in a decade. For example, the 2010 Census occurred the same year as Apple introduced the iPad!
The Census Bureau’s constitutional obligation is to identify where and how many people live in specific geographic areas; however, other aspects of the census like race/ethnicity and income help determine a wide range of federal, state, and local government budgets and other priorities. The demographic data also helps ensure that legislative districts are fairly drawn and don’t disenfranchise specific groups of people.
There are many changes to the 2020 Census that can make understanding population changes difficult. Some of these changes have a measurable impact on census population estimates, while other types of changes are unmeasurable. A measurable impact means that there is data or quality metrics that we can compare to understand and point to potential or actual impacts on the census. Some of the changes this year are not measurable or quantifiable, like the impact of the pandemic.
To further complicate census changes, some are consistent challenges that occur every ten years, while others are specific to the 2020 Census. Consistent challenges are ones that the census faces every decade, such as counting hard-to-count populations. These vary from specific 2020 Census challenges—new challenges like the impact of internet responses. We’ll classify each of the changes into these categories to help explain the impact on understanding the 2020 data and how D3 is starting to think about using the new enumeration data for the state, region, and city of Detroit.
Measurable and Consistent Challenges
This category is the easiest to understand because these changes happen every ten years so processes are already defined by organizations to analyze the changes and their impacts. The main measurable and consistent change in the 2020 Census is adapting geographic boundaries (block, block group, tract, etc.) to new population estimates.
Each census results in new geographic boundaries that impact our analyses over time, especially at the tract level because tracts have to maintain populations between 2,500-8,000 residents. This can impact our ability to analyze data around neighborhoods since tracts can be combined or split depending on population changes. Since this is a consistent and mostly measurable change, D3 has a process in place to analyze the changes. We’re going to release a statewide map soon that shows the new census tracts and whether they resulted from a tract split or combination of two tracts. We also analyze the impacts of these changes in the Metro Detroit region.
Measurable and 2020 Challenges
Similar to the previous section, these changes to the 2020 Census have data that can be used to measure their impact to some degree. However, they’re unique to the 2020 Census and require more thoughtful analysis as the data is released to ensure the appropriate nuances are understood to provide the right context to the data and impacts. The two major changes to the 2020 Census that are in this category are the shift to the online response option and the changes to the race/ethnicity question.
The 2020 census was the first time respondents had the option of responding via the internet. The transition to an internet response option was planned, but the potential impact on data collection was not clear moving into the census data collection. Years ago D3 started exploring the potential impacts, especially in Detroit where internet access is challenging for many community members. As more data from the Census Bureau is released, we will spend time looking at response rates, hard-to-count populations, and other information to provide more context around how the internet response might have impacted different communities. The pandemic might have had an even greater impact on this than expected since many community events, where residents might have been able to access the internet and fill out their census, were canceled.
Another measurable change that is unique in 2020 is a change to how the race/ethnicity questions were presented and coded. Given these changes, comparisons between the 2020 Census and 2010 Census race/ethnicity data should be made with caution and consideration for the improvements made to the Hispanic origin and race questions. While these changes are measurable, they also require a deeper analysis and developing new ways to compare and analyze data over time. This takes more time to develop appropriate context and we’re working on understanding how the changes impact the Metro Detroit region. It’s important to not take the race/ethnicity data at face value though, as detailed in this NPR article. The Census Bureau has changed race and ethnicity data collection before to ensure more detailed data collection about US residents. For example, in 2000, the Census expanded upon the question of how people identified as Hispanic/Latino, which required new analysis. Unlike some of the unpredictable impacts on the Census, this one should be easy to quantify and analyze when the appropriate methods are fully identified.
Unmeasurable and Consistent Challenges
Unmeasurable impacts are those that can’t really be truly measured. While there are models and imputations for understanding hard-to-count populations, it is truly impossible to know who the 2020 Census didn’t count, where they’re located, and what their demographics are. We examined some of the challenges around hard-to-count populations in our previous Census blog series. While this is a normal challenge for the Census and there are methods within the Bureau for better understanding the impacts of missing populations, it’s impossible to know exactly who is missing. We continue to work within the Detroit community to emphasize the importance of the decennial Census and ensure in the future the hard-to-count populations within our community are properly counted.
Unmeasurable and New 2020 Challenges
These are the biggest wild cards. From privacy algorithms to the pandemic to hyper politicization, these changes are unprecedented and unquantifiable. While we acknowledge the challenges that these sweeping issues have on understanding the census results, it’s impossible to truly measure their impacts and the uncertainty is going to become an accepted aspect of this new dataset.
The Census Bureau has used various methods to maintain the privacy of citizens filling out the census over time. In the 2020 Census, the bureau utilized differential privacy for the first time. For more about how differential privacy works, check out this video. The balance of public utility and privacy has been a hot topic leading up to the Census data release. The Urban Institute does a great job of outlining the issues related to this debate. D3 provided public comments to the Census Bureau about usability at more local geographies as early demonstration files made many of our use cases for census data impossible. It’s too soon to know for sure what impact this new privacy measure will have on some of the smaller geographies like block groups; however, D3 will be doing extensive research and analysis to ensure that our data is accurate and that we are confident with the information we provide our community.
One major concern with differential privacy is that running the algorithm 25 times, there will be 25 different estimates created. That means we can’t know with certainty exactly what the demographics are in any reported geography, but that has been the case for the census over time due to other privacy mechanisms like data swapping. However, one way to start to understand the impact of differential privacy will be to compare the 2020 Census data to 2020 Population Estimates for the City of Detroit and surrounding communities. D3 hopes to conduct this analysis over the coming months.
Another unique 2020 challenge is the politicization of the Census questionnaire itself with concerns about the citizenship question, court cases about pandemic responses, and more. These impacts are also unquantifiable because it’s impossible to know exactly why someone didn’t complete their census. However, it is highly likely that these political issues had impacts on census counts.
Possibly the biggest challenge and unknown to the Census is the impact of the pandemic on data collection and Census Bureau operations. The impacts on other datasets like the Current Population Survey and American Community Survey have been documented and commented on. The Current Population Survey had significant declines in response rates in 2020 and the Census Bureau recently announced that responses to the ACS were so low that they would not be able to publish one-year estimates. Over time, some of the impacts might be analyzed and explained. For example, understanding lost productivity of offices during mandatory work from home periods can help us understand where the Census Bureau lost productivity. There could also be some additional analysis conducted to understand where people were undercounted, such as students, due to moving home unexpectedly. However, the majority of the impacts of the pandemic on the Census will remain unknown.
While the 2020 Census has many changes and challenges with known and unknowable impacts, it is still the best understanding of who and where people are living in the United States. It will take some time to unpack all the nuances of the new privacy measures and impacts of different changes to the data collection process, but this is what the government will use to direct funds and redistrict, so understanding how Detroit residents and our communities are represented in the data is important to the future of our city. While the many different changes might make it impossible to understand why something changed, it is still good to know that it changed. For example, while a population decline is observed and validated, it might not be possible to know if it’s because of a pandemic-driven reason, differential privacy, or a problem with enumeration. D3 will continue to dive into the nuances of the 2020 Census and we invite you to reach out with any questions or findings of your own!