Understanding the Census Bureau’s Vintage 2023 Population Estimates for Connecticut
In May of 2024, the U.S. Census Bureau’s Population Estimates Program (PEP) released their vintage 2023 population estimates for Connecticut’s towns. This estimates series includes the first estimates for 2023, as well as updated estimates for 2020 through 2022.
You can explore these estimates on our new population estimates dashboard, which includes estimates of Connecticut’s total population at the state, planning region (Connecticut’s county-equivalent geography for Census Bureau data), and town levels; components of population change (births, deaths, and domestic and international migration); and new housing units permitted annually in each town according to the Census Bureau’s Building Permit Survey.
However, there are some important caveats to keep in mind when interpreting Connecticut’s population estimates. In particular, the vintage 2023 estimates series incorporates an adjustment to correct for a known error in the vintage 2022 series, which results in the vintage 2023 series displaying a substantial drop in the 2020 population estimate followed by a rebound in 2021. This adjustment also impacted the PEP domestic migration estimates for the state and planning regions in 2020 and 2021.
For reasons that we will explain below, we suggest that caution be used when interpreting the following estimates from PEP vintage 2023 for Connecticut:
Population as of July 1st, 2020
Population change from April to July of 2020
Population change from July of 2020 to 2021
Domestic and total migration between April and July of 2020
Domestic and total migration between July of 2020 and 2021
This blog post will:
Review the error in PEP vintage 2022
Describe the adjustment to PEP vintage 2023
Provide data on how the adjustment impacted state and town population estimates
Provide data on how the adjustment impacted net domestic migration estimates
Provide guidance for data users
What Was the Error in the Vintage 2022 Estimates?
The vintage 2022 estimates series provided artificially high population estimates for certain Connecticut towns in 2021 and 2022 due to a pandemic-related error in the count of residents in group quarters facilities such as college dormitories. For the 2022 vintage, the Census Bureau did not have access to counts of the number of Connecticut residents who had left group quarters facilities between Census Day (April 1st, 2020) and the reference date for the annual population estimates (July 1st, 2020). They therefore developed Connecticut’s population estimates under the assumption that the group quarters resident population had remained constant over these three months. In reality, however, many group quarters facilities had experienced a steep pandemic-related decline in residents in 2020, followed by a rebound in their resident population by 2021. (It is worth noting that counts of residents in college student housing actually reflect the number of residents in the fall of the reference year, which are then applied to the Census Bureau’s July 1st, 2020, annual population estimate. This accounts for the fact that students are not typically living in college dormitories over the summer, but they should have been counted in their college dormitory residence on Census Day).
For the vintage 2022 estimates, the Census Bureau did not receive data on the decline in group quarters residents between April and July of 2020, but they did receive data on the steep rebound between 2020 and 2021. The vintage 2022 population estimates therefore incorporated this rebound in group quarters residents from 2020 to 2021 without incorporating the decrease in residents that had occurred between Census Day and the 2020 annual group quarters count. Connecticut’s Department of Public Health had estimated that this error resulted in an overcount of Connecticut’s total population by about 20,000 to 25,000 residents. See our blog post on this topic for details.
How Were the Vintage 2023 Estimates Adjusted to Correct for this Error?
The vintage 2023 estimates series improves the accuracy of Connecticut’s total population estimates for 2021, 2022, and 2023 by accounting for the steep pandemic-related drop in group quarters residents between Census Day and the 2020 annual group quarters count. This was possible because the 2023 Group Quarters Report submitted to the Census Bureau by Connecticut’s Department of Public Health included special counts of group quarters residents on April 1st, 2020, as well as the usual counts for July 1st, 2020. The prior year’s report had not included counts of residents in college dormitories or nursing homes as of April 1st, 2020, as the group quarters population was typically assumed to be stable between April and July of the decennial census year.
In the vintage 2023 estimates series, residents who left group quarters facilities during the summer of 2020 are subtracted from the population estimate for July of 2020. This results in a drop of about 20,000 residents (-0.5%) in Connecticut’s total annual population estimates for 2020, 2021, and 2022, relative to the population estimates for these years in the vintage 2022 series (see the graph to the right).
The table to the right shows the change in population estimates for July 1st, 2020, between vintages 2022 and 2023 for each of Connecticut’s towns. By default, the table is sorted from the most negative to the most positive percent change between vintages. Use the arrows on the chart to scroll through the data for all towns, or you can search for a specific town name in the search box.
The greatest negative percent change between vintages was seen in Mansfield, where the population dropped by nearly 30% (7,714 residents) due to the temporary departure of students from University of Connecticut dormitories after the onset of the pandemic in 2020. The second-largest negative percent change was in New Haven, the home of Yale University, where the estimated population on July 1st, 2020, dropped by 3% (4,202 residents) between vintages.
Population Changes Over Time within Vintage 2023
Looking over time within the vintage 2023 estimates series, Connecticut’s total population was estimated to decline by 28,326 residents (-0.8%) between April (Census Day) and July of 2020, followed by an increase of 26,105 residents (0.7%) between July of 2020 and 2021. According to the vintage 2023 group quarters estimates released by PEP, changes over time in the group quarters population estimate accounted for 77% of the total population decline between April and July of 2020 (group quarters change of -21,761), and 48% of the total population increase between July of 2020 and 2021 (group quarters change of +12,565).
However, many residents who temporarily left group quarters facilities in 2020 likely moved back into homes in Connecticut. Likewise, many students who had left Connecticut to study in other states also would have temporarily moved back into Connecticut in 2020. The Census Bureau unfortunately is not able to track the movement of residents between group quarters facilities and the household population. Thus, the apparent population decline in 2020 and rebound in 2021 are substantially impacted by the temporary pandemic-related relocation of group quarters residents, but the Census Bureau’s estimates probably do not account for the movement of many of these group quarters residents into the household population in 2020.
For this reason, we suggest that data users use caution when interpreting Connecticut’s total population change between April and July of 2020, and between 2020 and 2021, in the vintage 2023 population estimates series.
The table to the right shows the percent changes over time in the population in each of Connecticut’s towns within the vintage 2023 estimates series. By default, the table is sorted from the most negative to the most positive percent change from April to July of 2020, although you can click on any column header to re-sort the table by that column. Use the arrows on the chart to scroll through the data for all towns, or you can search for a specific town name in the search box.
As expected, the most negative percent population change from April to July of 2020 was seen in Mansfield, where the population estimate dropped by nearly 30% due to students’ departure from University of Connecticut dormitories. Mansfield’s population estimate then rebounded by 33.6% from 2020 to 2021 as students returned to campus housing. The second-largest negative percent population change from April to July of 2020 was in New Haven, where the population estimate dropped by 3.2% and then rebounded by 3.7% in 2021 due to students’ departure from and return to Yale University housing. Similar drops and rebounds in town population estimates were seen in New London, Middletown, and Danbury, all of which have college or university student housing (Connecticut College, Wesleyan University, and Western Connecticut State University, respectively).
To explore total population estimates for 2023 at the state, planning region (Connecticut’s county-equivalent geography for Census Bureau data), and town levels, as well as changes in these estimates over time and across vintages, visit our population estimates dashboard.
Impacts on Domestic Migration Estimates
In addition to publishing total population estimates, the Census Bureau’s PEP publishes the estimated components of change (births, deaths, and net domestic and international migration) that underly these population estimates at the state and county (or county-equivalent) level.
According to the Census Bureau’s technical documentation, changes in the group quarters population contribute to their estimates of net domestic migration. It is therefore unsurprising that the adjustment to 2020 group quarters counts for vintage 2023 also impacts the PEP domestic migration estimates for 2020 and 2021. The magnitude of this impact can be seen in the table to the right, which shows net domestic migration estimates for Connecticut from the PEP vintage 2021, 2022, and 2023 estimates series.
The estimate of Connecticut’s net domestic migration for 7/1/2020 – 7/1/2021 increases by 13,885 between vintage 2021 and vintage 2022 (from 5,134 to 19,019), reflecting the increase in group quarters residents across these years that was first incorporated in the vintage 2022 estimates. This increased domestic migration estimate for 2020 to 2021 is also present in the vintage 2023 estimates series, although in this series the net domestic migration estimate for 4/1/2020 – 7/1/2020 decreases by 15,872 relative to the vintage 2022 estimate (from -5,548 to -21,420), reflecting the subtraction of residents who temporarily left group quarters facilities during the summer of 2020.
Since the pattern of changes in domestic migration estimates across vintages seems primarily attributable to the changes in estimates of group quarters residents, and since it is unclear what percentage of the group quarters change truly reflects migration into or out of Connecticut versus within the state, we suggest that caution be used in interpreting the following estimates of domestic and total migration for Connecticut:
Domestic and total migration between April and July of 2020
Domestic and total migration between July of 2020 and 2021
Summary of Guidance for Data Users
For Connecticut’s total population estimates in 2021, 2022, and 2023, the vintage 2023 estimates should be more accurate than prior vintages. However, caution should be used when interpreting total population changes from April (Census Day) to July 2020 or between 2020 and 2021, since the drop in population estimates between April and July of 2020 and the subsequent increase from 2020 to 2021 are impacted by the temporary movement of residents out of and back into group quarters facilities during the pandemic.
Since changes in the group quarters population also impact the PEP domestic migration estimates, we also urge that data users exercise caution when interpreting estimates of net domestic and total migration for Connecticut between April and July of 2020 as well as between July of 2020 and 2021.
For More Information
Visit our population estimates dashboard to explore estimates of Connecticut’s total population at the state, planning region, and town levels; components of population change; and new housing units permitted annually in each town.
For details on the Census Bureau’s methodology for creating the vintage 2023 population estimates, see their methodology statement.
To learn more about the Census and resources provided by CTData, head to our Census Data portal, where you can learn about the different sources of population data on our Population Statistics Hub and access Census Tools and Resources for your census data project. Explore other data sets and analysis at data by topic and data projects. You can stay up-to-date on the latest data and tools by subscribing to our newsletter and following CTData on Facebook, Twitter, and LinkedIn.