For deer killed by Michigan hunters during 2022, the official answers are 303,054 and 345,766. Reports submitted through mandatory state registration produce the first total. The Department of Natural Resources derives the second from its statewide survey of hunters. One measures reported harvest; the other estimates it. Both come from the same agency for that year. Their difference exceeds forty-two thousand deer. For 2023, that difference approached eighty thousand (274,282 compared with 353,914); for 2024, it exceeded eighty-one thousand (299,492 compared with 380,625). Registered totals remained below survey-based estimates throughout all three years: 12.4% in the first, 22.5% in the next and 21.3% in the last. This paper examines which figure deserves trust and why an honest reading requires keeping the measurements apart.
KEY FINDINGS
- Michigan's deer-harvest figures come from required registration and a statistical estimate based on a hunter survey. Across the three overlapping years (2022 to 2024), the registration total was 42,712 to 81,133 deer lower than the estimate from the survey.
- The difference was substantial, and its size changed. In 2022, registered harvest was 12.4% lower than survey-estimated harvest (303,054 compared with 345,766). In 2023, it was 22.5% lower (274,282 compared with 353,914). In 2024, it was 21.3% lower (299,492 compared with 380,625). A difference growing from 12% to 22% during mandatory reporting's second year cannot be treated as a constant adjustment. With only three seasons, we cannot tell whether it has stabilized.
- The methods also report results by different geographic units. Registration reports counts for 83 counties. The survey reports statewide and regional results, along with results for 91 deer management units; 60 of those units are county-sized. How the data are collected and the geographic units used to report them differ.
- For those reasons, we keep the measurements apart throughout our work and never calculate an average of the two. We select one preferred method for each use and never quietly add the other method's figures to it.
ITWO INSTRUMENTS, ONE HERD
Where Michigan's separate harvest totals come from
Most references to "the harvest number" mean a result from one particular measuring tool. Michigan uses two. With required harvest registration, hunters must report each deer they take. The state adds those submissions into counts for individual counties. With the survey-based harvest estimate, a sampled group of license holders answers questions about hunting success once the season is over. Statistical expansion turns those answers into estimates for the state and its regions. Each method is valid, but neither measures the herd itself. Registration records the deer hunters actually reported. Its coverage is complete in principle, although non-compliance and mistakes in reports can affect it. The survey estimates hunters' take without needing a report from every hunter. Sampling error and the assumptions used in its modeling still apply. Put an identical question to both tools, and getting separate answers should come as no surprise.
IITHE OVERLAP [2022, 2023, 2024]
How the measurements compared
Required harvest registration began recently in Michigan, leaving three seasons with results from both methods. Their difference is the focus of this paper. In 2022, the registered total was 303,054, while the survey placed statewide harvest at 345,766. Registration was 42,712 deer lower, a difference of 12.4%. The separation grew in 2023: 274,282 registered deer compared with 353,914 estimated deer, a shortfall of 79,632 deer, or 22.5%. For 2024, the figures were 299,492 and 380,625, leaving registration 81,133 deer lower, or 21.3%. Every one of the three years had a lower registration total. The first year's 12.4% difference became 22.5% in the second year of required reporting. The DNR's annual reports point in the same direction. For each year, they list fewer submitted harvest reports than deer estimated through the survey, with differences of 12%, 23% and 25%.
Swipe table to see every column
| Season | Registration | Survey estimate (statewide) | Gap (deer) | Gap % |
|---|---|---|---|---|
| 2022 | 303,054 | 345,766 | -42,712 | -12.4% |
| 2023 | 274,282 | 353,914 | -79,632 | -22.5% |
| 2024 | 299,492 | 380,625 | -81,133 | -21.3% |
For a given year, the agency publishes two official totals for deer harvested. The registered count was lower than the survey's estimate by 12.4% to 22.5%. Keeping them apart is the honest choice; averaging them is not.
IIIWHY BLENDING IS THE WRONG ANSWER
Why we never average the two harvest measurements
Taking the midpoint feels like an easy fix, but the data rule it out. An average makes sense when two measurements are noisy readings of an identical quantity. Here, the registration tally comes from deer that hunters reported, while the survey figure estimates harvest using a sample drawn from license buyers. With registration trailing the survey by 12.4% to 22.5%, averaging would mix a count and an estimate and carry that gap into the blended result. Their geographic units also differ: registration supplies counts across 83 counties; survey findings cover the entire state and 91 deer management units, with 60 county-sized. They also collect information differently: one tallies submitted reports, and the other estimates harvest using sampled license buyers. Combining them creates a third figure produced by neither method and conceals a disagreement worth seeing. A fixed-weight combination would also include a registration undercount that changed: about 12 percent became about 22 percent between the years 2022 and 2023. That would change what the combined figure meant from one year to the next. Where instruments differ substantially, a careful analysis chooses the one suited to the purpose, identifies it and shows the other beside it, without ever folding it in.
IVWHICH NUMBER, FOR WHICH QUESTION
How we choose the right measurement for the question
Choose the figure that answers your question. For a county-level look at a county's harvest or a comparison with nearby counties, mandatory registration provides figures reported by county. Across our work, we use a single registration total for every county in each year. For a long historical trend, the survey estimate offers a record that reaches farther back. Follow that record as a continuous series of its own, without joining it to the newer registration series. Both uses follow the same rule: do not label either measurement as the other or combine their figures. Hunters and analysts working with Michigan data should know which measurement they have in hand. Over these three seasons, the difference between the methods amounted to forty-three thousand to eighty-one thousand deer. Our position is that readers should treat the two harvest measurements separately.
NOTES AND SOURCES
- Data source: public deer harvest records from the Michigan Department of Natural Resources. We report two measurements: required harvest registration (mandatory_harvest_report) and the harvest estimate from the hunter survey conducted after the season (survey_estimate). We calculate registration totals by adding the 83 county rows in the state's public eLicense harvest summary. These counts reflect a July 21, 2026 snapshot and increase when late reports come in. The survey numbers are the statewide grand totals published by the DNR in its Deer Harvest Survey Reports for 2022, 2023 and 2024. They combine both sexes, cover every season and include deer harvested under DMA permits.
- Registration reports counts for 83 counties in Michigan. Survey findings are available for the whole state, management regions and 91 deer management units. Of those units, 60 share the boundaries of a single county. These county-sized results leave out DMA permits. Their combined share of the state estimate is about 75% to 77%. Information is collected differently for the two measures, and their geographic publication units differ. Here, each is shown as a measure of deer harvest across the state.
- Across the three shared years displayed here, registration was lower each time. Its shortfall was 12.4%, then 22.5%, then 21.3%. Three seasons do not provide enough evidence to declare this a systematic law. The DNR's annual reports likewise show submitted harvest counts below survey-estimated totals in each year: 297,132 compared with 339,189 for 2022; 267,393 compared with 347,008 for 2023; and 280,304 compared with 373,300 for 2024. These estimates omit DMA permits. The reports say the difference may come from unreported deer as well as survey overestimation caused by low response rates. Published registration totals are available for Michigan's 2025 season. When this analysis was conducted, the survey estimate for 2025 had not yet been released.
- Limits: neither measurement counts the deer herd, and both have sources of error. Registration is affected by failure to report and reporting mistakes. The survey is affected by sampling error, error in expanding the sample to a statewide estimate and low response. According to the DNR, response to the 2021 mail survey fell to 33%. Across our work, we keep the measurements apart and never use their average as a harvest figure.
Fully Drawn Research is an independent data analysis desk, not affiliated with any state wildlife agency, transportation agency, mapping provider or outfitter. Harvest figures are Michigan Department of Natural Resources public records analyzed independently by FDR. Michigan's two reporting instruments are presented separately and are never averaged or pooled; where a single figure is needed, one instrument is chosen and identified. FDR is an independent data analysis desk, not affiliated with any state wildlife agency. Figures current to the 2024 season (registration also available for 2025).
Keep reading
FDR Scout puts county harvest history like this on the map of the ground you hunt. Open Scout →