The Two-Tier Map: Why Half of America's Deer Data Isn't County Data

ITWO REGIMES

Which states record a deer kill's county, and which require a guess

Across the twelve states covered here, agencies generate harvest figures through two approaches. Under the first, hunters must report each animal through registration. They might call, use an online form, visit a checking station, or have a locker scan the tag. The resulting record includes which county the kill occurred in. Michigan, Ohio, Indiana, Wisconsin, Missouri, Iowa, and Illinois use this approach. Their county harvest tables therefore tally submitted reports as a census rather than modeling a total.

The second approach supplies no county-level harvest result. Four states within this group of five, South Dakota, Nebraska, North Dakota, and Kansas, send questionnaires after the season to sampled license holders. They ask about the animals killed and their approximate locations, then weight the replies to calculate an estimate for the whole state. Minnesota comes close to county reporting: every deer must be registered, making its totals counts rather than estimates. However, those counts are released solely for each permit area for deer. All of these states use management-unit geography, whether called a permit area, data analysis unit, or hunting district. For South Dakota, Nebraska, and Minnesota, the states examined, those borders trace habitat, roads, and rivers instead of county borders. The harvest results are entirely valid, but they contain no county-level figure.

IIWHAT COUNTY HARVEST MEANS

What remains unsaid

A registration state's county total of 4,318 deer tallies reported kills assigned to that county in the registration record. This administrative census carries the usual limitations of information people report themselves. A survey state's management-unit figure instead projects harvest through a model, includes sampling error, and applies to an area selected by the state. Some agencies also release the error interval. These measurements cannot substitute for each other. The distinction becomes especially important along borders: management units often include portions of several counties. Dividing a unit's estimated harvest along county borders requires an assumption, such as each county's share of land area or habitat, that the source information never provided. That added assumption produces the survey state's supposed 'county number.' It is also the reason we decline to produce one.

An analyst cannot retrieve a county that the state never recorded. Any total assigned to that county would be guessed.

IIIWHY THE UNIFORM MAP LIES

One set of colors for two measurements that cannot be treated alike

With one legend, a map can paint deer across twelve states in a gradual sweep between Ohio and the Dakotas. The display looks attractive, but its error is specific. Colors in the seven states using registration represent measured counts. For the five lacking county detail, those colors can represent only figures invented by the mapmaker through dividing a management-unit estimate into county portions. Readers encounter an unbroken display and give it an unbroken interpretation, whether deer density, harvest pressure, or hunting opportunity. Nothing tells them that half reflects measurement while half reflects modeling. The deception comes from making the map look consistent.

Wisconsin shows how reporting detail can change even among states with actual harvest counts. The current county series includes 61 counties. An older compilation from the state's wildlife agency provides figures for 72 counties through 2013, including eleven Northern-Forest counties missing from the current series. We give those eleven counties found only in the archive a separate source label. They come from another compilation that ends in 2013, so comparing them with the other 61 counties for the same year does not put both groups on equal footing. For the 61 counties covered by both sources, the figures differ by no more than about 0.2 percent. We will not quietly combine different vintages of actual counts from one state, and we will not make up county figures for a state that never recorded them.

IVTHE HONEST MAP

Match each state's display to the detail its records allow

Use the geographic detail supported by each state's information. Give county shading to the seven states using registration. Use management-unit shading for South Dakota, Nebraska, Kansas, and Minnesota, and statewide shading for North Dakota. Explain those scales in the legend. The display will look less consistent while being more accurate, showing readers precisely where finer detail stops. Comparing harvest between two counties in Iowa relies on measured figures. Asking which county in North Dakota has the highest deer harvest goes beyond what that state's own records can answer. The map should state that limitation plainly instead of hiding it behind a believable shade.

Before you trust a county harvest figure, check one thing: does the state register deer or survey hunters? If its deer registration records include a field that specifically names the county where each deer was killed, the county figure comes from recorded kills. If the state uses surveys, a county figure is either a management unit result relabeled as a county result or a made-up number. A polished map calls for closer checking of the data behind it.