Bias: data and people

St. Louis is a strange city. The city on the Mississippi River is divided into a northern and southern part by an invisible border. Both parts are roughly equal in population density, and there are a similar number of parks and green spaces on both sides. the animals Such as conspicuous warblers or gray squirrels are observed almost exclusively in the south. It's a real conundrum for anyone who looks at maps from app-based citizen science projects like iNaturalist or eBird, which collect biological species frequencies — that is, from research projects where ordinary people aren't interested in collecting data. What makes North St. Louis so unattractive to all Animal City residents?

Biologist Elizabeth Carlin of Washington University in St. Louis faced exactly this question in 2007 when she was new in town and sought help via a biology orientation app. “It seemed strange to me, especially since the trees seemed evenly distributed throughout the city,” Carlin said. So she looked for it herself. And I found lots of gray squirrels and other animals – even up north!

The solution to the puzzle soon became clear to her: the monitoring data said less about gray squirrels and more about those who collected the data. Residents of the poor, mostly black northern parts seemed to have other things to do than look for rodents or rare birds and then enter them into a cell phone app. Where no one looks, nothing will be found. Logical.

that it The classic case of distorted dataIn science we talk about bias. Such bias in projects involving citizen participation can arise not only because of the different socio-economic status of potential participants locally. Carlin used her experience as an opportunity to work with colleagues to collect other factors that can make data unreliable: showy or beautiful species are reported more often than inconspicuous or plain-looking species—and biological applications are more likely to be used in parks than in gardens. . Industrial areas. In the magazine People and nature Scientists warnTo consider all of this when working with citizen data and designing projects in a socially just and accessible way.

But bias doesn't just exist in citizen science. One example shows that the results of classical science sometimes say more about the researchers themselves than about their research objects The current study in Nature Communications. It has to do with the bias, dating back to Charles Darwin, that males are usually larger than females among mammals. According to the corresponding story, whoever is tallest is the one who wins the competition when choosing a partner.

One might think that an easy statement to test. But here too the answer depends on what you look at. By far the largest groups of mammals, according to the researchers from New York and Princeton, are rodents and bats, and they have traditionally been underrepresented in studies. Biologists have always been most interested in artiodactyls, carnivores, and primates. The big male theory applies to them.

American scientists led by Kaya Tombak have now used a comprehensive set of statistically high-quality data and have determined: Darwin's claim is completely untrue. In the majority of the 429 species examined, males were not larger. According to the researchers, the prevailing model of large males has led to alternative hypotheses being ignored or treated particularly critically. Conclusion: Data alone is not always enough for solid science. Often you also need to know who collected them – and with what preferences.