Animal behavior research is getting better at preventing observer biases from creeping in – but there’s still room for improvement

Research into animal behavior is based on careful observation of animals. Researchers can spend months in a jungle habitat watching tropical birds mate and raise their young. They could monitor the level of physical contact in cattle herds of different densities. Or they could record the sounds whales make as they migrate through the ocean.

Animal behavior research can provide fundamental insights into the natural processes that influence ecosystems around the world, as well as into our own human minds and behavior.

I study animal behavior – as well as the research reported by scientists in my field. One of the challenges of this kind of science is making sure that our own assumptions don’t influence what we think we see in laboratory animals. Like all people, the way scientists see the world is shaped by biases and expectations, which can influence the way data is recorded and reported. For example, scientists who live in a society with strict gender roles for women and men may interpret things they see animals do as reflecting those same divisions.

The scientific process corrects such errors over time, but scientists have faster methods to minimize potential observer bias. Animal behavior scientists haven’t always used these methods, but that is changing. A new study confirms that over the past decade, studies have increasingly adhered to rigorous best practices that can minimize potential biases in animal behavior research.

Prejudices and self-fulfilling prophecies

A German horse named Clever Hans is widely known in the history of animal behavior as a classic example of unconscious bias leading to a false outcome.

Around the turn of the century, it was claimed that Smart Hans could do math. For example, in response to his owner’s question “3 + 5,” Smart Hans tapped his hoof eight times. His owner then rewards him with his favorite vegetables. Early observers reported that the horse’s abilities were legitimate and that its owner was not deceitful.

However, careful analysis by a young scientist named Oskar Pfungst showed that if the horse could not see its owner, it could not respond correctly. So even though Clever Hans wasn’t good at math, he was incredibly good at observing his owner’s subtle and subconscious signals that gave away the mathematical answers.

In the 1960s, researchers asked participants in human studies to code the learning abilities of rats. Participants were told that their rats had been artificially selected as “smart” or “boring” students over many generations. Over a number of weeks, the participants put their rats through eight different learning experiments.

In seven of the eight experiments, human participants rated the “smart” rats as better learners than the “boring” rats, when in reality the researchers had randomly selected rats from their breeding colony. Bias led the human participants to see what they thought they should see.

Eliminating prejudices

Given the clear potential that human biases can distort scientific results, textbooks on animal behavior research methods from the 1980s onward have implored researchers to verify their work using at least one of two common-sense methods.

One of these is to ensure that the researcher observing the behavior does not know whether the subject comes from one research group or the other. For example, a researcher would measure the behavior of a cricket without knowing whether it came from the experimental or control group.

The other best practice is to bring in a second researcher, who has fresh eyes and no knowledge of the data, to observe the behavior and code the data. For example, while analyzing a video file, I count chickadees taking seeds from a feeder 15 times. Later, a second independent observer counts the same number.

Yet these methods to minimize potential biases are often not used by animal behavior researchers, perhaps because these best practices require more time and effort.

In 2012, my colleagues and I reviewed nearly a thousand articles published in five leading animal behavior journals between 1970 and 2010 to see how many articles reported using these methods to minimize potential bias. Less than 10% did this. In contrast, Infanty magazine, which focuses on human infant behavior, was much more rigorous, with more than 80% of its articles reporting methods to avoid bias.

It is a problem that is not just limited to my field. A 2015 review of published articles in the life sciences found that blind protocols are uncommon. It also found that studies using blind methods found smaller differences between the main groups observed compared to studies that did not use blind methods, suggesting that possible biases led to more notable results.

In the years after we published our article, it was frequently cited and we wondered whether there had been any improvement in this area. That’s why we recently reviewed 40 articles from each of the same five journals for the year 2020.

We found that the proportion of articles reporting bias control improved across all five journals, from less than 10% in our 2012 article to just over 50% in our new review. However, these coverage rates still lag behind Infancy magazine, which was 95% in 2020.

All in all, things are going well, but there is still room for improvement in the field of animal behavior. In practice, with increasingly portable and affordable audio and video recording technology, it is becoming easier to implement methods that minimize potential biases. The more the field of animal behavior adheres to these best practices, the stronger the foundation of knowledge and public trust in this science will become.