The more attentively people focus on crime video evidence, the less likely they are to arrive at objective conclusions about who is to blame, a new study suggests.

The study, published online in the Journal of Experimental Psychology in September, explains why observers reach different conclusions from the same video evidence. Researchers found that the more observers focused on out-group targets, the stronger the influence of their previous group identification on their decision.

“More attention doesn’t necessarily equate with more objectivity,” said lead author Yael Granot, a New York University graduate student.

Researchers conducted three experiments for the study. Participants viewed video evidence of altercations between police officers and civilians in the first two experiments, and video evidence of a staged conflict in the third. Who was to blame remained ambiguous in all videos. Participants were asked to determine who to punish and how severely to punish them after viewing the video footage.

According to Granot, the research was partly motivated by the 2007 Supreme Court case of police officer Timothy Scott. The primary evidence in the case was the officer’s dashboard footage of a car chase, in which Scott knocked the pursued car off the road. The judges struggled to reach a decision, despite being able to see the evidence with their own eyes.

Two years after the decision, which was called Scott v. Harris, researchers used the Scott video to better understand discrepancies in observers’ conclusions. They found that viewers’ group identifications, such as political beliefs and gender, impacted where they assigned culpability.

These 2009 findings “inspired the study,” said Dr. Granot.

“Sometimes there seems to be systematic group differences in legal judgment. We wanted to see how this interacts with how people are actually viewing video evidence,” she added.

In the two experiments that used video of civilians and policemen, viewers were first surveyed to determine the how strongly they identify with the officer depicted. They then watched footage of the altercation while researchers monitored their visual attention with eye-tracking software.

In general, the stronger the identification with one altercation participant, the harsher the penalty placed on the other participant.

During the second experiment, participants were randomly assigned to focus on either the officer or the out-group civilian.

“Only when those people were staring at that group member [did] polarization begin to happen,” Granot said. “The more we stare, the more we seem to make decisions in line with our group and our identity. If you are not staring at them intently … that difference goes away and everyone punishes equally,” she added.

The last of the three experiments utilized a video of a staged altercation between actors. Group identification was then assigned randomly by fake personality tests that aligned viewers with a side in the conflict.

Those who identified strongly with their group were less likely to accurately recall what the out-group member actually did. Researchers suggested that the lack of attention on the out-group member may be a potential cause of bias in punishment decisions.

“It makes you wonder what other domains it would generalize to,” said Jeff Kukucka, professor of psychology at Towson University.

According to Shana Cole of Rutgers University, the research suggests “persevering and prevalent biases how people take in, interpret, and encode visual information.”

Granot clarified that researchers were “not trying to say that video evidence is bad,” but rather that “video evidence has a unique way of bringing out existing biases.”

According to Kukucka, because people can view the same video and come away with very different interpretations, “we have to be careful to leave fact finders to their own devices,” especially as surveillance technology becomes cheaper and more feasible and video evidence becomes more prevalent.

In future research, Granot hopes to further study people’s motivations for focusing on either the out-group or in-group members as well as interracial altercations.