By Dr. Ken Broda Bahm:
U.S. employment numbers were up 178,000 last month, causing many to see an improving economy. The month before that, the number was down by more than 66,000, causing many to fear a worsening economic picture. Truth is, that number routinely fluctuates month to month for a variety of reasons, and the number that happens to be current end point on the chart isn’t so much a predictive bellwether, as it is simply the most recent data point. But we don’t treat it that way. Instead, through something called the “end point bias,” we tend to overvalue the most recent data point. Similar to the “peak-end rule,” the bias in evaluating experiences by giving undue weight to how they were at their peak and at the end, the end point bias also reflects a tendency to overvalue a change that restores a prior loss and the habit of treating a short-term fluctuation as an indicator of a long-term trend. That selective and partial use of information affects our perception whenever we are trying to understand the meaning of numbers over time. In litigation, that could apply, for example, to interpretations of a plaintiff’s past earnings, or a company’s minority hiring trend, or a manufacturer’s product safety record.
New research from professors Bruce Hardy and Kathleen Hall Jamieson takes a look at this end point bias as well as how to solve it. The study (Hardy & Jamieson, 2016) looks at the way we understand climate change issues, noting the phenomena where “unusually cold winters, a slowing in upward global temperatures, or an increase in Artic sea ice extent are often falsely cast as here-and-now disconfirmation of the scientific consensus on climate change.” The study is summarized in a recent piece in ScienceDaily and looks specifically at scientific reports on the levels of Artic sea ice in 2013. In that year, the same data was presented by the National Oceanic and Atmospheric Administration (NOAA) as, “Sea ice extent larger than 2012 record low, but still sixth smallest on record,” and by FoxNews as “Artic sea ice up 60 percent in 2013.” Both are accurate, but the Fox headline plays to end point bias by treating a small fluctuation as an indication of a favorable trend. Reading the latter story biased study participants by making them less likely to understand the actual downward trend in Artic sea ice, and while the bias worked especially well for conservatives, liberals and moderates were influenced as well.
Rather than just leaving this in the category of “depressing findings about popular understanding of science,” however, the research team looked into how the effect could be mitigated. Indeed, they found that end point bias can be significantly reduced with what they call the “LIVA” technique, which stands for “leveraging, involving, visualizing, and analogizing.” The influence of the end point is placed in context, they write, when “scientific content from a credible source…is leveraged in a visualized manner that invites the audience to centrally process a trend line that is contextualized with an illustrative analogy.” This approach might also apply to other biases based on numbering, like anchoring and estimating probability.
So let’s take a look at each step in the LIVA process.
Leverage
Because credibility often needs to come before comprehension, the first step of the process is to leverage the credibility of a scientific source. In the study, that meant emphasizing not just the data, but also the interpretations and opinions of respected scientists. In litigation, it means highlighting your expert’s credibility first and formost. As I’ve written recently, that requires not just stating the credentials but leveraging them so the expert’s knowledge and experience serves to give jurors a reason to believe that the expert can help them solve a problem at the heart of the case.
Involve
The strategy requires involvement, and that means considering the audience. In the study, the debiasing approach was less effective with conservatives, and one reason for that may have been a tendency of some conservatives to doubt government scientists (believing that these scientists are motivated to push the climate change agenda in order to win and to protect their sweet government paychecks). So greater involvement in that context would require finding sources conservatives trust.
Visualize
In the study, the researchers didn’t just present the data all at once. Instead, they built the following chart dynamically over 22 seconds to “invite careful central processing about the information in the iterative graph.”
Looking at this chart, you see the folly in focusing, as FoxNews did, only on the jump from 2012 to 2013 and not the overall downward trend from 1979. Adding one point at a time helps to counter end point bias by giving as much weight to each point as it gives to the end.
Analogize
A visual presentation works for some, but others will need a more verbal way of understanding it. For that, the analogy works well because it draws on a familiar scenario. In the study, the authors used the following analogy to contextualize the 2013 sea ice level:
Expecting sea ice to return to its 1979 level based on the improvement in 2013 is like earning a C on a first exam, a D on a second, an F on a third, a D on a fourth and as a result of that recent D, anticipating an A on the final.
Whether it is in the form of salary records, past profits, economic indicators, or any other data series that could be relevant in litigation, we are not always the best judges of long-term trends. For example, the authors point to past research showing that the outside temperature on the day a survey is taken influences how likely respondents are to say that global warming is occurring. We are biased by what is saliant and at hand. That’s the bad news. But the good news is that there are debiasing strategies that seem to improve our odds of getting a more accurate assessment.
______
Other Posts on Cognitive Bias:
______
Hardy, B. W., & Jamieson, K. H. (2016). Overcoming endpoint bias in climate change communication: the case of Arctic sea ice trends. Environmental Communication, 1-13.
Photo credit: 123rf.com, used under license, edited