Subscribe to the OSS Weekly Newsletter!

Register for the OSS 25th Anniversary Event

Tips for Better Thinking: Not-So-Total Recall

Taking a stroll down memory lane may be hazardous to certain types of scientific studies

How many eggs did you consume last month? For some, the answer is quick and easy: zero. But for those of us who consume eggs regularly but not according to a fixed schedule, answering this question requires a trip down memory lane, and that road unfortunately gets repaved fairly regularly.

Scientists love to study potential links between being exposed to something and developing a certain disease. For example, let’s imagine there is an affliction called Humpty Dumpty disease. Is eating a lot of eggs associated with developing Humpty Dumpty disease?

An easy way of attempting to shed light on this question is to give two groups of participants a questionnaire and ask them to think back on their egg consumption over the past month or year. One group has been diagnosed with Humpty Dumpty disease, the other has not. This is known as a case-control study. Scientists compare the two groups to see if there was on average more egg consumption in the group diagnosed with the disease.

This type of retrospective or “looking back” study can unfortunately fall victim to recall bias. Recall bias is a distortion that happens when, in this example, people with and without Humpty Dumpty disease remember their egg consumption differently. The people with the disease may be actively searching for something to blame. They may have seen headlines on news sites or chatter on social media saying eggs could have something to do with Humpty Dumpty disease. They may thus search their memory more thoroughly when asked about their egg consumption than someone without the disease, thus skewing the results of the study.

We don’t tap into our memories the way that our computer opens a file; we are always subtly recreating our memories as we access them, which means they can change over time. And just because we are confident in the memory we have of a surprising and consequential event in our lives (a so-called “flashbulb memory”), it does not mean the memory is accurate. A study on how well Americans remember the events of September 11, 2001 revealed that these memories can be altered after a single year. Indeed, on average, the people who were surveyed right after the event and then eleven months later were consistent in their answers 63% of the time on average.

Questionnaires that rely on people’s memories are still widely used in research because of how easy they are. No need to ask a group of people to document their egg-eating habits every single day; just one questionnaire asking them to self-report what they can remember. But this ease of use comes at a price: accuracy. One of the best ways to avoid recall bias in research is to instead follow people in time and ask them to document their exposure. These prospective studies are not as quick and dirty as asking someone to fill out a single piece of paper, but they have the potential to produce more reliable data.

So when we read about a new study showing a link between oral contraceptives and breast cancer, or talcum powder and ovarian cancer, or cell phone use and cancer of the pinky finger, we would do well to remember to ask the question: is this based on remembrances?

How many eggs did you consume last month?


Leave a comment!

Back to top