When considering the factors that contribute to higher earning, education level and occupation are no-brainers. However, a study published today in Frontiers in Psychology has found that a person’s ability to delay instant gratification – a behavioral trait popularized in the the famous Stanford marshmallow experiment, where children had to choose between one small reward provided immediately or two small rewards if they waited for 15 minutes – is actually one of the most important factors determining future affluence. We talked to the study’s lead author, Dr. William Hampton from the University of St. Gallen, to find out more.
Morressier: Briefly explain the findings of your study and their significance
Dr. William Hampton: Existing studies have determined that many factors play a role in how much money a person will make. Some are very obvious, such as education and occupation, while some are less so – taller people earn more, for example. Our study was the first, however, to create a validated rank ordering of these factors (age, occupation, education, geographic location, gender, race, ethnicity, height, age, delay discounting) using machine learning.
Our results were interesting because when we compared the importance of these factors we found that how much a person discounts the value of future rewards compared to immediate ones (known as delay discounting) is more predictive of income than some other ‘big’ variables such as age, ethnicity, and race.
Morressier: What real-world implications could the findings of your study have?
Dr. Hampton: For the general public our findings provide new insights into the factors that influence higher income. People already know that your job and education are important, but it’s likely that few people would guess that your ability to delay gratification is also very important – more important than your age, your ethnicity or your race.
This research may spur a greater interest in how we could train people, especially children, to be better at delaying gratification. That is, if you want your child to grow up to earn a good salary you might consider teaching them the importance of passing on smaller immediate rewards in favor of larger ones that they have to wait for. This is probably easier said than done as very few people naturally enjoy waiting, however our results suggest that those who do will likely be investing in their own earning potential!
Morressier: Are there any limitations to your research?
Dr. Hampton: Besides being limited to the US, we were also slightly limited by our online sample. We did not test, for example cognitive ability or generalized intelligence, which are currently still best measured in-person. We did however, collect education level, which has consistently been correlated with intelligence, so we don’t feel that this is a major flaw.
Finally, our study was cross-sectional, so we can’t say whether lower delay discounting leads to higher income or whether higher income leads to lower discounting. However, we speculate that this may be a consequence of the relationship between higher discounting and other undesirable life choices. For instance, the inability to delay gratification has been associated with the use and abuse of addictive substances such as cigarettes, alcohol, and opiates. Similarly, pathological gamblers have also been shown to exhibit heightened delay discounting. Inability to wait for delayed future rewards is also associated with lower intelligence, and poorer psychiatric health. In this way, one possibility is that delay discounting signals a cascade of negative behaviors that derail individuals from pursuing education and may ultimately preclude entry into certain lucrative occupational niches. Future longitudinal research should be designed to test this theory.
Morressier: Could you explain the significance of the use of machine learning in your study? How do you see this technology changing and improving the way research is carried out in the future?
Dr. Hampton: Using machine learning in this study allowed us to create a ranking of all the factors that we knew were important for predicting income. The methods historically used by social scientists, such as correlations and regression, do not allow for simultaneous comparison when the variables are so mixed – for example, how can you compare age to job category?
For our study, we collected data from over 2,500 participants and then split this data into a training set and a test set. We built our models using the training set and put the test set aside. Once we established a ranking of the variables, we then went back to our test set and tested the accuracy of this ranking. Being able to do this was amazing as it allowed us to check our findings and replicate them, giving us much greater confidence that our findings were accurate. This is particularly comforting given the recent wave of findings across the sciences that do not seem to replicate. Using this approach should lead to research that is highly replicable.
Morressier: What is the next step in this research?
Dr. Hampton: Our study showed which factors are most important for predicting income. So what do we do with this information? Some of these features are out of our control – we cannot alter our age or height. And most people are aware that education is a good investment. But what about delay discounting? I would also be very interested in more studies attempting to reduce delay discounting in people via some sort of training. Our group, and several others, have shown that discounting can be modified in the short-term, and there is some evidence that early childhood intervention can affect discounting behavior, but whether delay discounting is malleable or a stable trait is still hotly debated. I am excited about longitudinal and training studies that will help settle this debate.