David Trafimow, a Distinguished Achievement professor of psychology at New Mexico State University, was recently selected to receive the prestigious Humboldt Research Award for 2026. Only 100 Humboldts are granted worldwide each year across all disciplines. Scott Brocato recently spoke with Professor Trafimow about the honor, and his contributions to the field of psychology.
Scott Brocato:
Talk about the Humboldt Award, which is bestowed by the German government.
David Trafimow:
So it comes with a cash prize of 80,000 EUR, and there are two meetings. One meeting will be in March and another meeting will be in June. The first one in Bamberg and the other one in Berlin.
Scott Brocato:
How was your work selected? What is the process of receiving the award?
David Trafimow:
You have to be nominated and there's a...I don't know what to call it...a difficult process of that. There has to be a person with the right qualifications to nominate you, but you can't nominate yourself.
Scott Brocato:
And (the Humboldt winners) are from many different fields, correct?
David Trafimow:
Right. It's only one person in each field who gets one, but since there are lots of fields, there are multiple Humboldt winners. And so this is a chance where we can meet each other and have discussions and see if there's cross-fertilization. I personally am really looking forward to it because these are all top people in their fields, and the opportunity to get an idea of how they think is going to be irresistible.
Scott Brocato:
Let's talk about your work. How long have you been a professor here at NMSU?
David Trafimow:
Since, I believe, 1994.
Scott Brocato:
And when did you become interested in psychology, and how did you end up at NMSU?
David Trafimow:
A long and difficult road. I got interested in my senior year in high school when I just read a book at random from the library, and it was about psychology, and I thought it was interesting.
Scott Brocato:
You’ve received multiple recognitions for your contributions, particularly for your efforts in challenging conventional statistical practices and proposing alternative inferential methods. Could you explain that?
David Trafimow:
Sure. As of now, the sciences, including psychology, use something called Null Hypothesis Significance Testing. And the idea of it is, when you get your data, you do some sort of significance test. And if the significance test comes out at the right level, then you conclude that the results are statistically significant, and then you believe them. Whereas if the results are not statistically significant, then the idea is that perhaps they're reasonably likely to be due to chance, and then you won't believe them.
The problem with all of what I just said is that the whole procedure is unsound. And so along time ago, I wrote a paper that I published in Psychological Review, I think that was 2003, where I showed that it's unsound. And frankly, it was mostly ignored at the time. However, when I became editor of Basic and Applied Social Psychology in 2014, I realized that I could, as a journal editor, I could actually ban significance testing from the journal. And so I did. That was in 2015. And I think that's what really started this big revolution, because the published paper itself, even though I still like it--although I'd make changes if I were to rewrite it--I don't think that paper by itself had a big influence on the field.
Scott Brocato:
When you say the method is unsound, in what way?
David Trafimow:
Okay, that's a complex topic, but it's unsound for multiple reasons.
For those who conclude that if the result is statistically significant, therefore the hypothesis they want to support is likely right, there's a logical problem with that in the first place, and in this because the conclusion doesn't follow from the premises. And the second problem is that what you do when you do a significance test is you embed something called the null hypothesis. This is the hypothesis that nothing happened of interest. And so you want to disconfirm the null so you can say, “Aha! See? I win!”
However, what you're really doing is embedding the null hypothesis in a large set of additional assumptions. For example, one of the additional assumptions is that you've randomly selected from the population. And of course, that assumption is always false. So what it amounts to is that you're embedding the null hypothesis in a known wrong model, and then showing that--and this is at best because even this is questionable--but at best what you're showing is that the data are unlikely ,given the model. But since you already know the model is wrong, before you've even started, it's a rather pointless procedure. And you certainly wouldn't want to draw conclusions about a hypothesis embedded in a wrong model.
To give you an analogy, this might make it easier: imagine there was a museum, and the museum was an ugly museum. Now imagine you're concerned with a particular painting in the museum. Would you conclude that the painting is ugly because the museum is ugly? You wouldn't. So because the model is wrong, would you conclude that the hypothesis is wrong? You wouldn't.
Scott Brocato:
Well, this also may also tie in with another part of your research, which focuses on how the beliefs and opinions people hold about themselves are structured and how they influence their actions, decisions, and habits. Talk a little bit about that.
David Trafimow:
So this work actually goes all the way back to my days in graduate school. And there was a professor, Martin Fishbein, who's unfortunately no longer with us, but he invented what's called the Theory of Reasoned Action. And the idea is to try to figure out what causes behaviors. And according to this theory, it was attitudes and subjective norms. Attitudes are basically how you evaluate the behavior, and subjective norms are the opinions of others who are important to you about whether you should perform the behavior. And other people have added other variables, but that's the basics.
And I got into it because he was giving a talk, and one of the people in the audience--in fact, multiple people in the audience--gave him a hard time, saying, “Well, how do we know that attitude and subjective norms are really different variables? How do we know that they're not just different names for the same variable?” And he didn't really have a great answer. But an answer occurred to me, sitting in the audience as a graduate student. So I went to his office and I explained that I thought there was a way to actually test this definitively. And he agreed with me. And so we started doing that work, and that's how I got started in that area.
