Quitting vs Grit
Levitt and Dukes' Advice Makes No Sense and Begs the Question
Full disclosure: I really like Stephen Levitt and I enjoy his podcast people I mostly admire. It's one of the few I've listened to all the episodes. Recently I listened to an episode that I'm critical of. It's about quitting and grit. This piece will end with the best answer to the question of how to balance the two. You won't get an answer like this anywhere else, I promise.
This episode of the podcast tried to grapple with a very tough issue: when should you be a quitter, and when should you have grit and perseverance.
The two concepts are very simple: quitting, is frequently discussed on the show. Steve Levitt is a believer that we don't quit enough, not soon enough, not often enough. He has conducted a randomized trial that is pertinent to that question. The guest supports that general worldview, and has thought about tactics to help yourself quit.
Grit is a concept popularized by Angela Duckworth, and is the idea that for something very difficult, doing it can be challenging, and you may want to quit along the way, but you should have the perseverance to stick through it. You will be happy you did.
Ultimately, Steve Levitt tries to harmonize these two ideas by saying: you should quit if the goal is actually something you don't really want, and if so, quit early, and you should have grit, if the goal is something you really do want.
But this just begs the question: should I write a novel? It depends on what happens. On the one hand, it could become a splendid success, on the other hand, it could be a modest failure. I want to be a successful writer (not true for me just FYI), but is it a good goal or bad goal? How do I know? The episode just begs the question. It says quitting and grit can be individually deployed depending on the net expected outcome, but the truth of life is you often don't know the net expected outcome. So what do you do? When do you be gritty, and when you quit? We are as far away from the answer as when we started.
Let me first talk about the strongest piece of evidence offered in the podcast. It is a summary of the randomized control trial run by Steve Levitt. It's a very clever study. He enrolled people who face a decision in life to which they're willing to subject it to a coin flip. And then he randomly assigns them: stick with it, or give it up. And it is clearly the case that give it up is the right answer. But this only applies for these people. Who are these people? These are the kinds of people who have followed Steve Levitt for years, and like him. I put myself in that category. But they're also the sort of people who have a decision in their life that they're willing to subject to randomization.
I'm not in that category.
Don't get me wrong. I would happily be a participant in many randomized studies. I would happily be randomized to drugs, testing strategies, masking, a whole load of things. But when it comes to decisions like should I stick with this job, should I stick with this friend, decisions that only applied to my life, and I am the only volitional actor, I have a hard time imagining any decision that I would subject to randomization. I usually know one way or the other. I'm not on the fence kind of person. So, my point is, the people enroll in a study are not only fans of him, but very likely people who face this unique decision that they're willing to subject to randomization. As such, it's hard to believe his study has generalizability beyond that rare instance. Certainly not to my life.
The study participants have been inculcated in the view that quitting is okay, and good. Naturally, if they get that point of view they may be more happy. It would be interesting, if somebody who is a proponent of grit ran a similar trial. Or someone in whom you don't know what their worldview is. And it would be interesting to do it with different recruitment strategies.
As for the anecdotes presented, they are just talk. Steve's son quit a sport that he wasn't good at, and then he quit a sport that he was good at. One was rationalized by saying better to quit early, the other was rationalized by saying he must really not have liked it. But it's impossible to know the counterfactual. What would have happened had he persisted? In either? And he's also probably brought up in an environment where quitting is validated.
An example given by the guest was she quit her PhD program, but now has re-enrolled in a PhD program. So just because you quit doesn't mean you're out forever. But this just begs the question of might she have been happier had she stuck with the original program? Who knows?
Here's the messy reality:
This topic, should I quit or should I stick with it is of massive public interest. Many people in all stages of decision making want to listen to this topic, even if they themselves are not willing to randomize their choices. It's all part of the genre: How can I be more successful and happier?
And the bitter truth is, we lack robust, large scale, replicated, randomized trials. Steve is the only one who's actually made a good faith attempt at trying to answer this question rigorously. In the absence of that, people are just talking bullshit. They have no idea if you should quit your PhD program, trade in your car, divorce your spouse, break a friendship, make a friendship, move in together, sleep with her, or break it off—-they just have no idea.
No one can say: if you have a question about your marriage, better to divorce. They don't know that to be true. No one can say: if you have a question about your job, better to find a new one. They have no evidence for that.
In the absence of this, how do we make decisions? Probably, we all do the same. We talk to people we know and trust. We make lists of pros and cons, if not on paper, then in our minds. And we have no idea if our process is optimal or suboptimal.
I suspect that for many of us, depending on our mood, we dream of a different counterfactual. When I'm feeling happy, and validated, I'm so grateful I switched jobs. When I'm feeling blue, and stressed, I imagine things would have been nice had I stayed. Steve's randomized trial tells me that if I was the sort of person who would have subjected this question to his coin flip, then on average I made the right choice. But I'm not that type of person. I wouldn't have subjected this choice to a coin flip. Because I was pretty sure moving was the right choice. I didn't want to stay.
I don't want to say we will never know generalizable answers to the questions of should I stay and should I persevere. We may. But we must admit the current science is in its infancy. Proof of that is we just have one very small study. Imagine what would happen if we had a thousand, or 10,000 such studies. Now you're talking. Maybe there's some interaction effects. Maybe it's better to quit under some circumstances, but not under others. Maybe we can improve upon the imperfect method of asking advice from a trusted friend, and sleeping on it. Then again, maybe not. Maybe our method, which feels incredibly ad hoc, and unscientific is actually the best people can do. After all, all decisions are personalized decisions. Perhaps they don't yield to generalizable principles.
I'm interested in what you think, but also what Steve levitt thinks.
Finally, let me conclude with one observational: among the most popular non-fiction books, perhaps with the rare exception of Freakonomics itself, an incredibly popular theme is how to improve my life. Some books are explicitly marketed as self-improvement, but other books promise you a perspective that will help your day-to-day decision making. Books by Adam Grant, Angela Duckworth, Harari and others fit this bill. The appetite to go out and read something to transform your life is tremendous. As such, I would contend a major bias in this space is the desire to have something to offer.
Your book has to have a theme. Be nice, be perseverant, quit, have a range of interest, focus-- You have to offer something. As such, I would imagine the biases tremendously against the null. That it actually doesn't matter that much, and that we actually don't have very broad generalizable rules that can help you.
I suspect that point of view is deeply underrepresented. I worry then the temptation is to offer something that is flimsy. But all this said should still acknowledge that Steve has done the best work to date. Because he actually ran a randomized trial.
Just some thoughts on the topic from the lens of evidence-based medicine and policy. If you enjoy, subscribe to the substack.