The Marshmallow Test: What Does It Really Measure? • Hacker News Board

Re: The Marshmallow Test: What Does It Really Measure?

#221

Post by dang » Sun, Jun 03, 2018, 7:47 PM UTC

troupe wrote: post #216

comboy wrote: post #130

Earlier quoted context omitted.

Marshmallows represent different value for kids with different background. Some can have them every day, for some it's something they only saw other kids eating. It's similar to how it would be easy to risk $50K for somebody with $100M net worth compared to somebody whose net worth is $100K.

In general access to candy every day is something I see more at the low end of the income spectrum and less at the high end.

That may have changed quite a bit in the decades since the original study.

Re: The Marshmallow Test: What Does It Really Measure?

#222

Post by rhapsodic » Sun, Jun 03, 2018, 7:47 PM UTC

comboy wrote: post #130

abainbridge wrote: post #9

> The researchers also, when analyzing their test’s results, controlled for certain factors—such as the income of a child’s household Is that a valid thing to do? If "will power" was a strongly inherited trait, and the theory is that will power leads to increase household income in adult life, then this controlling would remove most of the signal. The original paper relied partly on a questionnaire that asked the par…

Marshmallows represent different value for kids with different background. Some can have them every day, for some it's something they only saw other kids eating. It's similar to how it would be easy to risk $50K for somebody with $100M net worth compared to somebody whose net worth is $100K.

>Marshmallows represent different value for kids with different background.

Or kids with different tastes. I liked candy and cookies as much as any other kid, I suppose, but I never much cared for marshmallows. For me to leave one sit there would not have required much strength of will.

I wonder why they didn't ask each kid what their absolute favorite candy or sweet treat in the world was, and use that for the test instead.

Re: The Marshmallow Test: What Does It Really Measure?

#223

Post by jabgrabdthrow » Sun, Jun 03, 2018, 7:57 PM UTC

yk wrote: post #128

4bpp wrote: post #107

So this is suggesting that childhood poverty may be a strong correlate of whatever the marshmallow test tests (and childhood poverty is plausibly a strong correlate of life outcomes, so the effect might be completely mediated via childhood poverty rather than any notion of "affinity towards delayed gratification"), but I'm not sure I understand how/whether it rules out the following two. (1) whatever quality X the ma…

The study is suggesting that the marshmallow test is a test of mother without college degree, and not testing something about the children beyond that. [0] And in general, my suspicion about psychology is, that since you can always invent an objection in the form of your (1) and (2), I doubt that statistic is an actually good tool to look at these tests. If we assume for a moment that we somehow get a ground truth of…

> I suspect that we would see that we need as many categories as there are children to get anything predictive.

Making a category to predict human behavior generally hurts those who are “miscategorized” much more than it helps those who are accurately categorized (there aren’t really “positive stereotypes” that jump into active-decision processes like with negative stereotypes).

Therefore, we should not try to predict behavior by making even fuzzy-categories other than one-child-per-category, since this may literally hurt children, especially systemically underprivileged and underrepresented children.

It’s important to control your own decision factors against your local collection of “known non-causations”!

Re: The Marshmallow Test: What Does It Really Measure?

#224

Post by arpa » Sun, Jun 03, 2018, 8:10 PM UTC

paulddraper wrote: post #207

jakobegger wrote: post #159

Earlier quoted context omitted.

> this controlling would remove most of the signal That's the whole point of controlling. If all of your signal can be explained by a correllation, then you aren't measuring what you think you are. If "willpower" (waiting for second marshmallow) was important, it would be important regardless of whether your parents are rich or not. This means that among poor kids, those with willpower would be more successful, and a…

> The important factor is your parents income, not your "willpower". Or perhaps the opposite is true. Maybe all the studies that reference parental income "should" be referencing the child's willpower. Maybe willpower is the most important/influential trait, and we should be looking at the several factors increase willpower: parental income, parenting technique X, an inherited willpower gene, etc.

Yes, because the more willpower your child has, the more income you will have!

Re: The Marshmallow Test: What Does It Really Measure?

#225

Post by gnicholas » Sun, Jun 03, 2018, 8:20 PM UTC

Did the new researchers check to see if the parents were aware of the marshmallow test or had ever tried it with their kids?

My guess is that at least some of the parents would have heard of it since it is very well-known. These days, it's even in kid-focused apps [1] that are designed to help with emotional development. I know many parents who have tried it with their kids to see how they'd do.

It's likely that knowledge of the marshmallow test correlates with education and income, so this could be part of the reason for the effects the researchers saw.

1: https://eqtainment.com/

Re: The Marshmallow Test: What Does It Really Measure?

#226

Post by watwut » Sun, Jun 03, 2018, 8:24 PM UTC

taneq wrote: post #36

zeth___ wrote: post #4

Earlier quoted context omitted.

If you think poor people have time for doing that you haven't been poor.

So maybe well-off people score better on the test because they have time to teach their kids some self-restraint, not because they're genetically better able to resist marshmallows?

The question was not whether delayed gratification is genetic or not. The question was whether not eating candy when stranger promiss you more candy for not eating it makes you more successful later in life.

The correlation was found weak.

Re: The Marshmallow Test: What Does It Really Measure?

#227

Post by learc83 » Sun, Jun 03, 2018, 8:25 PM UTC

mynameishere wrote: post #100

cimmanom wrote: post #62

Earlier quoted context omitted.

Because they have access to only poor quality foods. There may not be calorie scarcity but there is nutrition scarcity. And yes, for some there's also "food insecurity" which is not knowing where your next meal is coming from or having to choose between food and other necessities (shelter, warm winter clothing, medical care, transportation, etc.)

This just isn't true. It's so obviously not true it's really very amazing that people repeat it. Go to a Dollar General in a poor area sometime. They have lots of potato chips and assorted crap. Yep. They also have canned beans, fruit, refrigerated sausage, cheese, chicken, etc. None of this is unaffordable. That's Dollar General. Most places have access to a Walmart which is as good nutritionally as any grocery stor…

Cheap junk food is much more palatable than cheap healthy food.

> Most places have access to a Walmart which is as good nutritionally as any grocery store that ever existed.

Walmart tends to avoid the inner city. If you don't have a car there generally isn't one close enough to shop at more than once a week or so. Bringing home a week of groceries for a family on bus isn't a trivial task.

It's much easier to pick up some cheap food from a convenience store.

> Christ, even your typical gas station sells salad these days.

If you compare the per calorie price of that salad to junk food, it's much more expensive.

Re: The Marshmallow Test: What Does It Really Measure?

#228

Post by dahart » Sun, Jun 03, 2018, 8:34 PM UTC

ravar wrote: post #147

dahart wrote: post #143

Earlier quoted context omitted.

> Is that a valid thing to do? If "will power" was a strongly inherited trait, and the theory is that will power leads to increase household income in adult life, then this controlling would remove most of the signal. I would think so. If what seems like “will power” is strongly correllated with your parents’ income, that seems to demonstrate that will power is not a strongly heritable trait, thus the hypothesis is l…

Just because something correlates with parental income doesn't mean that it isn't heritable. Take an example that you agree with. Being white (or Asian) correlates with parental income. Those traits are also rather heritable.

I agree with that, which is precisely why you should control for it, no?

You’re example is correct, but not what I was talking about. You’re only talking about one kind of correlation, and assuming that the data is otherwise ignored. What I imagined was will power being easy to test for heritability if it changes from generation to generation. Being white or Asian might correlate with average income for a population, but Asianness & whiteness don’t change with income. On the other hand, if “will power” changes from generation to generation and the sign and magnitude of the change correllates with the sign & magnitude change of income, then the likelihood that will power is heritable drops rapidly.

Re: The Marshmallow Test: What Does It Really Measure?

#229

Post by gnicholas » Sun, Jun 03, 2018, 8:36 PM UTC

ouid wrote: post #158

thedevil wrote: post #153

> The failed replication of the marshmallow test Actually, the result DID replicate here even though the experiment was slightly different. Other variables were introduced that might suggest additional explanations of the replicated result.

Marshmallow willpower is still correlated with future success, but only because it is correlated with wealth. This means that there's isn't a causal link between eating marshmallow willpower and success. which is a different result, as the previous result included that as a possibility.

If success is correlated with wealth, then wouldn't the original study (done with kids at a Stanford preschool) have shown a small/null result? That is, most of the kids have wealthy and well-educated parents.

Re: The Marshmallow Test: What Does It Really Measure?

#230

Post by white-flame » Sun, Jun 03, 2018, 8:37 PM UTC

chiefalchemist wrote: post #163

white-flame wrote: post #160

Earlier quoted context omitted.

Seriously, who in western countries does not have availability to "healthy food"? Everybody has basic grocery stores with cheap simple stuff just as available to them as convenience junk shops. It's education about how food affects you that makes all the difference. People literally don't know what they're putting in themselves. There's a lot of food purchasing decisions based on stressors related to your very reason…

Not that I trust the CDC per se but if you query yourself you'll see there are plenty of other sources. https://www.cdc.gov/features/FoodDeserts/index.html Regardless, what you're naively discounting is the effects of poverty on the mind (and the body). Imagine, if you can, a state of permanent stress, fear, worry, etc. The human mind is __not__ wired to make optimal decisions under such circumstances. Now let's just…

A main quote from that link is "a small percentage of American consumers are limited in their ability to access" healthier food, seemingly mostly tied to lack of transportation to otherwise close-by stores in densely populated areas. I would posit that's a much smaller percentage than those eating unhealthily in low/poverty income situations.

Your post here coincides with mine. Somebody running between jobs does have means of transportation, but is in a highly stressed life situation. The main problem is not one of literal access to healthy food, yet that is always the first supposition made. Healthy food is cheap, and it is not far away. The problems are beyond that.