4 ms·

> it is only bi-modal if there’s a single large dip in the middle of the histogram. I don’t think this is correct, bi-modal just means there are two modes. But

> it is only bi-modal if there’s a single large dip in the middle of the histogram. I don’t think this is correct, bi-modal just means there are two modes. But we don’t have to argue over a definition, I’ll just use more words. What I mean is very simple - if you go to college and get a better job due to your degree, you are undoubtably better off. If you go to college and do not obtain a better job with your degree, you are not better off (and are now worse off due to crushing debt). So there are broadly two types of people with degrees, and pay is radically different for the two groups. When discussing these two groups of people, median is not a very useful concept - you are one group or the other, you are not going to fall in the middle. Sure, college might be a statistically beneficial choice, but it is a risky choice - if you lose, you lose big. In case we are still not on the same page, I’ll give a very simple example - a person goes to college for an IT degree. The person is unable to land a job in IT despite their degree, and remains in unskilled labor. How is this person better off for their degree? And how are they supposed to come up with hundreds in monthly payments for the next few decades with their low paying job? Or are you really contesting the notion that a person may not obtain a job/career with their degree?

dahart 4y ago
> bi-model just means there are two modes. Which confirms what I said. Two modes means two peaks, which means there’s a dip between them. You’re not disagreeing with my definition, you’re confirming it. https://en.wikipedia.org/wiki/Multimodal_distribution https://en.wikipedia.org/wiki/Multimodal_distribution This definition is important because this is what you’re misunderstanding about both my argument and your own. You are drawing a line between the inputs in the distribution, but failing to understand there is no line in the outputs. The distribution of pay and outcomes is not bi-modal, regardless of the fact that any given individual either did or did not go to college. The collective aggregate behavior of the system has a smooth distribution, where you cannot see two different lumps (modes) that identify the degree holders and non-degree holders. https://en.wikipedia.org/wiki/Household_income_in_the_United_States#/media/File:Education_Income.jpg https://en.wikipedia.org/wiki/Household_income_in_the_United... https://www.statista.com/statistics/203183/percentage-distribution-of-household-income-in-the-us/ https://www.statista.com/statistics/203183/percentage-distri... > Or are you really contesting the notion that a person may not obtain a job/career with their degree? This isn’t just a straw man, I already explicitly addressed this at least twice. I did not claim that people with degrees can’t fail; they can and do sometimes. I even gave an example of how they sometimes fail, no need to misrepresent me on the very thing I’m agreeing with you about. My point, to state it yet again, is that people getting degrees and ending up no better off is a small minority of cases. Similarly, people who don’t get degrees and end up rich is a small minority of cases. The majority of cases, as the data proves, shows that people with a degree are statistically better off than people without. Not always, just the majority of the time. Unlike your argument, this isn’t logic, it’s just a fact. > median is not a very useful concept […] college might be a statistically beneficial choice, but it is a risky choice Hehe this argument is getting funnier. You’re effectively saying don’t look at the data, don’t look at the expected value or the probability, only look at the loss of a low-probability event. This is wrong, IMO. The whole reason that median is a useful concept is because it tells you something about the expected value. Both the probability of success of getting a degree, and the expected value, according to the data we have, show that it’s quite a bit riskier to forego the degree than to get it. You can fret about how risky getting a degree is, but that’s not helpful if you ignore the risks of not getting one. I also don’t agree with your conclusion, and the data does not support your assertion that people either win or lose, nor that they “lose big” when they “lose”. It’s not binary. There’s a wide range of outcomes for people with degrees, and the whole distribution skews toward positive outcomes compared to not getting a degree.

dahart 4y ago

> bi-model just means there are two modes. Which confirms what I said. Two modes means two peaks, which means there’s a dip between them. You’re not disagreeing with my definition, you’re confirming it. https://en.wikipedia.org/wiki/Multimodal_distribution https://en.wikipedia.org/wiki/Multimodal_distribution This definition is important because this is what you’re misunderstanding about both my argument and your own. You are drawing a line between the inputs in the distribution, but failing to understand there is no line in the outputs. The distribution of pay and outcomes is not bi-modal, regardless of the fact that any given individual either did or did not go to college. The collective aggregate behavior of the system has a smooth distribution, where you cannot see two different lumps (modes) that identify the degree holders and non-degree holders. https://en.wikipedia.org/wiki/Household_income_in_the_United_States#/media/File:Education_Income.jpg https://en.wikipedia.org/wiki/Household_income_in_the_United... https://www.statista.com/statistics/203183/percentage-distribution-of-household-income-in-the-us/ https://www.statista.com/statistics/203183/percentage-distri... > Or are you really contesting the notion that a person may not obtain a job/career with their degree? This isn’t just a straw man, I already explicitly addressed this at least twice. I did not claim that people with degrees can’t fail; they can and do sometimes. I even gave an example of how they sometimes fail, no need to misrepresent me on the very thing I’m agreeing with you about. My point, to state it yet again, is that people getting degrees and ending up no better off is a small minority of cases. Similarly, people who don’t get degrees and end up rich is a small minority of cases. The majority of cases, as the data proves, shows that people with a degree are statistically better off than people without. Not always, just the majority of the time. Unlike your argument, this isn’t logic, it’s just a fact. > median is not a very useful concept […] college might be a statistically beneficial choice, but it is a risky choice Hehe this argument is getting funnier. You’re effectively saying don’t look at the data, don’t look at the expected value or the probability, only look at the loss of a low-probability event. This is wrong, IMO. The whole reason that median is a useful concept is because it tells you something about the expected value. Both the probability of success of getting a degree, and the expected value, according to the data we have, show that it’s quite a bit riskier to forego the degree than to get it. You can fret about how risky getting a degree is, but that’s not helpful if you ignore the risks of not getting one. I also don’t agree with your conclusion, and the data does not support your assertion that people either win or lose, nor that they “lose big” when they “lose”. It’s not binary. There’s a wide range of outcomes for people with degrees, and the whole distribution skews toward positive outcomes compared to not getting a degree.