But we can see very definitively that a higher-than-average rate of education does nothing to increase the partisan character of the voting block.
Mormons have more bachelor degrees than average: Yes.
Mormons are more conservative than average: Yes.
We can see very definitively that a higher-than-average rate of education does nothing to increase the partisan character of the voting block: No.
That’s not what this shows. The two facts, together, are not enough to show a correlation between education and conservatism or to show a lack of correlation between education and liberal/leftist attitudes. To make a claim about correlation, as you are trying to do, you have two options:
Show a correlation or lack of correlation within this block,
Show a correlation or lack of correlation between average education level and average political leanings, using voting blocks as data points.
I have a perfect analogy:
I claim that smoking will decrease your life expectancy.
You claim that the smoking rate in the United States is 9.6%. The smoking rate in Japan is 16.4%. Life Expectancy in the United States is 79.8. Life Expectancy in Japan is 85.2. Therefore “we can see very definitively that a higher rate of smoking does nothing to decrease the life expectancy of the country”.
No. That doesn’t show that. Because if you look at data within Japan, you will still see a lower life expectancy among the people who smoke, and if you look at data globally, using countries as data points, you will also see lower life expectancy among the countries that smoke the most.
When you take broad averages like this, you very often lose useful information. In this case, just being “educated” in the abstract carries a weak correlation to political beliefs.
That’s true, but unfortunately, once you started trying to create a more complex model, you started making mistakes. And you aren’t even an idiot. Imagine your dumbest neighbors trying to make sense of this conversation.
No. I start off by giving you one variable correlated with one other variable. Once someone understands that, THEN we can start talking about nuance.
What does Japan have that America lacks, which might give them better long term health outcomes and a higher life expectancy?
Therefore “we can see very definitively that a higher rate of smoking does nothing to decrease the life expectancy of the country”.
Not relative to a public universal health care system, no.
If your diagnosis is “we need to cut cigarette use to save lives”, but we’re ignoring the very successful early medical interventions against cancer that save lives, then we’ve traded actual useful policy for This One Neat Trick that clearly doesn’t deliver the intended results.
Not relative to a public universal health care system, no.
I’m saying not at all. With public universal health care, without public universal health care… it does not matter. There is a negative correlation between smoking and life expectancy.
If your diagnosis is “we need to cut cigarette use to save lives”,
Yes, that is a conclusion one could draw from this.
but we’re ignoring the…
For the purposes of discussing this correlation, it does not matter. If we have universal health care, then cutting cigarette use will save lives. If we do not have universal health care, then cutting cigarette use will save lives. Conclusion: cutting cigarette use will save lives. That is true. No amount of discussion about unrelated third variables will make that not true.
If a third variable makes the model better, go ahead and add a third variable. But if someone doesn’t even understand the two-variable model, we should make sure they understand THAT model before we add even more stuff that they don’t understand.
With public universal health care, without public universal health care… it does not matter.
It matters enormously. Literally the difference between surviving and not surviving.
You know, the thing you’re measuring for.
But if someone doesn’t even understand the two-variable model, we should make sure they understand THAT model before we add even more stuff that they don’t understand.
Not for the purpose of establishing whether there is a negative correlation between smoking and life expectancy.
Also that comic is saying that adding a third variable does not always make the model better. Also p-hacking is a thing. It’s the opposite of what you’re trying to establish, which is the existence of a third variable which improves the model. And once again, I will say that if you find two correlated variables confusing, adding a third variable is NOT going to help you.
Mormons have more bachelor degrees than average: Yes.
Mormons are more conservative than average: Yes.
We can see very definitively that a higher-than-average rate of education does nothing to increase the partisan character of the voting block: No.
That’s not what this shows. The two facts, together, are not enough to show a correlation between education and conservatism or to show a lack of correlation between education and liberal/leftist attitudes. To make a claim about correlation, as you are trying to do, you have two options:
I have a perfect analogy:
I claim that smoking will decrease your life expectancy.
You claim that the smoking rate in the United States is 9.6%. The smoking rate in Japan is 16.4%. Life Expectancy in the United States is 79.8. Life Expectancy in Japan is 85.2. Therefore “we can see very definitively that a higher rate of smoking does nothing to decrease the life expectancy of the country”.
No. That doesn’t show that. Because if you look at data within Japan, you will still see a lower life expectancy among the people who smoke, and if you look at data globally, using countries as data points, you will also see lower life expectancy among the countries that smoke the most.
https://worldpopulationreview.com/country-rankings/smoking-rates-by-country https://worldpopulationreview.com/country-rankings/life-expectancy-by-country
That’s true, but unfortunately, once you started trying to create a more complex model, you started making mistakes. And you aren’t even an idiot. Imagine your dumbest neighbors trying to make sense of this conversation.
No. I start off by giving you one variable correlated with one other variable. Once someone understands that, THEN we can start talking about nuance.
What does Japan have that America lacks, which might give them better long term health outcomes and a higher life expectancy?
Not relative to a public universal health care system, no.
If your diagnosis is “we need to cut cigarette use to save lives”, but we’re ignoring the very successful early medical interventions against cancer that save lives, then we’ve traded actual useful policy for This One Neat Trick that clearly doesn’t deliver the intended results.
I’m saying not at all. With public universal health care, without public universal health care… it does not matter. There is a negative correlation between smoking and life expectancy.
Yes, that is a conclusion one could draw from this.
For the purposes of discussing this correlation, it does not matter. If we have universal health care, then cutting cigarette use will save lives. If we do not have universal health care, then cutting cigarette use will save lives. Conclusion: cutting cigarette use will save lives. That is true. No amount of discussion about unrelated third variables will make that not true.
If a third variable makes the model better, go ahead and add a third variable. But if someone doesn’t even understand the two-variable model, we should make sure they understand THAT model before we add even more stuff that they don’t understand.
It matters enormously. Literally the difference between surviving and not surviving.
You know, the thing you’re measuring for.
Not for the purpose of establishing whether there is a negative correlation between smoking and life expectancy.
Also that comic is saying that adding a third variable does not always make the model better. Also p-hacking is a thing. It’s the opposite of what you’re trying to establish, which is the existence of a third variable which improves the model. And once again, I will say that if you find two correlated variables confusing, adding a third variable is NOT going to help you.