Multiple regression (what people usually have in mind when talking about controlling for different variables) is actually not a very good tool for causal inference in social science. It's much better than raw correlations, of course, but it still doesn't really work. You can never be sure that you're controlling for everything you need to control for, and often controlling for things you shouldn't control for can cause problems, too. Multiple regression has produced a lot of really bad social science papers.
Really you need a find and exploit a source of exogenous variation in the variable of interest. This is a lot harder, but it gives much stronger evidence than multiple regression. Economists have made a lot of progress in developing techniques for doing this over the past few decades.
I've been wondering about this myself. All the evidence we have from twin and family studies suggests that genetics contributes much more than upbringing to variation in adult traits and outcomes. It's quite possible that single parenthood is just a proxy for heritable traits that influence outcomes, rather than a major causal driver of outcomes.
We aren’t going back to sharing rooms and even new apartments aren’t built for each kid to have their own room.
What do you mean? Three- and four-bedroom apartments have been around for a long time.
Modern American cities have two major shortcomings. First, you can't have a private yard. This is a problem inherent to high density. Parks kind of make up for this, but it's not quite the same, and also, making a segue to the next major problem, they're full of drug addicts.
More generally, cities are full of criminals and drug addicts. As anyone who's ever been to a major East Asian city can tell you, this isn't inherent to city life. It's a policy choice. Voters in American cities have made a deliberate choice to tolerate crime and public drug abuse, with predictable results.
I get that for logistical reasons there are some regulations that need to be uniform, but there are many that do not need to be, as evidenced by the fact that municipalities have any regulatory power at all.
The minimum wage is an obvious example: There's no reason certain municipalities couldn't have a minimum wage lower than the $15.50 statewide minimum. California's recent ban on pork produced in certain ways is another example. There's absolutely no reason municipalities shouldn't be able to override that if they choose, but California's state government has stripped them of the authority to do so.
Democratic states like California strip municipalities of authority all the time. When California imposes a regulation, municipalities are stripped of the ability to choose not to impose that regulation. When California raises taxes, municipalities are stripped of the ability to choose to have low taxes and spending. When California raised the minimum wage to $15.50, this stripped municipalities of the ability to choose to have a lower minimum wage.
There's a major asymmetry in the framing here, where a state imposing government interventions statewide (and thus prohibiting municipalities from not having those interventions) is seen as totally legitimate, while a state preventing municipalities from imposing interventions is seen as an anti-democratic assault on the right of self-government.
If you're in favor of more taxes and more regulation, this framing is very convenient, because you get four legitimate chances to get the taxes and regulations you want: Federal, state, county, and city.
If you want to stay that states should allow more local control, fine. But be consistent about it. If municipalities should be allowed to add more government control, then they should also be able to subtract it.
It also had a slightly below average suicide rate. It does have a demographic advantage there, with the high Hispanic population; on the other hand, this is a slight handicap when it comes to homicide.
Well, if “WE” means “Liberals” the answer is a slew of red flag laws, registration and licensing proposals and improved mental health care.
And releasing repeat violent offenders, giving them as many chances as it takes to kill someone.
This is a bipartisan problem: The right won't accept crime-control measures that burden law-abiding people, and the left won't accept crime-control measures that burden criminals.
ThTh3 is sophistry. The basic point the meme is making is completely correct. This strikes a nerve, so people want to say it's wrong; since it isn't, they pretend that it's saying something else, and critique that instead. The point is that while the media obsessively focus on white-on-black violence and promote the idea that it's a major threat to black people, while black-on-white violence is an order of magnitude more common.
Yes, intraracial violence is much more common, but the chart is explicitly about interracial violence, and how the media chooses to exaggerate the prevalence of one kind while ignoring other kinds. And I can't help noticing that right up until this meme went viral, the left was saying that talking about black-on-black homicide was a racist distraction from the real problem, white-on-black homicide.
Responding to a couple of specific points:
Tweets 6 and 7: It makes no sense to adjust for population. If non-Hispanic whites are 60% of the population and blacks are 15%, and they both offend at the same rate and choose victims at random, then the rates of white-on-black and black-on-white offending will be 0.6 * 0.15 and 0.15 * 0.6, both equal to 9%. If we assume a same-race bias such that 80% of crime is same-race, and the other 20% is random-victim, then both should be 0.2 * 0.09 = 4.5%. If a fixed percentage of people of each race specifically target the other race for violence, then we should expect to see much more white-on-black crime, because the number of X-on-Y crimes depends only on the population of X and not at all on Y. The data shown in the original chart cannot be explained away by unequal populations.
The caption doesn't make it clear, but the chart in tweet 7 (from the report "Race and Hispanic Origin of
Victims and Offenders, 2012-15") is adjusting for the population of the victim's race. Figure 4 in the same report shows absolute numbers, which reveals that for 2012-15, white-on-black crime was about a sixth as common as the reverse. The correct interpretation of that chart is that interracial violence poses a similar threat to both black and white people, not that white and black people commit interracial violence at similar rates.
If you switched it to use the population of the race of the offender as the denominator, the chart would show a huge gulf between black-on-white and white-on-black, but that wouldn't be appropriate, either. I believe that that's what that (incorrect) viral chart about interracial homicide from a couple of years ago was doing.
The study linked in tweet 10 was published in 2014, with data from 2008-2012. It may have been true that in those years the media downplayed white-on-black crime while exaggerating black-on-white crime, but a) this isn't actually contradicted by the excerpts he posted, and b) this was before the BLM movement, and before the media went all in on gaslighting us with the new blood libel.
I agree that interracial violence is not a major threat to white Americans right now. But it's at least as much of a concern as it is for black Americans. I don't want the media to exaggerate black-on-white violence---I want them to stop lying about white-on-black violence.
And I want people to take crime seriously. Since 2015, half of the progress in reducing homicide rates since the early 90s has been reversed. In some sense we're all in this together, but black people are bearing a wildly disproportionate share of the impact, well over half of all homicides. The people who are hurt most by black crime denialism and the idea that black overrepresentation in prison is inherently racist are the law-abiding black people who have to deal with it in their neighborhoods. Whitey just gets the spillover.
Good response from Richard Hanania here: https://www.richardhanania.com/p/interracial-crime-and-perspective
There's a common swindle pulled by anti-gun activists, where they compare gun deaths in the US, or in states with high gun ownership, to gun deaths in countries or states with low gun ownership. Despite the obvious fallacy, I see this all the time. Much more often than I see comparisons of total homicides and suicides.
The thing is, we don't care about gun suicides or gun homicides as such. We care about suicides and homicides. The US doesn't have 250 times as many suicides as the UK overall. It has about ten times as many: A fivefold increase because the US has five times the population, and a twofold increase because the US has roughly twice the per-capita suicide rate.
Twice the suicide per capita is still pretty bad, but you understand that 250 is two full orders of magnitude greater than two, right? Seriously, just don't do this. It's intellectually trashy. That aside, there are two things you have to take into account here:
1. The UK has a relatively low suicide rate by European standards. I don't know why this is, but this is not the norm.
2. As you can see in the chart above, the US currently has a relatively high suicide rate compared to other wealthy countries, but in 2000 it was at the low end. Over the past twenty years, suicide has risen in the US, while falling in many European countries (also some good news from Japan, which had long been notorious for high suicide rates).
It's not entirely clear why suicide rates have increased over the last 20 years in the in the US. Gun ownership has been stable or declining over that period. Socioeconomic explanations, popular for ideological reasons, run into the inconvenient fact that the suicide rate has been steadily rising through good times and bad. Note also in the linked chart that Greece has maintained a rock-bottom suicide rate throughout the implosion of their economy.
My pet theory is opioid abuse, which has been rising over that time period and is a known risk factor for suicide, but this is speculative. One point of evidence in favor of this is that the black suicide rate in the US, traditionally much lower than the white suicide rate, has been climbing over the past few years in tandem with the increase in their opioid overdose rates.
The fifth and six lines, “Here in this room” and “martyrdom of fire” in particular, give me the kind of cringe that usually follows twenty to thirty seconds after a groomsman announces that he’s written a poem for the bride and groom.
Was it cliche when he wrote it, or is this like the guy complaining about Shakespeare just chaining a bunch of quotes together?
Pro Publica has a long history of manufacturing scandals out of whole cloth. Without having really dug into the details myself, if the Biden and Obama appointees are backing Thomas, I'm inclined to believe that this is another such case.
High inflation alone does not necessarily lead to falling real wages, so even if inflation were really bad it might not meaningfully affect affordability of goods and services. It depends on whether inflation is driven by excessive growth in the money supply (wages rise along with prices) or by negative supply shocks (wages don't keep up).
On a related note, the advance estimate for Q1 GDP came out yesterday, and it looks like quarterly nominal GDP growth is down to an annualized 5%. While that's not quite where we want to be, especially given the mediocre real GDP growth, it's getting close and shows continued progress on controlling monetary inflation.
Also, here's a chart of real wage growth:
https://fred.stlouisfed.org/graph/?g=12VmE
Populists keep claiming that wages haven't kept up with inflation, and it's just not true.
This is the Yangtze River. In Mandarin it's called Chángjiāng, where cháng (長) means long and jiāng (江) means river. For some reason, possibly related to regional dialects, ji was sometimes romanized as ki in the past, e.g. Beijing => Peking, Nanjiing => Nanking.
So why's it called the Yangtze river? As its other name suggests, it's a very long river, and people in different areas gave it different names; Yangtze (揚子) is the name that stuck in English.
Anyway, "long Kiang" is a mistranslation; the "long" is part of the river's proper name, not just a descriptor.
Also, "kojin" here (故人; gùrén in Mandarin) is a common noun meaning old friend; in the original Chinese title, he's referred to by name as 孟浩然 (Mèng Hàorán), another famous poet. Kokakuro is 黃鶴樓 (Huánghèlóu), the Yellow Crane Tower. If you look it up in Wikipedia, you can find a more accurate translation.
It's not sustainable. We learned back in the 70s that there's no long-run trade-off between inflation and unemployment. We cannot keep unemployment permanently low by keeping inflation permanently high, and the result of trying to do so is that we get recessions anyway, and have high inflation on top of that.
The Fed doesn't want to slow real wage growth. High real wage growth is good. High nominal wage growth is bad, because consistently high nominal wage growth is incompatible with 2% inflation. If the amount of money people have to spend on things is growing at 6% per year and the supply of things to spend money on is only growing at 2% per year, you're going to get 4% inflation, give or take.
Also, lying is the least harmful form of misinformation, because it's easier to call out. Prestige media promote false narratives through selective presentation of "the truth...and nothing but the truth." No need to lie. You just lead your audience down a winding path through the parts of the truth consistent with the narrative and away from the parts that contradict it.
This is just as effective as lying, but much more resistant to debunking. "That's technically true, but when you account for X, Y, and Z, you'll see that the full story is very different from what you've been led to believe" is a much less effective rebuttal than "That's a lie and here's the proof."
On “Open Mic for the week of 5/22/2023”
The difference between good racism and bad racism is something you should learn in the first week of DEI 101.
On “TSN Open Mic for the week of 5/15/2023”
Multiple regression (what people usually have in mind when talking about controlling for different variables) is actually not a very good tool for causal inference in social science. It's much better than raw correlations, of course, but it still doesn't really work. You can never be sure that you're controlling for everything you need to control for, and often controlling for things you shouldn't control for can cause problems, too. Multiple regression has produced a lot of really bad social science papers.
Really you need a find and exploit a source of exogenous variation in the variable of interest. This is a lot harder, but it gives much stronger evidence than multiple regression. Economists have made a lot of progress in developing techniques for doing this over the past few decades.
"
I've been wondering about this myself. All the evidence we have from twin and family studies suggests that genetics contributes much more than upbringing to variation in adult traits and outcomes. It's quite possible that single parenthood is just a proxy for heritable traits that influence outcomes, rather than a major causal driver of outcomes.
"
What do you mean? Three- and four-bedroom apartments have been around for a long time.
Modern American cities have two major shortcomings. First, you can't have a private yard. This is a problem inherent to high density. Parks kind of make up for this, but it's not quite the same, and also, making a segue to the next major problem, they're full of drug addicts.
More generally, cities are full of criminals and drug addicts. As anyone who's ever been to a major East Asian city can tell you, this isn't inherent to city life. It's a policy choice. Voters in American cities have made a deliberate choice to tolerate crime and public drug abuse, with predictable results.
"
https://twitter.com/LP_CLC/status/1659562485794365440
"
Was it Phil who said that if a black man did something like this he would be killed by cops?
"
I get that for logistical reasons there are some regulations that need to be uniform, but there are many that do not need to be, as evidenced by the fact that municipalities have any regulatory power at all.
The minimum wage is an obvious example: There's no reason certain municipalities couldn't have a minimum wage lower than the $15.50 statewide minimum. California's recent ban on pork produced in certain ways is another example. There's absolutely no reason municipalities shouldn't be able to override that if they choose, but California's state government has stripped them of the authority to do so.
"
Democratic states like California strip municipalities of authority all the time. When California imposes a regulation, municipalities are stripped of the ability to choose not to impose that regulation. When California raises taxes, municipalities are stripped of the ability to choose to have low taxes and spending. When California raised the minimum wage to $15.50, this stripped municipalities of the ability to choose to have a lower minimum wage.
There's a major asymmetry in the framing here, where a state imposing government interventions statewide (and thus prohibiting municipalities from not having those interventions) is seen as totally legitimate, while a state preventing municipalities from imposing interventions is seen as an anti-democratic assault on the right of self-government.
If you're in favor of more taxes and more regulation, this framing is very convenient, because you get four legitimate chances to get the taxes and regulations you want: Federal, state, county, and city.
If you want to stay that states should allow more local control, fine. But be consistent about it. If municipalities should be allowed to add more government control, then they should also be able to subtract it.
"
In 2021, Texas had a homicide rate almost exactly equal to the national average, lower than Maryland, Delaware, Illinois, and Michigan.
https://www.cdc.gov/nchs/pressroom/sosmap/homicide_mortality/homicide.htm
It also had a slightly below average suicide rate. It does have a demographic advantage there, with the high Hispanic population; on the other hand, this is a slight handicap when it comes to homicide.
"
Typical NOMSGs.
On “Mass Shooting Facts and Figures”
Well, if “WE” means “Liberals” the answer is a slew of red flag laws, registration and licensing proposals and improved mental health care.
And releasing repeat violent offenders, giving them as many chances as it takes to kill someone.
This is a bipartisan problem: The right won't accept crime-control measures that burden law-abiding people, and the left won't accept crime-control measures that burden criminals.
On “Thursday Throughput: Vaccines Vs Cancer Edition”
ThTh3 is sophistry. The basic point the meme is making is completely correct. This strikes a nerve, so people want to say it's wrong; since it isn't, they pretend that it's saying something else, and critique that instead. The point is that while the media obsessively focus on white-on-black violence and promote the idea that it's a major threat to black people, while black-on-white violence is an order of magnitude more common.
Yes, intraracial violence is much more common, but the chart is explicitly about interracial violence, and how the media chooses to exaggerate the prevalence of one kind while ignoring other kinds. And I can't help noticing that right up until this meme went viral, the left was saying that talking about black-on-black homicide was a racist distraction from the real problem, white-on-black homicide.
Responding to a couple of specific points:
Tweets 6 and 7: It makes no sense to adjust for population. If non-Hispanic whites are 60% of the population and blacks are 15%, and they both offend at the same rate and choose victims at random, then the rates of white-on-black and black-on-white offending will be 0.6 * 0.15 and 0.15 * 0.6, both equal to 9%. If we assume a same-race bias such that 80% of crime is same-race, and the other 20% is random-victim, then both should be 0.2 * 0.09 = 4.5%. If a fixed percentage of people of each race specifically target the other race for violence, then we should expect to see much more white-on-black crime, because the number of X-on-Y crimes depends only on the population of X and not at all on Y. The data shown in the original chart cannot be explained away by unequal populations.
The caption doesn't make it clear, but the chart in tweet 7 (from the report "Race and Hispanic Origin of
Victims and Offenders, 2012-15") is adjusting for the population of the victim's race. Figure 4 in the same report shows absolute numbers, which reveals that for 2012-15, white-on-black crime was about a sixth as common as the reverse. The correct interpretation of that chart is that interracial violence poses a similar threat to both black and white people, not that white and black people commit interracial violence at similar rates.
If you switched it to use the population of the race of the offender as the denominator, the chart would show a huge gulf between black-on-white and white-on-black, but that wouldn't be appropriate, either. I believe that that's what that (incorrect) viral chart about interracial homicide from a couple of years ago was doing.
The study linked in tweet 10 was published in 2014, with data from 2008-2012. It may have been true that in those years the media downplayed white-on-black crime while exaggerating black-on-white crime, but a) this isn't actually contradicted by the excerpts he posted, and b) this was before the BLM movement, and before the media went all in on gaslighting us with the new blood libel.
I agree that interracial violence is not a major threat to white Americans right now. But it's at least as much of a concern as it is for black Americans. I don't want the media to exaggerate black-on-white violence---I want them to stop lying about white-on-black violence.
And I want people to take crime seriously. Since 2015, half of the progress in reducing homicide rates since the early 90s has been reversed. In some sense we're all in this together, but black people are bearing a wildly disproportionate share of the impact, well over half of all homicides. The people who are hurt most by black crime denialism and the idea that black overrepresentation in prison is inherently racist are the law-abiding black people who have to deal with it in their neighborhoods. Whitey just gets the spillover.
Good response from Richard Hanania here: https://www.richardhanania.com/p/interracial-crime-and-perspective
On “Triple Terror In Texas”
There's a common swindle pulled by anti-gun activists, where they compare gun deaths in the US, or in states with high gun ownership, to gun deaths in countries or states with low gun ownership. Despite the obvious fallacy, I see this all the time. Much more often than I see comparisons of total homicides and suicides.
The thing is, we don't care about gun suicides or gun homicides as such. We care about suicides and homicides. The US doesn't have 250 times as many suicides as the UK overall. It has about ten times as many: A fivefold increase because the US has five times the population, and a twofold increase because the US has roughly twice the per-capita suicide rate.
Twice the suicide per capita is still pretty bad, but you understand that 250 is two full orders of magnitude greater than two, right? Seriously, just don't do this. It's intellectually trashy. That aside, there are two things you have to take into account here:
1. The UK has a relatively low suicide rate by European standards. I don't know why this is, but this is not the norm.
2. As you can see in the chart above, the US currently has a relatively high suicide rate compared to other wealthy countries, but in 2000 it was at the low end. Over the past twenty years, suicide has risen in the US, while falling in many European countries (also some good news from Japan, which had long been notorious for high suicide rates).
It's not entirely clear why suicide rates have increased over the last 20 years in the in the US. Gun ownership has been stable or declining over that period. Socioeconomic explanations, popular for ideological reasons, run into the inconvenient fact that the suicide rate has been steadily rising through good times and bad. Note also in the linked chart that Greece has maintained a rock-bottom suicide rate throughout the implosion of their economy.
My pet theory is opioid abuse, which has been rising over that time period and is a known risk factor for suicide, but this is speculative. One point of evidence in favor of this is that the black suicide rate in the US, traditionally much lower than the white suicide rate, has been climbing over the past few years in tandem with the increase in their opioid overdose rates.
On “POETS Day! Henry Wadsworth Longfellow”
The fifth and six lines, “Here in this room” and “martyrdom of fire” in particular, give me the kind of cringe that usually follows twenty to thirty seconds after a groomsman announces that he’s written a poem for the bride and groom.
Was it cliche when he wrote it, or is this like the guy complaining about Shakespeare just chaining a bunch of quotes together?
"
It wasn't because of his height.
On “TSN Open Mic for the week of 5/1/2023”
And "N times less."
I think it means 1/N times as much, but I'm not sure.
"
Pro Publica has a long history of manufacturing scandals out of whole cloth. Without having really dug into the details myself, if the Biden and Obama appointees are backing Thomas, I'm inclined to believe that this is another such case.
On “Of Beer, Ammo, and Wages”
High inflation alone does not necessarily lead to falling real wages, so even if inflation were really bad it might not meaningfully affect affordability of goods and services. It depends on whether inflation is driven by excessive growth in the money supply (wages rise along with prices) or by negative supply shocks (wages don't keep up).
On a related note, the advance estimate for Q1 GDP came out yesterday, and it looks like quarterly nominal GDP growth is down to an annualized 5%. While that's not quite where we want to be, especially given the mediocre real GDP growth, it's getting close and shows continued progress on controlling monetary inflation.
Also, here's a chart of real wage growth:
https://fred.stlouisfed.org/graph/?g=12VmE
Populists keep claiming that wages haven't kept up with inflation, and it's just not true.
On “POETS Day! Li Bai, Ernest Fenollosa, and Ezra Pound”
"The long Kiang"
This is the Yangtze River. In Mandarin it's called Chángjiāng, where cháng (長) means long and jiāng (江) means river. For some reason, possibly related to regional dialects, ji was sometimes romanized as ki in the past, e.g. Beijing => Peking, Nanjiing => Nanking.
So why's it called the Yangtze river? As its other name suggests, it's a very long river, and people in different areas gave it different names; Yangtze (揚子) is the name that stuck in English.
Anyway, "long Kiang" is a mistranslation; the "long" is part of the river's proper name, not just a descriptor.
Also, "kojin" here (故人; gùrén in Mandarin) is a common noun meaning old friend; in the original Chinese title, he's referred to by name as 孟浩然 (Mèng Hàorán), another famous poet. Kokakuro is 黃鶴樓 (Huánghèlóu), the Yellow Crane Tower. If you look it up in Wikipedia, you can find a more accurate translation.
On “TSN Open Mic for the week of 4/17/2023”
It's not sustainable. We learned back in the 70s that there's no long-run trade-off between inflation and unemployment. We cannot keep unemployment permanently low by keeping inflation permanently high, and the result of trying to do so is that we get recessions anyway, and have high inflation on top of that.
"
The Fed doesn't want to slow real wage growth. High real wage growth is good. High nominal wage growth is bad, because consistently high nominal wage growth is incompatible with 2% inflation. If the amount of money people have to spend on things is growing at 6% per year and the supply of things to spend money on is only growing at 2% per year, you're going to get 4% inflation, give or take.
On “Freddie Gets Practical, As In “Practical Majors””
If you can handle Fortran, you can do better than government pay scale.
On “Dominion Settles With Fox News”
Also, lying is the least harmful form of misinformation, because it's easier to call out. Prestige media promote false narratives through selective presentation of "the truth...and nothing but the truth." No need to lie. You just lead your audience down a winding path through the parts of the truth consistent with the narrative and away from the parts that contradict it.
This is just as effective as lying, but much more resistant to debunking. "That's technically true, but when you account for X, Y, and Z, you'll see that the full story is very different from what you've been led to believe" is a much less effective rebuttal than "That's a lie and here's the proof."
"
Also, couldn't it be trivially subverted by inviting on particularly dim-witted or uncharismatic person to take the opposing side?
On “Freddie Gets Practical, As In “Practical Majors””
You went to all that effort to set yourself up to blame WordPress, and then you just threw it away.
*Comment archive for non-registered commenters assembled by email address as provided.