Starting a new thread to shed more light on this topic. The original thread (https://texags.com/forums/16/topics/3358357) stirred my curiosity, and I dug into it deeper. First thing was to look closer at the source data from the CDC for the year 2020. If you try to drill down into a state to understand it better, it became obvious there are holes in the data.
Texas for example, has 254 counties, but the CDC only listed the data for 30 counties. The other 224 were "suppressed" due to "statistical imperfections." Their totals however, were lumped into the state total, and you can calculate their aggregate statistics by simple math. By being "suppressed", you could be led to believe those 224 counties are like the 30. But not so. They had a lower murder rate than the Texas average, and unsurprisingly most of them are rural, red counties.
I began to smell a rat with cherry-picked analysis from the authors in the original thread. Since the state total data is the only thing reliable, I focused there. A huge flaw in the original approach is all Biden and Trump states were thrown together as if the same. For example, a state that voted 50.1% Biden was treated the same as one that voted 70%. Same for the Trump category. Common sense tells you that you can't lump things together like that and assume they're the same.
So I used the source data to calculate "% Biden voters" and "% Trump voters" by state to be more precise and understand if there's a correlation with murder rates. There is not, and the graphs below show it. Each has an R-square of essentially zero, which means no correlation.
This means you can't take who won the state for President and make broad sweeping assumptions about how that relates to murder rates in the state, because there is no relationship there. States are very large populations with a very diverse group of people, cultures, and behaviors. You have to dig deeper. The authors in the original thread carefully found a set of group statistics to make a claim that is unsupported by the facts.
I looked into cultural issues and one jumped out… the percentage of babies in a state that are born out of wedlock. The graph below shows the relationship. The R-square of 68% is high and suggests that roughly two thirds of the murder rate difference between states is connected in some way to the percentage of babies born without both parents present.
Makes sense to me, as kids that grow up without both parents to insist on a good education and staying out of trouble are at a huge disadvantage and more likely to follow a lifestyle of crime. Doesn't mean there aren't single parents who do a great job because there are but they are an exception to the trend.
Interestingly, you can point directly at Democrats' social policies (generational welfare, abortion, soft on crime, etc) as contributors to a culture that doesn't value life, hard work, or personal responsibility. That's where the conversation needs to focus, instead of deflecting away to who-voted-for-who.
Texas for example, has 254 counties, but the CDC only listed the data for 30 counties. The other 224 were "suppressed" due to "statistical imperfections." Their totals however, were lumped into the state total, and you can calculate their aggregate statistics by simple math. By being "suppressed", you could be led to believe those 224 counties are like the 30. But not so. They had a lower murder rate than the Texas average, and unsurprisingly most of them are rural, red counties.
I began to smell a rat with cherry-picked analysis from the authors in the original thread. Since the state total data is the only thing reliable, I focused there. A huge flaw in the original approach is all Biden and Trump states were thrown together as if the same. For example, a state that voted 50.1% Biden was treated the same as one that voted 70%. Same for the Trump category. Common sense tells you that you can't lump things together like that and assume they're the same.
So I used the source data to calculate "% Biden voters" and "% Trump voters" by state to be more precise and understand if there's a correlation with murder rates. There is not, and the graphs below show it. Each has an R-square of essentially zero, which means no correlation.
This means you can't take who won the state for President and make broad sweeping assumptions about how that relates to murder rates in the state, because there is no relationship there. States are very large populations with a very diverse group of people, cultures, and behaviors. You have to dig deeper. The authors in the original thread carefully found a set of group statistics to make a claim that is unsupported by the facts.
I looked into cultural issues and one jumped out… the percentage of babies in a state that are born out of wedlock. The graph below shows the relationship. The R-square of 68% is high and suggests that roughly two thirds of the murder rate difference between states is connected in some way to the percentage of babies born without both parents present.
Makes sense to me, as kids that grow up without both parents to insist on a good education and staying out of trouble are at a huge disadvantage and more likely to follow a lifestyle of crime. Doesn't mean there aren't single parents who do a great job because there are but they are an exception to the trend.
Interestingly, you can point directly at Democrats' social policies (generational welfare, abortion, soft on crime, etc) as contributors to a culture that doesn't value life, hard work, or personal responsibility. That's where the conversation needs to focus, instead of deflecting away to who-voted-for-who.