White racial violence problem?

[Michael Hardner]

...the independent variables include the different races as a percentage of the population, income per capita, the gini coefficient, life expectancy, average level of education + other factors that can explain differences in outcome in the crime rate between countries.

These are not independent variables. Income and education levels, for example. This is a 1st year student error.

[-1=e^ipi]

@Michael Hardner - I was talking in the context of a linear regression model, whether independent variable = explanatory factor. Income and education are correlated and cause each other, but that doesn't result in a biased estimate for the specific regression model I was suggesting.

You are playing semantics. I didn't invent the terminology, I just use it.

"In statistics, linear regression is an approach for modeling the relationship between a scalar dependent variable y and one or more explanatory variables (or independent variable) denoted X."

http://en.wikipedia.org/wiki/Linear_regression

Edited May 27, 2015 by -1=e^ipi

[Michael Hardner]

You are playing semantics. I didn't invent the terminology, I just use it.

Yes, but you don't understand the terminology is the point. It's actually a mathematical definition. "Playing semantics" with this is like saying 2+2=5....

As I suspected, you're part way through undergrad and have discovered something that you think explains the universe, but you probably understand it as much as the chimps who commandeered the spaceship in Return to the Planet of the Apes, with respect.

[-1=e^ipi]

Yes, but you don't understand the terminology is the point. It's actually a mathematical definition. "Playing semantics" with this is like saying 2+2=5....

I understand the meaning of independent variable in the context of what I said, you are the one who doesn't understand the meaning.

Look, if I do a linear regression, y = XB + error, where y, B and error are mx1 matrices and X is an nxm matrix, then X is the independent variable(s) and y is the dependent variable, regardless of what I choose y and X to be. See wiki link you math n00b: http://en.wikipedia.org/wiki/Dependent_and_independent_variables#Independent_variable

As I suspected, you're part way through undergrad

Incorrect, I finished my undergrad years ago.

[cybercoma]

Yes, but you don't understand the terminology is the point.

Euler is using the definition correctly. Independent variables are sometimes called explanatory variables or predictors. In statistics, we look at what effect they have on the dependant variable or outcome. That doesn't mean those variables have nothing influencing them and it doesn't mean that there may not be a problem using linear regression if the outcome also influences independent variables. Those are all separate issues though. The terminology that Euler is using is correct.

[Michael Hardner]

I understand the meaning of independent variable in the context of what I said, you are the one who doesn't understand the meaning.

Ah... the old "I know you are but what am I" argument... yes you have used this one before.

See wiki link you math n00b:

Thanks. A true expert would acknowledge his shortcomings. You committed an outright HOWLER, and yet you try to opine on these topics like you're a professor. Regression analysis is a fine tool in the hands of capable people, I will say that.

[Michael Hardner]

The terminology that Euler is using is correct.

Not if the variables aren't independent. These words mean things.

[Michael Hardner]

The terminology that Euler is using is correct.

http://www.statisticssolutions.com/assumptions-of-multiple-linear-regression/

"Multiple linear regression assumes that there is little or no multicollinearity in the data. Multicollinearity occurs when the independent variables are not independent from each other. "

Euler's use of "independent" variables for Education and Income, for example, would result in Multicollinearity.

http://blog.minitab.com/blog/adventures-in-statistics/what-are-the-effects-of-multicollinearity-and-when-can-i-ignore-them

Jim Frost:

"It refers to predictors that are correlated with other predictors in the model. Unfortunately, the effects of multicollinearity can feel murky and intangible, which makes it unclear whether it’s important to fix."

Multicollinearity is problem that you can run into when you’re fitting a regression model, or other linear model.

[-1=e^ipi]

Euler is using the definition correctly.

Thank you for backing me up.

"Multiple linear regression assumes that there is little or no multicollinearity in the data. Multicollinearity occurs when the independent variables are not independent from each other. "

Sigh,

http://en.wikipedia.org/wiki/Multicollinearity

"Multicollinearity does not reduce the predictive power or reliability of the model as a whole, at least within the sample data set"

Multicollinearity doesn't cause the OLS estimators to be biased, or the estimates of their errors to be biased. It doesn't cause hypothesis tests using OLS estimators to be invalid. Things like endogeneity do, but not multicollinearity.

Euler's use of "independent" variables for Education and Income, for example, would result in Multicollinearity.

Yes, yes it does. So what?

[Argus]

Together, Blacks and Hispanics committed 98% of all shootings in New York according to crime stats. Though you'd be hard pressed to see that reported in the media.

http://www.city-journal.org/2010/eon0514hm.html

According to FBI stats almost half of all murders in the US were committed by blacks.

http://conservative-headlines.com/2014/12/fbi-half-of-all-people-arrested-for-murder-in-2011-were-black/

Young Black men commit murder 14 times more often than young white men.

http://www.frontpagemag.com/2013/dgreenfield/time-young-black-men-murder-14-times-more-than-young-white-men/

But things are different in the UK, right? Well... not so much. Half of all violent crime and two thirds of shootings are committed by... blacks.

http://www.dailymail.co.uk/news/article-1290047/Metropolitan-Police-crime-statistics-reveal-violent-criminals-black--victims.html

Edited May 28, 2015 by Argus

[Michael Hardner]

Yes, yes it does. So what?

So you account for it in your model. I don't remember the ways, but there's a cross-product approach where you multiply them... I don't remember all the approaches as I took the course over 30 years ago. You'll take it next year, you can school me then.

[bush_cheney2004]

Do you mean sums of squares, cross products, analysis of variance (ANOVA/MANOVA), etc. ?

Fun stuff !!

[-1=e^ipi]

So you account for it in your model. I don't remember the ways, but there's a cross-product approach where you multiply them... I don't remember all the approaches as I took the course over 30 years ago. You'll take it next year, you can school me then.

You mean adding a quadratic interaction term? That doesn't 'solve' multicollinearity, which isn't even a problem.

[Bonam]

You don't need any fancy regressions for this stuff, you just need decent data, which for the most part is not allowed to be gathered, or, if gathered, is difficult to find/access.

[cybercoma]

You don't need any fancy regressions for this stuff, you just need decent data, which for the most part is not allowed to be gathered, or, if gathered, is difficult to find/access.

That's the bigger barrier: understanding the limitations of the data. Regressions do help us understand how various indicators are related to the outcomes though, but understanding how the data was gathered in order to run those outcomes and what that means for the models is key. For instance, there are a number of considerations with Uniform Police Statistics that need to be made. One large issue with them is that they don't measure actual crime, but only crime that comes to the attention of the police. Crimes that come to the attention of the police are in no small part determined by policing methods and coverage. Further still, we've seen how police will doctor reports for their benefit and how they disproportionately and more aggressively police poor and black neighbourhoods. There's less leniency given to people of colour as well, so they're more likely to be booked and become part of the statistics. For instance, white people are more likely to abuse drugs, but black people are incarcerated for drug-related offences at a rate that's 10x greater (News article on the study here). This is just one example and one small aspect of understanding the data, where it comes from, how it's gathered, and what it means before conducting any sort of statistical analysis with it.

Edited May 28, 2015 by cybercoma

[cybercoma]

http://www.statisticssolutions.com/assumptions-of-multiple-linear-regression/

"Multiple linear regression assumes that there is little or no multicollinearity in the data. Multicollinearity occurs when the independent variables are not independent from each other. "

Euler's use of "independent" variables for Education and Income, for example, would result in Multicollinearity.

http://blog.minitab.com/blog/adventures-in-statistics/what-are-the-effects-of-multicollinearity-and-when-can-i-ignore-them

Jim Frost:

"It refers to predictors that are correlated with other predictors in the model. Unfortunately, the effects of multicollinearity can feel murky and intangible, which makes it unclear whether it’s important to fix."

Multicollinearity is problem that you can run into when you’re fitting a regression model, or other linear model.

Multicollinearity doesn't change the name of the terms independent/dependent variables. It just changes the interpretation of the models. You're using "independent" in a colloquial sense and Euler is using "independent" in the statical analysis jargon sense. Consequently, you two are talking past each other. You may have a point about multicollinearity or endogeneity, but I didn't read far enough back in the thread to see what you two are specifically talking about. I'm just trying to point out that Euler is simply talking about terms on the right side of an equation when xe says "independent variables." It's only meant to distinguish predictors from outcomes. Calling these things independent variables simply means that they can add to our ability to more accurately predict a given outcome. What specific variables you guys are talking about, I don't know. Like I said, I haven't gone back through the thread and probably won't. I just wanted to point out that the terminology Euler is using is correct here, but also that it probably doesn't take away from the point that you're trying to make about the data itself and how there are possibly other confounding variables at play here that affect both the predictors (independent variables) and the outcomes (dependent variables). That would cause an endogeneity issue. This is what makes social statistics so difficult to capture and interpret.

[cybercoma]

Thank you for backing me up.

I'm not really sure what point you were making in all of this, i.e., what model you're using or what variables you're talking about. All I'm trying to do is point out that you guys are talking past each other and getting hung up on jargon. I think MH is using "independent" in a colloquial sense, whereas you're using it in a stats sense. You should meet him halfway and try to understand the point of what he's saying, rather than him misunderstanding the particular terminology you're using. I think what he's trying to get it is that the predictors may suffer from endogeneity due to confounding variables, but I'm not positive. To move the discussion forward, you should focus on what he's trying to get at, rather than the silly misunderstanding over terminology.

[Michael Hardner]

ok then, thanks for clearing up Cyber... I maintain that the variables he used in his example would need some further manipulation but my mathematics studies (pure statistics, mind you, not applied) are too dim to remember what that is.

Thanks again

[-1=e^ipi]

I think MH is using "independent" in a colloquial sense, whereas you're using it in a stats sense. You should meet him halfway and try to understand the point of what he's saying

I know what MH is referring to by using the word independent. MH is the one purposely misinterpreting me in an attempt to do a 'gotcha'. I suggested to Hernanday that his perception of higher white crime might be due to confirmation bias, and perhaps he should test his hypothesis by performing a regression where crime is the dependent variable, and various other factors such as race, income, education, etc. are independent variables. From there, Michael just kept trying to misinterpret me.

I think what he's trying to get it is that the predictors may suffer from endogeneity due to confounding variables, but I'm not positive.

Except he isn't talking about endogeneity, but multicollinearity. Michael's under the impression that you can't have 2 independent variables in a regression that are correlated with each other. And having 2 independent variables cause each other in a regression doesn't cause endogeneity.

To move the discussion forward, you should focus on what he's trying to get at, rather than the silly misunderstanding over terminology.

He's trying to do some sort of 'gotcha' by misinterpreting what I write.

[cybercoma]

He's trying to do some sort of 'gotcha' by misinterpreting what I write.

I get that you think that, which is why I elaborated. This isn't MH's style to play semantic "gotcha" games. He's literally the only poster on this forum who's genuinely interested in furthering his knowledge and facilitating debate in a reasonable way. You're attributing to malice something that was a simple misunderstanding.

[H10]

http://news.nationalpost.com/news/canada/white-calgary-mom-hopes-community-bbq-will-stop-her-family-from-being-racially-slurred

Racism is only a white problem.

Racism is a white problem.

http://www.thestar.com/yourtoronto/education/2015/05/29/black-teachers-still-face-racism-on-the-job-in-ontario.html

[H10]

@hernanday - I'm not particularly convinced that 'white people' have a disproportionately high rate of crime. Perhaps you should perform a linear regression where the dependent variable is the crime rate, and the independent variables include the different races as a percentage of the population, income per capita, the gini coefficient, life expectancy, average level of education + other factors that can explain differences in outcome in the crime rate between countries. Then you can test the hypothesis of if having a larger white population has a positive effect on the crime rate or not. If you can show that 'whiteness' has a positive effect on the crime rate at the 95% confidence level, then I will accept your claim (assuming the model used takes into account all of the relevant explanatory variables).

It doesn't matter if whites commit disproportionate crime. That is not the central claim of my thread. 12 pages and constant derailing by trolls who want to discuss everything but the thread title problem. I think there have been only 2 poster who answered. Its not about you agreeing or accpeting my claim, there is already substantial data to support this,

I'm not particularly concerned if you agree with the data or not. I am far more interested in learning why white people do these crimes in the first place. What are the cultural factors that lead to a some of the richest countries in the world like USA, UK, Australia and Canada, to having huge populations of whites being arrested and incarcerated in cultures that were largely constructed by whites. Whites make the rules of the game so to speak. You would not expect them to be the majority of jail, they make the rules, they are stopped by their own kind of police, tried by their own judges, and own juries and own white prosecutors.

Strangely most people want to duck the explanation

I get that you think that, which is why I elaborated. This isn't MH's style to play semantic "gotcha" games. He's literally the only poster on this forum who's genuinely interested in furthering his knowledge and facilitating debate in a reasonable way. You're attributing to malice something that was a simple misunderstanding.

thank you.

[Moonlight Graham]

Racism is a white problem.

Do you really mean that?

Edited May 29, 2015 by Moonlight Graham

[H10]

Do you really mean that?

No, I was just teasing him using satire. Racism is a complex system. It isn't only a white people problem. Some whites are not racist and actually have a good concept of what is going on like Tim Wise. There are black who embrace white supremacy and are ironically more racist than whites. In fact often the most effective white racism comes from using a black puppet.

[Argus]

No, I was just teasing him using satire. Racism is a complex system. It isn't only a white people problem. Some whites are not racist and actually have a good concept of what is going on like Tim Wise. There are black who embrace white supremacy and are ironically more racist than whites. In fact often the most effective white racism comes from using a black puppet.

The worst racists I've ever known in my life were Black and Arabic. The Blacks hated other Blacks from different regions. Ie, Caribbean Blacks despised African Blacks, and African Blacks felt contemptuous towards Caribbean Blacks. Asians tended to feel that all Blacks were basically subhuman, and Arabs tended to feel the same. The racism of these people was overt, matter of fact, and unashamed. Racism in non-White countries tends to be extremely violent. It's not the kind where security watches you when you come into a store. It's the kind where you get your head cut off and your house set on fire. In India, darker skinned individuals can be beaten to death for walking in the shadow of a higher caste individual. In places like Myannmar, Sri Lanka and Vietnam, minority members are driven out to sea while in Pakistan they're butchered in job lots.

Calling racism a White issue is idiocy.

Edited May 30, 2015 by Argus