13% of the population commit 53% of US murders

[Epikur]

@Forestluv

I need more pictures

like this:
 

 

[ajasatya]

@Epikur My point is: study more.

Those number are dangerous, yes, not because of the truth behind them, but because even the smartest ones among lay people do not know how to make proper sense of them. Let me come up with an example to clarify what I mean by the need of profound understanding of selection biases and causal inference.

Let's suppose we're trying to check whether race is a predictor of violence or not.

The first assumption I'd come up with: "social strata affects how people behave and place themselves on the spectrum of violence". This is very likely to be true, but if we want to show it numerically, we can run a hypothesis test and compute the p-value. Lots of reading required just for this.

Okay, now we have a problem. Black people, for instance, have a long history of oppression so the poorest strata of society might be skewed towards them. Again, we can check this hypothesis.

If it comes out to be true, then we will need to control our experiment by social stratum. What does it mean? It means that we can't just sample random people carelessly, but rather in a way that our populations of black vs non-black people are equally represented across social strata. Here lies another challenge: how do we sample those groups properly? What are the pros and cons of each sampling method? The study of sampling techniques, alone, requires a few months of intense study to master.

Alright, we have your samples controlled by social stratum and we (hopefully) haven't messed up anything until now. Then what about age? Maybe the population of non-black people is relatively skewed towards more advanced ages, where people are usually less violent in general?

Then what about degrees of education? You see? Every way to dissect our population that potentially affects how violent people are adds a whole new level of complexity. The pursuit of the perfect samples, in which our phenomenon does not manifest in a biased way, is extremely complex.

Now, finally, if we haven't messed up until now, which is highly unlikely unless we're very skilled statisticians or maybe just a brilliant mathematicians, which I am not, it's time to run our regression. And this last step also requires serious studies.

In the end, our results will depend on the choices we made. Which predictors did we use? Which sampling techniques? Which regression model? Which source of data? What is the time frame of our data?

There will be many more questions to answer, like: how do we make sure that we didn't select data from a bad time frame? How trustworthy is our data? Etc etc.

[Forestluv]

@ajasatya Would this be a correct hypothetical for one of the dilemmas you point to?

Let’s say a poor neighborhood is 90% black and 10% white. A rich neighborhood is 10% black and 90% white.

If we take a representative sample from each group it would be: Poor group 90% B : 10% White vs Rich Group 10% B : 90% white, it’s not a completely fair comparison because composition the two groups are unequal. Yet if we equalize the two groups such as 90% B : 10% W, it is not a fair comparison in the sense that the two groups no longer representative of the original populations. Each method would have value in one context, yet would be biased in another context. . . We would have a similar dilemma if we added in things like education level, obesity frequency etc. 

[ajasatya]

8 minutes ago, Forestluv said:

@ajasatya Would this be a correct hypothetical for one of the dilemmas you point to?

Definitely, you got it. I am not surprised though, since you're a professor .

And just to give a glimpse of how the complexity of new variables stacks up:

In a very simplistic approach, we may divide the population in two social strata. If, then, we want to divide the same population in two groups of education levels, that would make up for 4 groups! Two times two.

If, instead, on the pursuit of a better granularity, we divide each dimension in 3 groups, we would end up with 9 groups! Three times three.

And this is just for two variables!

[Epikur]

@ajasatya

Well what you are saying is nothing new in principle. You can make it even more tricky and say science is just historical patterns. There is no "law". We think that a pattern will be a certain way in the future but our "prove" is only the past. 

Anyway the thing is we have to  make actions, judgments, theories with missing variables. So we must do rely on intuition, automated procceses, in our habits and so on.

These numbers are not that much sloppier than the statistics that are used from the left imho.




 

[Preety_India]

9 hours ago, Consept said:

These threads are simple it comes down to 2 scenarios 

1. Black people are inherently more violent and aggressive. If you believe that poverty is the main cause then black people are inherently lazy and dont want to work or better themselves. Either theres something inherent in them that causes the stats we're seeing 

2. There are complexed socio-economic and historical issues that have led and contributed to the situation that black people find themselves in. 

If you present these stats as the be all and end all (not saying op is doing that), then you subscribe to number 1. If youre interested in why this is the case and want to get to the route of it then most likely you subscribe to number 2. I think if youre making an argument with these stats you should be clear what scenario you are arguing for. 

Those stats are not true. Just consider this much. 10 years ago they had the same stats. 10 years later that is now, they still keep pedaling the same stats. 

How can stats remain the same over a 10 year period and more.. No change at all, lesser or greater? This means the stats are bogus. 

I'm clever enough. I don't need to be a statistician to know that these stats are defective. It's just logic and common sense. 

Don't believe such stats. They have become like a broken record that keeps playing. 

 

[ajasatya]

21 minutes ago, Epikur said:

These numbers are not that much sloppier than the statistics that are used from the left imho.

People in general do not know statistics, regardless of being from the left or the right.

Let them who do not know be. Be not like them.

And drop this ridiculous comparison, wandering around levels of solid sloppiness. This just harms the credibility in your words.

[Epikur]

10 minutes ago, ajasatya said:

People in general do not know statistics, regardless of being from the left or the right.

Let them who do not know be. Be not like them.

And drop this ridiculous comparison, wandering around levels of solid sloppiness. This just harms the credibility in your words.

Life is sloppiness and messyness. You have to work with what you have. 

I think people know statistics the way they know math. If it works in their practical life it's ok. 

Btw. do you believe in these numbers? What do they mean to you? Did you see the video about it from Vaush and what is your take on it? Vaush studied sociology and seems confident with statistics.
 

[Epikur]

22 minutes ago, Preety_India said:

Those stats are not true. Just consider this much. 10 years ago they had the same stats. 10 years later that is now, they still keep pedaling the same stats. 

How can stats remain the same over a 10 year period and more.. No change at all, lesser or greater? This means the stats are bogus. 

I'm clever enough. I don't need to be a statistician to know that these stats are defective. It's just logic and common sense. 

Don't believe such stats. They have become like a broken record that keeps playing. 

 

I should believe you because you said so?

That reminds me of flat earthers. They have also some "prove". Do you take them serious?

 

[Preety_India]

@Epikur  I have stated the logic 

Btw nobody can change what you want to believe. 

Your belief also doesn't make much sense because it doesn't solve anything. You can't get anything done clutching onto a statistic forever. 

True leaders move beyond numbers and bring actual change.

What will you do with those numbers? 

 

[Epikur]

3 minutes ago, Preety_India said:

@Epikur  I have stated the logic 

Btw nobody can change what you want to believe. 

Your belief also doesn't make much sense because it doesn't solve anything. You can't get anything done clutching onto a statistic forever. 

True leaders move beyond numbers and bring actual change.

What will you do with those numbers? 

 

It's part of my filtering process. People who can deal with tabu subjects against their interests get a special trust from me. Trust is important to build something of value. 

I get to discuss the subject and like ajasatja said try to reduce my sloppiness a bit. 

If you are emotionally invested in a subject it's mostly ego I guess. That means we have always this bias. That is why I want to actively look at the opposite perspective of what I have and try to find people who try that too.

It's the hegelian dialectic process I guess. 

 

[Preety_India]

@Epikur  I have stated the logic. I don't know what more you are asking for. 

I'm not emotionally invested in it. 

I have used my own thinking and come to the conclusion that it's bogus

 

If you don't agree, that's fine. I stand with my opinion 

[Forestluv]

@ajasatya I’m at a beginner level of statistics. ANOVA is as high as I go. I’m interested how a statistical analysis can reveal an insight in one context, yet it may be limited to that context and biased in other contexts. I usually just do basic stats, yet if I enter more complexity I try to be careful in understanding what it’s revealing and it’s limits. Sometimes, I need assistance from a statistician about what I am allowed to conclude from a particular analysis and the caveats. And with the more complex stuff, I need to find someone I trust to translate it for me. It’s almost like a foreign language. I know enough basic phrases to get around, yet once it gets into complex grammar and different dialects I need someone I trust who is fluent in the language to help translate for me. 

[Epikur]

1 minute ago, Preety_India said:

@Epikur  I have stated the logic. I don't know what more you are asking for. 

I'm not emotionally invested in it. 

I have used my own thinking and come to the conclusion that it's bogus

 

If you don't agree, that's fine. I stand with my opinion 

It's not logic because you say so. I would say it seems to you it's logic. 

It looks to me that you are emotionally invested and biased. 

It's obvious we disagree.

Take it easy life is an illusion anyway.
 

[Leo Gura]

Getting sick of even entertaining these racist talking points.

[ajasatya]

8 minutes ago, Epikur said:

Life is sloppiness and messyness. You have to work with what you have. 

I think people know statistics the way they know math. If it works in their practical life it's ok. 

Btw. do you believe in these numbers? What do they mean to you? Did you see the video about it from Vaush and what is your take on it? Vaush studied sociology and seems confident with statistics.

Saying that "life is sloppy" to justify deliberate (or even unintentional) lack of depth of analysis where significantly less sloppiness is achievable is a shame.

Math, as in algebra or arithmetic, is much less prone to drastic errors. It's very different from statistics, in the sense that statistics is full of counter-intuitive concepts. I'd say that people get to know enough math so that it works in their lives, but they get to know enough statistics so that it's convenient for them, and it opens room for many sorts of devilry, so watch out.

I don't have strong reasons to doubt those numbers. What they mean to me is that the society in the US is marked with severe inequality of opportunities.