Evaluation of statistical data.

[carson8]

Let's say I have the following statistical data:

9 out of 10 murders are committed by men.

Can someone please tell me, if all the phrases below accurately describe the statistical data above?

A- "People have 9 times more chances to get killed by a man than by a woman"
B- "The probability that a given randomly chosen homicide was committed by a man is 90%"
C- "One is 9 times as likely to be killed by a man as by a woman"
D- "If one is to be murdered, then one is 9 times as likely to be killed by a man as by a woman"

Thank you.

 

[hlsmith]

What is this for?

 

[carson8]

It's just about the correct way to speak about the statistical data in terms of probabilities (or opportunities). My opinion is there's a lot of variables and conditions involved in a murder, and so B and D somehow consider those variables to have a value, or those conditions to be met when they specify that: "a murder has been committed already" (B), or that there's certainty the murder will be committed, "If one is to be murdered" (D). Following this same line of thoughts, I think A and C are not correct because those phrases assume everybody has the same chances to be killed equally, or have the same chances, whether those variables and conditions are met or not, or are likely to be met or not.

The underlying argument is about race, which is always a topic, so let's say the real data is:

9 out of 10 murders where the victim is of race X are committed by men of the X race

If A and C are correct, then I can say.

"A person of race X has 9 times more chances to get killed by a man of race X than by a woman of race X"

And I think this is wrong because the phrase assumes all persons of race X share the same probabilities of being murdered.

So I would say it like this.

"If a person of race X is to be murdered, then that person is 9 times as likely to be killed by a man of race X as by a woman of race X"

This is the argument about, how to say things accurately.
Thank you

 

[hlsmith]

Another point is that you are extrapolating into the future on a murder that was not in the sample or prior population murders were sampled from. There is some risk in generalizing to the future, since things may change. Just a question, do you need to mess with the stat and infer, what is wrong with just stating, "9 out of 10 murders where the victim is of race X are committed by men of the X race"? Also, how many murders go unsolved or may not be deemed a murder, some doubt?

 

[carson8]

Another point is that you are extrapolating into the future on a murder that was not in the sample or prior population murders were sampled from. There is some risk in generalizing to the future, since things may change. Just a question, do you need to mess with the stat and infer, what is wrong with just stating, "9 out of 10 murders where the victim is of race X are committed by men of the X race"? Also, how many murders go unsolved or may not be deemed a murder, some doubt?

No, I don't need to mess with the data to get to that phrase, let's say the phrase describes the data accurately.
In my opinion, that's the only phrase ever to be used, without even trying to estimate the future. But if I have to do it, my question is if B and D are the right way to do so. Also, let's assume no case goes unsolved.

 

[hlsmith]

B seems alright. D, I would phrase it in regards to your sample of existing murders, not if you were to get murdered, as mentioned, things change, your sampling may be off, you didn't control for particulars, and who is you (Americans? from blank to blank, etc.).

Welcome to the forum.

 

[noetsi]

A critical point I think is there is no agreed on way to comment on statistics that I am aware of. There is no authority that decides the valid way to make such statements. You would likely find different program, different professors, and different journals give different answers.

I think all 4 statements are functionally the same although probability is something hlsmith knows better than I.

 

[carson8]

Thank you all