Likelihood of Rape

[fls]

Yes, this refers to Elevatorgate. No, you don't have to know or care about it to answer the poll. :)

The set-up is that a one man and one woman are on an elevator at 4 in the morning in a hotel.

I posed this question to TraneWreck in another thread and I wanted to get some additional opinions. If you feel compelled to explain your response, please consider using spoiler tags. ;)

Thanks.

[Complexity]

Errr... what poll?

ETA: Must be the first or last option, because the second and third options are subsumed by the first.

I voted for the first option.

[fls]

Errr... what poll?

Sorry. You don't get to make the poll until after the thread has been started, so there's always a bit of a delay.

Linda

[ImaginalDisc]

So the assumption is that the rapist is the man?

[ehcks]

Maybe I'm an optimist, or maybe I hate this topic, but I don't see rape as likely in any of those.

Also, most elevators have security cameras. Terrible place to do anything illegal.

[blobru]

Not sure the distinction between the first choice -- "a woman is on an elevator with a man" -- and the next two? Is it:

1. (no conversation)
2. (conversation with proposition)
3. (conversation without proposition)?

[fls]

So the assumption is that the rapist is the man?

Yes.

Linda

[USEagle13]

Yes, this refers to Elevatorgate. No, you don't have to know or care about it to answer the poll. :)

The set-up is that a one man and one woman are on an elevator at 4 in the morning in a hotel.

I posed this question to TraneWreck in another thread and I wanted to get some additional opinions. If you feel compelled to explain your response, please consider using spoiler tags. ;)

Thanks.

Was pretty biased poll as 2 of the options were pretty much the same thing. Almost seemed like some sort of feminist type thingy.

[ImaginalDisc]

Yes.

Linda

Why?

[fls]

Not sure the distinction between the first choice -- "a woman is on an elevator with a man" -- and the next two? Is it:

1. (no conversation)
2. (conversation with proposition)
3. (conversation without proposition)?

There may or may not be conversation with the first option. You're just getting more information with the second and third option that you don't get with the first.

Linda

[fls]

Why?

You can answer as though either can be the rapist, if you prefer.

Linda

[Pup]

Must be the first or last option, because the second and third options are subsumed by the first.

What he said.

[ImaginalDisc]

You can answer as though either can be the rapist, if you prefer.

Linda

In that case I think. . .

. . .this poll will provide no useful information at all because it's chock full of misplaced assumptions and the opinions of a bunch of internet posters don't amount to a hill of beans in this crazy, mixed up world.

[bigred]

This is an inside joke I'm not up on, right? I mean you don't seriously think any of these minor changes in the circumstances make rape more or less likely? I've never come so close to voting for the planet X silliness.

[Complexity]

So the assumption is that the rapist is the man?


There is no need for such an assumption given the choices listed.

[fls]

This is an inside joke I'm not up on, right? I mean you don't seriously think any of these minor changes in the circumstances make rape more or less likely?

Some people think that it does make a difference, so I wondered what others thought.

I've never come so close to voting for the planet X silliness.

I don't know what it is, but for some reason, it's more satisfying to vote silly than just not vote. :)

Linda

[Pup]

There may or may not be conversation with the first option. You're just getting more information with the second and third option that you don't get with the first.

Linda


I'm taking back my agreement with Complexity. That would only be true if the question was about the number of rapes.

Man propositions woman means x% chance of rape.

Man doesn't proposition woman means y% chance of rape.

Man and woman on elevator is a weighted average of x and y. It will be higher than or equal to one and lower than or equal to the other, so it's actually the worst of the three choices, and can at best only tie for the greatest chance of rape.

I think.

This is worse than the Monty Hall problem.

[Garrette]

I'm taking back my agreement with Complexity. That would only be true if the question was about the number of rapes.

Man propositions woman means x% chance of rape.

Man doesn't proposition woman means y% chance of rape.

Man and woman on elevator is a weighted average of x and y. It will be higher than or equal to one and lower than or equal to the other, so it's actually the worst of the three choices, and can at best only tie for the greatest chance of rape.

I think.

This is worse than the Monty Hall problem.
Until recently my thought process would have been along these lines, though perhaps not quite so precise. But lately I am trying to make myself simplify my thought process and the best way for me to do that is sets. Viewing the question this way, the full set is "Man and Woman on Elevator." Subset is "Man Propositions Woman on Elevator." Second subset is "Man Does Not Proposition Woman on Elevator." Final, unstated subset, is "Man Rapes Woman." That final subset will include some of the first two subsets, but all three will be smaller than the full set.

[quarky]

Is this the new math?

I don't like elevators much; prefer the stairs.

But I don't like rape even more.

[fls]

Pup, to avoid introducing confusion between numbers and probability (or likelihood, etc.), all choices would be in reference to the same group - women who are riding in an elevator alone with a man.

Linda

[blobru]

It's the Sleeping Beauty Paradox (http://forums.randi.org/showthread.php?t=214200) again, quarky, except now Sleeping Beauty's merely sleepy, and Prince Charming ain't so charming (unless "creeper" means charming in Gaelic). Still, he rides to her rescue by sharing an elevator and offering her some "coffee". When she wakes up Monday, what are the odds that "coffee" meant coffee?

Sorry to derail, fls. I went with:
the third guy, only because he's the silent-type and you can't trust them silent-types; a bit glib, but I was thinking of the case where he might be planning to abduct her and doesn't want to draw attention to himself; honestly, wasn't quite sure how to reason about it... can i have a do-over? :confused:

[Cavemonster]

I voted for the second option.
The question is how does a proposition or the lack of a proposition effect a risk assessment at that moment.

All other things being unknown, any person is a possible rapist, in the same way they are a possible nobel laureate, stamp collector or mailman.

If I know nothing else about a person, and I learn that they can read, the probability of them being a nobel laureate increases. The mass of humans has been divided into two categories, the literate and the illiterate. One of those two groups is more likely to contain a nobel laureate than the other and our man is in that group. From the wider set, our chances have increased.

Let's return to the elevator. The most obvious information that a proposition imparts is that this man wants to have sex with this woman. Dividing the world into people who want to have sex with her and people who don't, which camp is more likely to contain the rapist? By definition, the rapist wants to have sex with her, (leaving aside whether rape is "about" sex) So this piece of information increases the likelyhood.

What else do we learn from the proposition? We learn that the man is willing to take some action to fullfill his desire for sex with her. Again, as compared to the camp who would take no action whatsoever, we've increased our chances.

Is there anything in a proposition that decreases the chances of a rape? I'm sure arguments can be made, but I honestly can't think of any with the strength and clarity of the two above that increase the chances, I'd be hay to hear them though.

Is there anything in the lack of a proposition that increases the chances? Again, not with the clarity of the increases above. People have mentioned a "silent type" but the fear of the rapist as mugger, coming out of the darkness to attack doesn't match the numbers. Of course, the greater number of rapes are by known people. and I can say anecdotally that the majority of rape reports I've read and heard that involved a waking, undrugged victim began with a "request".

Now the amount of the increases is still very small, and ultimately, unknown, but yes, I'd call it a net increase.

[Aepervius]

Cavemnonster you can use the same type of argument and lead to the contrary conclusion.

"man want to rape her not make love"
"man should not annoucne intention but take her by surprise otherwise an alarmed woman (alarmed by asking for sex) might start already grasping at mace or self defense when asked and thus is more difficult to surprise or overwhelm".
"conclusion : the 3rd option is more likely : the man which do not speak to the woman in the elevator"

PS: I don't think either of the reasonning is valid. Just saying that you are making argument out of wool cloth.

[JJM 777]

Elevator does not make a man a rapist, methinks.

First you need the rapist, then you need to have bad luck to end up alone with him while he feels safe to assault you. There are places that rapists seek, those would be the risky places, maybe a dimly lit park in the quiet hours of the night. Or the bedroom of an acquaintance who sometimes gives and sometimes not, today frustratingly not.

[Cavemonster]

Cavemnonster you can use the same type of argument and lead to the contrary conclusion.

"man want to rape her not make love"
"man should not annoucne intention but take her by surprise otherwise an alarmed woman (alarmed by asking for sex) might start already grasping at mace or self defense when asked and thus is more difficult to surprise or overwhelm".
"conclusion : the 3rd option is more likely : the man which do not speak to the woman in the elevator"

PS: I don't think either of the reasonning is valid. Just saying that you are making argument out of wool cloth.

Yes, I very clearly state in my post that you CAN make the same kind of argument, but that to my mind they aren't as strong.

For instance, "man want to rape her not make love"
For one, to proposition is not to want to "make love" unless you're being really poetic about your phrasing. From the proposition alone, all we know is that he wants sex.
They aren't mutually exclusive categories. Date rape is incredibly common. Julian Assange is in trouble for supposedly raping someone he had had consensual intercourse with hours previously.

In my example, there is a clear set of categories category you can create from the new information. Those who want to have sex with her and those who don't, and we fully eliminate the category that was far less likely to contain a rapist.

"man should not annoucne intention but take her by surprise otherwise an alarmed woman (alarmed by asking for sex) might start already grasping at mace or self defense when asked and thus is more difficult to surprise or overwhelm".

This is again counter to the majority of rape reports I've heard, which I actually mentioned in my post.
What are the two equivalent categories you imagine for "man want to rape her not make love"?

[Garrette]

I've been thinking a bit and may have to modify my answer, given in post #18, though I also may not. Still thinking.

Here is a recent real world example from my life which I think is analogous. I will leave out details.


I have a background in security--providing it, planning it, conducting assessments of it, etc.



Someone I know recently fired someone. The person fired had been a royal pain for over a year, but institutional policy eliminated a quick termination as a possibility, so it dragged on this long.

During the year+ of the saga, there were many occasions--coinciding with disciplinary/personnel actions--on which my friend felt concerned about the employee taking violent retaliatory action. On those occasions I offered my professional opinion that there was really very little (virtually nothing) about which to be concerned. However, because my friend was distressed, I offered to spend time (unobtrusively) at my friend's workplace during the times of concern; the offer was always declined.

A few weeks ago the actual firing took place. My friend's concern skyrocketed. I explained and offered again. The fired employee exercised the right to appeal. Concern about actions at the appeal were raised. I explained again.

On the day of the appeal I received a call from my friend: The employee came in, refused to take part in the appeal and instead took the offer of resigning in lieu of termination.

My friend was immensely relieved. My response: I drove over immediately and spent the remainder of the day there. My friend tried to talk me out of it, but I insisted. Since the employee had expressly given up on the system, the likelihood of violence had increased. It was still minimal, but it had increased. Just as importantly from the perspective of managing the violence, the window of greatest risk was a manageable two or three business days. So while I expected nothing to happen (it didn't), the increased risk along with precautions I could reasonably take on behalf of my friend, I acted.

So here's the analogy which doesn't perfectly mirror my example since I tailored it to mirror the OP:

Under which circumstance do you think violence is more likely?

1. An employee is fired

2. An employee is fired expresses frustration with the system

3. An employee is fired and does not express frustration with the system


Based on my own actions, I'd have to say I think it's #2, but that's not consistent with the answer I gave to the OP.

Still thinking.

[fls]

How about:

Under which condition has the most rape occurred?

Linda

[sphenisc]

Based on personal observation the first three options are exactly equally like to result in rape. There doesn't seem to be an option for that.

[Lothian]

How about:

Under which condition has the most rape occurred?

Lindac.v. number of times each condition has occured?

[brodski]

There is no need for such an assumption given the choices listed.

Well, the idea hat a person is more or less' likely to be raped by someone who they proposition is a different question as to whether they are more or less likely to be raped by someone who propositions them.
Which question are we trying to address?

[fls]

c.v. number of times each condition has occured?

I think so (not sure what "c.v." means).

Linda

[Lothian]

I think so (not sure what "c.v." means).

LindaSorry. Shorthand for compare with. I meant c.f

[Garrette]

Not sure if this will help, but I thought I'd put completely arbitrary numbers to it:

1. Set (Men and Women on Elevator) = 100

2. Set (Men and Women on Elevator and Man Propositions Woman) = 80

3. Set (Men and Women on Elevator and Man Does Not Proposition Woman) = 20

4. Set (2, above, and Man Rapes Woman) = 5

5. Set (3, above, and Man Rapes Woman) = 2


It's obvious that Sets 2 through 5 are subsets of 1, so the greatest number of rapes happens in Set 1 as opposed to any subset.

However, as probabilities within a set or subset I think it changes.

P(Rape) in Set(1) = 7/100 = .07

P(Rape) in Set(2) = 5/80 = .0625

P(Rape) in Set(3) = 2/20 = .1


The numbers don't mean anything since I made them up EXCEPT to indicate that the probability of rape can indeed be higher in a subset than in the large set.

May have to revise my original answer.

[Garrette]

How about:

Under which condition has the most rape occurred?

LindaAh. Different question. Obviously number 1.

[icerat]

Given the size of the superset of all occurrences in a year of a man riding in an elevator with a woman, I suspect that to several decimal places the likelihood of each all round out to about zero.

(note: this doesn't mean they are zero!)

[Pup]

The numbers don't mean anything since I made them up EXCEPT to indicate that the probability of rape can indeed be higher in a subset than in the large set.

May have to revise my original answer.

See my spoiler-tagged post above. I came to the same conclusion, and in fact, I think that the probability will always be higher in one of the subgroups (unless all three happen to be exactly equal). So one can immediately reject the first option.

That's assuming we're talking probability and not actual numbers, of course. If it's actual numbers, it must be number 1, since it's a total of 2 and 3.

I haven't voted, because it seems to me the most significant factor is missing: what does the woman do afterwards. I'm assuming the vast majority of rapes will be by the male, thus that choice will dominate the probabilities.

I'd think that by far the highest chance of rape would be if he propositions her and she says yes, because then he can steer her to a suitable location--his car, an isolated area, his room, her room, etc. I just can't imagine actual rape occurring between floors on an elevator, due to the time constraints and risk (he has to leave the building, probably through a lobby, at some point). Groping, maybe, yeah, but that's not the question.

While that would mean the greatest chance would be if he propositioned her, I'd think the probability would drop back to less than the other option if she turned him down, since the chances are that he'd move on to another victim that he could manipulate more easily to a safe (for him) location.

This is based on a guess (don't actually know statistics) that most rapes are either by rapists trolling for random handy victims who might at most meet a vague profile--brown hair or something--or are date rapes where what seems like a yes turns into a no but the rapist won't accept that.

[Roadtoad]

See my spoiler-tagged post above. I came to the same conclusion, and in fact, I think that the probability will always be higher in one of the subgroups (unless all three happen to be exactly equal). So one can immediately reject the first option.

That's assuming we're talking probability and not actual numbers, of course. If it's actual numbers, it must be number 1, since it's a total of 2 and 3.

I haven't voted, because it seems to me the most significant factor is missing: what does the woman do afterwards. I'm assuming the vast majority of rapes will be by the male, thus that choice will dominate the probabilities.

I'd think that by far the highest chance of rape would be if he propositions her and she says yes, because then he can steer her to a suitable location--his car, an isolated area, his room, her room, etc. I just can't imagine actual rape occurring between floors on an elevator, due to the time constraints and risk (he has to leave the building, probably through a lobby, at some point). Groping, maybe, yeah, but that's not the question.

While that would mean the greatest chance would be if he propositioned her, I'd think the probability would drop back to less than the other option if she turned him down, since the chances are that he'd move on to another victim that he could manipulate more easily to a safe (for him) location.

This is based on a guess (don't actually know statistics) that most rapes are either by rapists trolling for random handy victims who might at most meet a vague profile--brown hair or something--or are date rapes where what seems like a yes turns into a no but the rapist won't accept that.

Note the high lighted parts: IIRC, most rapes are committed by people who actually know the victim.

[Cuddles]

Planet X. I don't think there's any meaningful way to calculate the probabilities, and therefore no way to compare them.

[AvalonXQ]

As others have pointed out, 1 cannot be the most likely out of 1, 2, and 3, as all instances of 1 are either 2 or 3. If either propositioning or not changes the probability at all, 1 has to be in the middle.

[fls]

Please note I started this thread in Science, Mathematics, Medicine, and Technology.

Linda

[lomiller]

Planet X. I don't think there's any meaningful way to calculate the probabilities, and therefore no way to compare them.

This

[fls]

Please note that the likelihood of rape in any particular set of circumstances includes the likelihood of finding oneself in those circumstances in the first place.

Linda

[brodski]

Please note I started this thread in Science, Mathematics, Medicine, and Technology.

Linda

In that case, should we be looking at research which has been conducted into this question, or just answering based on gut instinct with post hoc rationalisations thrown in?

[roger]

Linda is 31 years old, single, outspoken, and very bright. She majored in philosophy. As a student she was deeply concerned with issues of discrimination and social justice and also participated in anti-nuclear demonstrations.
Now which of two alternatives is more probable?
A) Linda is a bank teller
B) Linda is a bank teller who is active in the feminist movement

The answer is A, of course. Most people answer B. If that is not clear, consider
c) Linda is a bank teller, has $574.34 in the bank, has a Doberman named fluffy, has 11 toes and 9 fingers, is 6'3" tall, and is active in the feminist movement. The more conditions you add, the less likely the choice.

I read poll identically to this rationality test. It is more probable for a rape to occur in an elevator with a M&W in it, vs M&W and some other event.

I recognize Linda is not trying to create a statistics wording test, but that is how it fell out.

And, as an aside, people that are doing conditional probability are assuming a positive correlation between asking and rape. You can't assume your conclusion.

[AvalonXQ]

Please note that the likelihood of rape in any particular set of circumstances includes the likelihood of finding oneself in those circumstances in the first place.

No, actually. It doesn't.
"The likelihood of rape in any particular set of circumstances" = P(rape|circumstances), which is to say, we are looking for how often rape occurs when the established circumstances occur.

To illustrate this, we can make the circumstance something more extreme, like: "A woman gets on an elevator with a serial rapist intending to rape her." Despite the fact that such a circumstance occurs extremely infrequently, we would agree that the given circumstance makes rape much more likely than the simple circumstance of "A woman gets on an elevator with a man".

[AvalonXQ]

And, as an aside, people that are doing conditional probability are assuming a positive correlation between asking and rape. You can't assume your conclusion.

No, I am just assuming some correlation between asking and rape. Positive or negative, it will always put #1 in the middle unless it has absolutely no effect at all (in which case the three will be identical).

[fls]

In that case, should we be looking at research which has been conducted into this question, or just answering based on gut instinct with post hoc rationalisations thrown in?


How do you usually answer these questions? Or alternatively, how do you think it was answered in the threads which spawned the idea?

(ETA: I'm not implying that you listed all the options.)

Linda

[brodski]

How do you usually answer these questions? Or alternatively, how do you think it was answered in the threads which spawned the idea?

Linda

Without data I don't attempt to answer these types of questions.
In the absence of meaningful data with regards to your question, what aspect of Science, Medicine, Mathematics or Technology do you wish to discuss?

[Last of the Fraggles]

I'm going to vote for the absent 'same probability in each case' option. Can't see how propositioning someone makes any impact on the likelihood of rape.

Not really sure where this thread is going or what the objective is but I will wait and see.

[fls]

Without data I don't attempt to answer these types of questions.
In the absence of meaningful data with regards to your question, what aspect of Science, Medicine, Mathematics or Technology do you wish to discuss?


It will be mathematics. I wanted to let it run for a bit before discussing it.

Linda

[Foolmewunz]

Based on the only known statistics outside of the poll in the OP, I'm going with "...propositions her".

Why? Because the one thing that we have as fairly common knowledge is that most rapes are perpetrated by someone the victim knows. The closest to "getting to know" in those scenarios is at least talking to her, even if an awkard come-on. So if there's a "more likely" scenario, to me it would be if the guy propositions her, thereby establishing some sort of minimal definition of "knowing" the other person.

This, of course, assumes something not in the OP - that they don't know each other when they enter the elevator.

[dogjones]

Errr... what poll?

ETA: Must be the first or last option, because the second and third options are subsumed by the first.

I voted for the first option.

I agree, but I voted for the third option. This because I believed the spirit of the question to be "Is a rapist likely to try a proposition first?" - the answer to which I thought was "Probably not, much like a dog does not bark before he bites." However, on reflection I realise that this is not necessarily true.

I should have gone for Planet X I think.

[case#46cw39]

Until recently my thought process would have been along these lines, though perhaps not quite so precise. But lately I am trying to make myself simplify my thought process and the best way for me to do that is sets. Viewing the question this way, the full set is "Man and Woman on Elevator." Subset is "Man Propositions Woman on Elevator." Second subset is "Man Does Not Proposition Woman on Elevator." Final, unstated subset, is "Man Rapes Woman." That final subset will include some of the first two subsets, but all three will be smaller than the full set.


Using sets is always a good idea. By division any genus X must be dividable into some of it Y and some of it -Y. (where could be empty set)

X {Man and woman on elevator} must be divisible into

Y {X with proposition}

-Y {X without proposition}

In other words "X is Y or not Y" is always true.

Thus X must always be the correct answer, since it contains all the possibilities, namely all of Y and -Y

A similar poll would be:

What's more likely to bite?

[]dogs

[]boxers

[]non-boxers

The answer can only be dogs.

ETA unless one of the subsets is empty at which time it would be a tie between genus and non-empty set.

[Jekyll's Guest]

Is a man a rapist?

1) Yes.
2) Maybe.
3) Not so far.

That's the poll.

[brodski]

Using sets is always a good idea. By division any genus X must be dividable into some of it Y and some of it -Y. (where could be empty set)

X {Man and woman on elevator} must be divisible into

Y {X with proposition}

-Y {X without proposition}

In other words "X is Y or not Y" is always true.

Thus X must always be the correct answer, since it contains all the possibilities, namely all of Y and -Y

A similar poll would be:

What's more likely to bite?

[]dogs

[]boxers

[]non-boxers

The answer can only be dogs.

imagine if all boxers and only boxers bite and that they bite on every encounter.
Imagine that boxers make up 50% of the dog population.

You encounter a boxer, what is the likelihood you will be bitten?
You encounter a non-boxer, what is the likelihood you will be bitten?
You encounter a dog, what is the likelihood you will be bitten?

Imagine that we have no information about the ratio of boxers to non-boxers, and have no information about the breeds involved in biting incidents.


You encounter a boxer, what is the likelihood you will be bitten?
You encounter a non-boxer, what is the likelihood you will be bitten?
You encounter a dog, what is the likelihood you will be bitten?

[Ron_Tomkins]

Can't answer unless you specify the height of the building and thus, the length of time the elevator will be closed ;)

[case#46cw39]

imagine if all boxers and only boxers bite and that they bite on every encounter.
Imagine that boxers make up 50% of the dog population.

You encounter a boxer, what is the likelihood you will be bitten?
You encounter a non-boxer, what is the likelihood you will be bitten?
You encounter a dog, what is the likelihood you will be bitten?

Imagine that we have no information about the ratio of boxers to non-boxers, and have no information about the breeds involved in biting incidents.


You encounter a boxer, what is the likelihood you will be bitten?
You encounter a non-boxer, what is the likelihood you will be bitten?
You encounter a dog, what is the likelihood you will be bitten?

Well, we don't see genus dog just their species. So the only way to make it so we encounter a dog, but don't know if it's a boxer or non-boxer is if its behind a curtain.

What's more likely to bite?
[] dog behind curtain (might be a boxer or non-boxer, can't be anything else.)
[] boxer
[] non-boxer

Only answer: Don't know. because don't know what kind of dog behind curtain, maybe it's a vicious, snarling type boxer behind curtain. Still flawed poll. Needs separate, non-overlapping choices.

[AvalonXQ]

Using sets is always a good idea. By division any genus X must be dividable into some of it Y and some of it -Y. (where could be empty set)

X {Man and woman on elevator} must be divisible into

Y {X with proposition}

-Y {X without proposition}

In other words "X is Y or not Y" is always true.

Thus X must always be the correct answer, since it contains all the possibilities, namely all of Y and -Y

A similar poll would be:

What's more likely to bite?

[]dogs

[]boxers

[]non-boxers

The answer can only be dogs.

ETA unless one of the subsets is empty at which time it would be a tie between genus and non-empty set.

As explained earlier in this thread, you have it backwards.
The question you're trying to answer, is:

A man rapes a woman. Which is most likely?
A. The man is on an elevator.
B. The man is on an elevator and propositions the woman.
C. The man is on an elevator and doesn't proposition the woman.

The answer to this question is, of course, A. But that's not the question that was asked.
The question that was asked can be rephrased as follows:

Which number is largest -- A, B, or C?
A = P(rape|(man and woman in elevator))
B = P(rape|(man and woman in elevator) AND (man propositions woman))
C = P(rape|(man and woman in elevator) AND NOT(man propositions woman))

It is always true that A is between B and C. The only possibilities are...
C<=A<=B
B<=A<=C
This is very straightforward conditional probability stuff.

[case#46cw39]

As explained earlier in this thread, you have it backwards.
The question you're trying to answer, is:


The answer to this question is, of course, A. But that's not the question that was asked.
The question that was asked can be rephrased as follows:

It is always true that A is between B and C. The only possibilities are...
C<=A<=B
B<=A<=C
This is very straightforward conditional probability stuff.

Which number is largest -- A, B, or C?
A = P(rape|(man and woman in elevator))
B = P(rape|(man and woman in elevator) AND (man propositions woman))
C = P(rape|(man and woman in elevator) AND NOT(man propositions woman))


How is event A possible, without it being an example of either event B or event C?

It's not possible so is not a unique choice. That's the problem I'm having with this poll. But maybe poll experts know better then me. Wouldn't be surprised.

[AvalonXQ]

How is event A possible, without it being an example of either event B or event C?

It's not. But additional information can make something either more or less likely -- that's a basic component of Bayesian probability that you're missing.
As an easy example, a person who smokes is more likely to get lung cancer. If we know nothing about Person X, and then learn that Person X is a smoker, the chance of Person X having lung cancer goes up.

[fls]

This is fun. :). I will admit that I forgot about the extent to which people manage to drag out these questions (I forgot about all those Monty Hall threads).

From a mathematical perspective, the answer is simple. As Complexity, Pup, Garrette, Roger, and case#46cw39 (apologies if I forgot someone) pointed out. Option 1 subsumes options 2 and 3, so it represents the larger likelihood. The problem seems to be in picturing the question as referring to the group of women who get on an elevator alone with a man. Instead, as you start adding conditions, the tendency seems to be to look at the probability within each condition.

Linda

[AvalonXQ]

From a mathematical perspective, the answer is simple. As Complexity, Pup, Garrette, Roger, and case#46cw39 (apologies if I forgot someone) pointed out. Option 1 subsumes options 2 and 3, so it represents the larger likelihood.
No; that is backwards.
The man propositioning the woman either makes it more likely or less likely that a rape will occur. If it makes it more likely, then #3 is the correct answer. If it makes it less likely, then #2 is the correct answer. #1 is never the correct answer.


The problem seems to be in picturing the question as referring to the group of women who get on an elevator alone with a man. Instead, as you start adding conditions, the tendency seems to be to look at the probability within each condition.

That is the question you asked -- the likelihood of rape given the situation, not the total number of rapes in each situation.

Would it help for me to present the question re-written so that the answer is as you describe?

[AvalonXQ]

If you had written this question, your analysis would be correct:

"Under which circumstance do you think more rapes occur?"

The total number of rapes will be higher under #1 than either #2 or #3, but the likelihood of rape (that is, the number of rapes per woman in this situation) will be higher in either #2 or #3.

[quarky]

How about:

Under which condition has the most rape occurred?

Linda

War.

War and rape go together like love and marriage.

or did you mean within the parameters of the poll?

(Only at jref could such a sensitive issue be turned into a math puzzle. Somewhat charming I suppose.)

[caniswalensis]

All iI can really say is that I do not ever want to visit planet X! :eek:

[RenaissanceBiker]

Note the high lighted parts: IIRC, most rapes are committed by people who actually know the victim.

Yes, but I do know of one case involving a stranger and an elevator. A young lady I know worked in a professional office in a tall downtown building. One morning she got into an elevator with a stranger. She pushed the button for her floor and he pushed a button for a floor lower than hers. When the door opened it turned out to be an unoccupied floor. It was being renovated for a new tenant. The stanger pushed the young lady out of the elevator and raped her there on the floor. He then left by the same elevator and was never caught. There was no conversation or proposition.

[fls]

If you had written this question, your analysis would be correct:

"Under which circumstance do you think more rapes occur?"

The total number of rapes will be higher under #1 than either #2 or #3, but the likelihood of rape (that is, the number of rapes per woman in this situation) will be higher in either #2 or #3.

Yeah, I stated it that way in post #27. I think stating it that way makes it easier to keep the situation in mind (a woman got on an elevator alone with a man) when considering conditions. The likelihood of rape can only change in the way you describe if, in addition to adding conditions, you change which group you are considering at risk (i.e. it is no longer the group of women who got on an elevator alone with a man).

Linda

[Pup]

From a mathematical perspective, the answer is simple. As Complexity, Pup, Garrette, Roger, and case#46cw39 (apologies if I forgot someone) pointed out. Option 1 subsumes options 2 and 3, so it represents the larger likelihood.

That's what I thought at first, but I changed my mind. It's got to be 2 or 3, but not 1, since 1 will be in between (or equal to) 2 and 3.

This is so begging for a parody thread: Likelihood of rape, if Monty Hall were on an elevator with a woman...

[fls]

That's what I thought at first, but I changed my mind. It's got to be 2 or 3, but not 1, since 1 will be in between (or equal to) 2 and 3.

This is so begging for a parody thread: Likelihood of rape, if Monty Hall were on an elevator with a woman...

Does it help if the question is presented in the form...


Under which condition has the most rape occurred?


Linda

[Pup]

Does it help if the question is presented in the form...


Under which condition has the most rape occurred?


Linda

Changes it completely. Then it's number 1, no doubt, since as Case and everybody said, 2 and 3 are a subset. Unless all three are zero, but I doubt that.

Well, I'm also not sure what things are like on Planet X. Or what Monty Hall's been up to.

[blobru]

Does it help if the question is presented in the form...


Under which condition has the most rape occurred?


Linda


I think I would have gone with -- Given a man and woman on an elevator, would you expect the man's propositioning the woman makes rape:

more likely; (your option 2)
less likely; (option 3)
no correlation; (sort of option 1)
on Planet X, elevators only go down.

[fls]

I think I would have gone with -- Given a man and woman on an elevator, would you expect the man's propositioning the woman makes rape:

more likely; (your option 2)
less likely; (option 3)
no correlation; (sort of option 1)

I think this is how we are inclined to decide it (it explains why many commenters made it a story, when a story about RW riding the elevator with a man would have been a non-starter).

on Planet X, elevators only go down.

You are a bad man.

Linda

[blobru]

- You are a bad man.

Linda


:whistling

[trvlr2]

This is an inside joke I'm not up on, right? I mean you don't seriously think any of these minor changes in the circumstances make rape more or less likely? I've never come so close to voting for the planet X silliness.

I rarely agree with BR. So, Planet X won.

[Roboramma]

I don't think anyone knows the answer to this question. Any answer I give would be pure guesswork, and not worth much.

[epepke]

Linda is 31 years old, single, outspoken, and very bright. She majored in philosophy. As a student she was deeply concerned with issues of discrimination and social justice and also participated in anti-nuclear demonstrations.
Now which of two alternatives is more probable?
A) Linda is a bank teller
[LEFT]B) Linda is a bank teller who is active in the feminist movement
[COLOR=#000000]
The answer is A, of course. Most people answer B. If that is not clear, consider
c) Linda is a bank teller, has $574.34 in the bank, has a Doberman named fluffy, has 11 toes and 9 fingers, is 6'3" tall, and is active in the feminist movement. The more conditions you add, the less likely the choice.

Yup. Also, that one was in John Allen Paulos' Innumeracy.

[Lothian]

imagine if all boxers and only boxers bite and that they bite on every encounter.
Imagine that boxers make up 50% of the dog population.

You encounter a boxer, what is the likelihood you will be bitten?Is it called Tyson?







apologies to Brodski. I know all sporting jokes sail over his head

[Cavemonster]

Does it help if the question is presented in the form...


Under which condition has the most rape occurred?


Linda

Linda,
I think it's very very important to understand that this isn't just changing the presentation of the question, it is changing the question entirely.

Not only that, but the question of probability is far more relevant to the situation you're referring to than raw numbers.

Under which condition has more rape occurred?
A) In the United States?
B) In Mike the rapist's Rape van where he rapes ten people every night?

The answer is clearly A. As many people as Mike rapes, more people are raped in the whole country. But this piece of information has no useful bearing in personal risk management. Where would you be more worried, A or B?

Now if we say, as the original poll did, "Under which circumstance do you think rape is more likely?"
A) A woman is in the United states
B) A woman is in Mike the rapist's rape van

We know that something like 1 in 6 women are raped over their lifetime, but 100% of women who wind up in Mike's van are raped. The answer can only be B.

[Pup]

Linda,
I think it's very very important to understand that this isn't just changing the presentation of the question, it is changing the question entirely.

Exactly. The original poll, or its variations, seem to invite two kinds of answerers, the Monty Hall responders, who see the questions as a logical problem and answer accordingly, or the intuitive responders, who see the questions as a social poll and answer what they think seems to be the purpose of the questions (which may or may not actually be the purpose).

I don't see how one can get useful data with such a mix of votes, since the Monty Hall responders will skew the answers in one direction, while the intuitive responders may or may not add to or detract from the Monty Hall answers.

So you won't get a clear picture of the intuitive responses, since the Monty Hall results will skew that, and you won't get a clear picture of how well people can solve Monty Hall puzzles, since there's no clear "wrong" answer that most people would "think" is right, like there is in Cavemonster's example:


Under which condition has more rape occurred?
A) In the United States?
B) In Mike the rapist's Rape van where he rapes ten people every night?

Planet X is the right response. I just know it is.

[Ivor the Engineer]

I don’t think asking for the likelihood (http://en.wikipedia.org/wiki/Likelihood_function) of rape makes sense because rape an outcome of the scenario rather than a parameter. What the question appears to be asking for is the probability that a rape occurs given the particular value of a parameter.

If a man propositioning a woman in an elevator is correlated either positively or negatively with rape, then, as others have already pointed out, the probability of a rape occurring with the additional parameter specified in one of the options 2 or 3 will be higher than option 1.

If a man propositioning a woman in an elevator is uncorrelated with rape then the probability of rape is identical between the three options, since the additional parameter reduces the numerator and the denominator in exact proportion to each other in options 2 and 3.

[fls]

Linda,
I think it's very very important to understand that this isn't just changing the presentation of the question, it is changing the question entirely.

I agree that it changes what question most people were trying to answer (especially if I include people who second-guessed themselves and changed their answer) - that is, the question people were trying to answer and the question they should have been answering were different. And the second way of stating the question more clearly directs them to that question.

Not only that, but the question of probability is far more relevant to the situation you're referring to than raw numbers.

As I hinted at earlier...it doesn't matter whether you are referring to raw numbers or whether you are referring to probability unless you are changing the number of women in your at risk group (i.e. the denominator). That is, when you are asking whether I mean likelihood or whether I mean numbers, what you are really asking is "does adding a condition mean that I change the number of at risk women to only those women with that condition?"

Now, it looks like the intuitive answer to that question is "yes", because it looks like the bulk of the people thought about the probability of each group separately, depending upon the conditions attached to the group. At the start of the poll, the second option had the most votes - more than double the votes for option 1. As the discussion went on, and as I offered some suggestions on ways to approach the question, the votes began to even out.

So the question which needs to be answered/justified is "does adding a condition change the number of women who are at risk in the scenario?"

Under which condition has more rape occurred?
A) In the United States?
B) In Mike the rapist's Rape van where he rapes ten people every night?

The answer is clearly A. As many people as Mike rapes, more people are raped in the whole country. But this piece of information has no useful bearing in personal risk management. Where would you be more worried, A or B?

Now if we say, as the original poll did, "Under which circumstance do you think rape is more likely?"
A) A woman is in the United states
B) A woman is in Mike the rapist's rape van

We know that something like 1 in 6 women are raped over their lifetime, but 100% of women who wind up in Mike's van are raped. The answer can only be B.

Right. Because you are clearly talking about two different groups of at risk women - the group of women who are in the US and the group of women who enter Mike the rapist's rape van. Those groups clearly have nothing to do with each other (even if Mike the rapist's rape van is in the US). This isn't anything like the scenario in the OP.

Linda

[Cavemonster]

I don’t think asking for the likelihood (http://en.wikipedia.org/wiki/Likelihood_function) of rape makes sense because rape an outcome of the scenario rather than a parameter.

From the wiki:
In non-technical parlance, "likelihood" is usually a synonym for "probability"

[fls]

Exactly. The original poll, or its variations, seem to invite two kinds of answerers, the Monty Hall responders, who see the questions as a logical problem and answer accordingly, or the intuitive responders, who see the questions as a social poll and answer what they think seems to be the purpose of the questions (which may or may not actually be the purpose).

I don't see how one can get useful data with such a mix of votes, since the Monty Hall responders will skew the answers in one direction, while the intuitive responders may or may not add to or detract from the Monty Hall answers.

So you won't get a clear picture of the intuitive responses, since the Monty Hall results will skew that, and you won't get a clear picture of how well people can solve Monty Hall puzzles, since there's no clear "wrong" answer that most people would "think" is right, like there is in Cavemonster's example:

Actually I suspect that both the Monty Hall responders and the intuitive responders are answering by trying to guess at the probability within each set of conditions.

Planet X is the right response. I just know it is.

That may be the case. Somebody should go through the polls and form a description of Planet X based on all the characteristics which have been assigned to it. We may be able to guess at the likelihood from that.

Linda

[fls]

Yeah, likelihood and probability (and chance and frequency and rate) can pretty much be used interchangeably here.

Linda

[Pup]

Actually I suspect that both the Monty Hall responders and the intuitive responders are answering by trying to guess at the probability within each set of conditions.

The Monty Hall responders, however, don't need to guess the probability within all sets in order to choose or eliminate answers.

The correct Monty Hall responders (I think) can eliminate #1 arbitrarily, regardless of the probability, so it should get zero votes from correct Monty Hall responders. Other Monty Hall responders (including me, initially), can mistake #1 as the guaranteed correct answer, regardless of probabilities.

[fls]

The Monty Hall responders, however, don't need to guess the probability within all sets in order to choose or eliminate answers.

The correct Monty Hall responders (I think) can eliminate #1 arbitrarily, regardless of the probability, so it should get zero votes from correct Monty Hall responders. Other Monty Hall responders (including me, initially), can mistake #1 as the guaranteed correct answer, regardless of probabilities.

Whether they have to guess at all or some of the sets is irrelevant. It's the process of taking the conditional approach which leads both groups astray (regardless of how well it's applied). It wasn't a conditional question to begin with.

Linda

[Ivor the Engineer]

<snip>

So the question which needs to be answered/justified is "does adding a condition change the number of women who are at risk in the scenario?"

<snip>

Yes, because the scenario changes based on whether or not the man propositions. If he does, the denominator is changed from:

the number of women who have been in an elevator

to:

the number of women who have been propositioned in an elevator

in option 2 and:

the number of women who have been not propositioned in an elevator

in option 3.

[Pup]

Whether they have to guess at all or some of the sets is irrelevant. It's the process of taking the conditional approach which leads both groups astray (regardless of how well it's applied). It wasn't a conditional question to begin with.

Is there some meaning of "conditional question" I'm unaware of? I thought it meant a question in which a condition is assumed: If X, then is Y...?

A normal reading of the poll seems to naturally suggest conditional questions, such as: if a man and a woman are in an elevator, then what is the likelihood a rape will occur?

I don't see how one could approach answering the poll, without mentally forming conditional questions like that. This is certainly going to be a strange and obscure gotcha.

[AvalonXQ]

It's the process of taking the conditional approach which leads both groups astray (regardless of how well it's applied). It wasn't a conditional question to begin with.


Sorry, but "under which circumstance" is a conditional question.

You asked a conditional question whether you intended to or not.

[Elfman]

I disagree with the implications of your survey.

In order to have a rape, or possibility of a rape, you need a rapist. Having a penis doesn't automatically qualify you as one.

[AvalonXQ]

I disagree with the implications of your survey.

In order to have a rape, or possibility of a rape, you need a rapist. Having a penis doesn't automatically qualify you as one.

Nor is it even a requirement to be one.

[Elfman]

"Nor is it even a requirement to be one." - AvalonXQ


Agreed

[brodski]

apologies to Brodski. I know all sporting jokes sail over his head

In this day and age do your really think it is morally, or socially, acceptable to indulge homophones?

I hope the duck on your head is less bigoted.

[brodski]

Nor is it even a requirement to be one.

That depends on your jurisdiction.

[Lothian]

In this day and age do your really think it is morally, or socially, acceptable to indulge homophones?Know eye do knot.

[AvalonXQ]

That depends on your jurisdiction.

No, it doesn't.

If you force someone to have sex with you, you are a rapist.

I don't care if you're in a state with outdated sexist laws, in Ethiopia, or on the high seas -- having non-consensual sex with another person is rape.

Whether or not a certain rape is a crime may be a reasonable question. Debating the meaning of rape or sexual assault within the law doesn't change whether or not someone has committed rape, which has a reasonable non-legal definition which we can apply to women as well as men.

[fls]

Yes, because the scenario changes based on whether or not the man propositions. If he does, the denominator is changed from:

the number of women who have been in an elevator

to:

the number of women who have been propositioned in an elevator

in option 2 and:

the number of women who have been not propositioned in an elevator

in option 3.

So when a man and a woman are in an elevator and he doesn't proposition her, she disappears in a puff of smoke?

Linda

[Dani]

I don't understand this thread.

Likelihood = probability.

Then it has to be 2 or 3.

[fls]

Is there some meaning of "conditional question" I'm unaware of? I thought it meant a question in which a condition is assumed: If X, then is Y...?

A normal reading of the poll seems to naturally suggest conditional questions, such as: if a man and a woman are in an elevator, then what is the likelihood a rape will occur?

I don't see how one could approach answering the poll, without mentally forming conditional questions like that. This is certainly going to be a strange and obscure gotcha.

What if the choices in the poll were:

A woman is on an elevator with a man.
A woman is on an elevator with a man and he is missing his left little finger.
A woman is on an elevator with a man and he is missing his left little finger and he was born in Springfield, Missouri.

I think you would be less inclined to consider the probabilities only under those conditions and guess that the these conditions were slightly more or less likely than the first. Instead, you might guess that the second and third options were quite unlikely - that is, you rarely (if ever) see all those conditions present along with rape.

Linda

[fls]

Sorry, but "under which circumstance" is a conditional question.

You asked a conditional question whether you intended to or not.

It's not a matter of what I intended, it's a matter of which question represents the set-up.

Under which condition has the most rape occurred?

How likely is rape given these conditions?

Linda

[Ivor the Engineer]

So when a man and a woman are in an elevator and he doesn't proposition her, she disappears in a puff of smoke?

Linda

No, they're in the third scenario outlined in your poll.

What if the selection function was perfect? I.e. What if *only* men who are rapists propositioned women in elevators? In that case P(Rape) = 1 in scenario 2, P(Rape) = 0 in scenario 3 and 0<P(Rape)<=1 in scenario 1.

[AvalonXQ]

It's not a matter of what I intended, it's a matter of which question represents the set-up.
And the answer is...
How likely is rape given these conditions?

[Pup]

What if the choices in the poll were:

A woman is on an elevator with a man.
A woman is on an elevator with a man and he is missing his left little finger.
A woman is on an elevator with a man and he is missing his left little finger and he was born in Springfield, Missouri.

I think you would be less inclined to consider the probabilities only under those conditions and guess that the these conditions were slightly more or less likely than the first. Instead,you might guess that the second and third options were quite unlikely - that is, you rarely (if ever) see all those conditions present along with rape.

Linda

:confused: That still requires one to consider conditional questions, like "If a man and woman are on an elevator, then what is the probability a rape will occur," or "If a man from Springfield missing a finger is on an elevator with a woman, then..."

But we've stirred up Monty Hall again, because those options are fundamentally different from the options in the first poll. In the first poll, answers 2 and 3 included all the subsets of answer 1, so the probability of #1 had to be between 2 and 3, and therefore could not be the most likely.

In this case, though, that's not true, so #1 is back in the running again, and any of the three could be the correct answer. In fact, I think it would only take one serial rapist from Springfield who's missing a finger to make #3 the obvious winner. Um, right? I'm so not good at Monty Hall problems.


I think so, because let's say the chance of #1 is, I dunno, 1 in 10 million. But if there are only 20 Springfield finger amputees, and each has ridden an elevator 100 times in his life, and the one who's a serial rapist raped 10 women, that's 10 rapes out of 2000 elevator rides if a woman is on the elevator with a Springfield amputee, or 1:10,000,000 for #1 vs. 1:200, for #3, an easy winner for #3.


Whereas in the first poll, #2 and #3 were such large sets, no single serial rapist could skew the results.

But if the question in the new poll were changed to, under which of the three conditions have the most rapes occurred, #1 would still be the definite winner.

[Ivor the Engineer]

It's not a matter of what I intended, it's a matter of which question represents the set-up.

Under which condition has the most rape occurred?

How likely is rape given these conditions?

Linda

I think you're asking a slightly different question. Obviously rape occurs more frequently/often under the less restrictive (compared to options 2 and 3) conditions of option 1.

But your question wasn't about the frequency of occurrence of rape in elevators, it was about the probability of rape in elevators.

[fls]

:confused:

Yeah, sorry about that. I realized just after I sent it that it was unnecessary. You had no problem with the question below, so I didn't need some elaborate and confusing example to illustrate what I was getting at.

But if the question in the new poll were changed to, under which of the three conditions have the most rapes occurred, #1 would still be the definite winner.

Okay. And does that question represent the set-up (wondering about the safety of a woman getting on the elevator alone with a man)?

Linda

[Jensen]

I have never engaged in conversation before raping a woman in an elevator. I think.

[Skeptic Ginger]

Errr... what poll?

ETA: Must be the first or last option, because the second and third options are subsumed by the first.

I voted for the first option.
Ooops, can I change my vote? ;)

[Skeptic Ginger]

....

Okay. And does that question represent the set-up (wondering about the safety of a woman getting on the elevator alone with a man)?

LindaMaybe the elevator is safer than the hallway. :)

[Pup]

But if the question in the new poll were changed to, under which of the three conditions have the most rapes occurred, #1 would still be the definite winner
Okay. And does that question represent the set-up (wondering about the safety of a woman getting on the elevator alone with a man)?

Well, I'd read it to mean which of the three specified situations had the most rapes. There's really no other way to approach the question, because one doesn't know what other sets of women to consider.

For example, the set of "women not on elevators with men" would surely have the most rapes, since that's a much larger set at any given time than women on elevators with men.

[fls]

Well, I'd read it to mean which of the three specified situations had the most rapes. There's really no other way to approach the question, because one doesn't know what other sets of women to consider.

For example, the set of "women not on elevators with men" would surely have the most rapes, since that's a much larger set at any given time than women on elevators with men.

The set under consideration would be "women on an elevator alone with a man" (as mentioned in the OP).

Linda

[jasonpatterson]

Honestly this question strikes me as completely silly, but what is driving me nuts as I read it (why am I bothering? self loathing I suppose...) is the idea of "a woman getting on an elevator alone with a man."

If she's getting on the elevator with a man, she's not alone. If you are with someone, stranger or otherwise, you're not alone.

[Marduk]

ha I know this one, the elevator got raped because neither the woman or the man had its consent to enter it
;)

[ftl]

Quite likely in case 2. The woman may be drunk; if a man follows a drunk woman out of a bar and propositions her, knowing that she's too drunk to think straight, and then they have sex, it'll be rape even if there wasn't physical coercion.

[AvalonXQ]

Honestly this question strikes me as completely silly, but what is driving me nuts as I read it (why am I bothering? self loathing I suppose...) is the idea of "a woman getting on an elevator alone with a man."

If she's getting on the elevator with a man, she's not alone. If you are with someone, stranger or otherwise, you're not alone.

"Alone" doesn't have to mean "one person and no one else"; it can mean "the specified group and no one else". Two people, or even ten people, can be "alone" if they are the only people there -- if they are there to the exclusion of anyone else.

[Nosi]

So the assumption is that the rapist is the man?

And the assumption that the raped is a woman? ;) There are a LOT of men in the closet who have suffered rape by both men and women.

[epepke]

Nah, it's a simple probability question. Since every instance of 2 or 3 has to be an instance of 1, then 1 cannot be smaller than either 2 or 3.

It's like saying, Who votes more Republican?

1) Americans
2) American Blondes
3) American Brunettes

Since American Blondes and American Brunettes are both Americans, their numbers cannot be higher than the number of Americans total. Simple.

What this question and the answers and the discussion may say is that when a highly emotive subject like rape is brought up, people's brains squirt out of their ears like pudding in a breached pressure cooker. Also, they seem to be enormously proud of this fact and think it makes them really skeptical.

ETA: The Monty Hall problem gets a lot of threads for several reasons.

One is that there are a lot of stupid people.
Another is that vos Savant cannot write, and in her explanation, the question is unanswerable
But the third is that some non-stupid people who can read go on to analyze the problem and try to figure out under just what conditions the question could be answerable

[Ivor the Engineer]

I don't think it's a simple question at all.

Imagine a container filled with rapeseed oil and water. You are offered to drink from a small glass containing fluid extracted from the container in one of the following ways:

1) The container is shaken and part of the contents is poured into the glass.
2) A ladle is used to extract fluid from the top-half of the container and poured into the glass.
3) The glass is filled from a tap at the bottom of the container.

Which method results in the lowest probability of having to drink any rapeseed oil?

[Pup]

I don't think it's a simple question at all.

It's simple if 2 and 3 are subsets of 1 and it's a question of numbers, not probability, which is epepke's example.

In the original poll, 2 and 3 are subsets of 1, but it's a question of probability, not numbers. Still, one can at least definitely eliminate the first of the three choices.


Imagine a container filled with rapeseed oil and water. You are offered to drink from a small glass containing fluid extracted from the container in one of the following ways:

1) The container is shaken and part of the contents is poured into the glass.
2) A ladle is used to extract fluid from the top-half of the container and poured into the glass.
3) The glass is filled from a tap at the bottom of the container.

Which method results in the lowest probability of having to drink any rapeseed oil?

The fundamental difference is that 2 and 3 aren't subsets of 1. To make the question the same, it would need to be:

1 You drink an eight ounce glass drawn from the container.

2 You drink an eight ounce glass drawn from the top of the container

3 You drink an eight ounce glass drawn from the bottom of the container.

4 On Planet X, "rapeseed" is bleeped by the autocensor.

In that case, you could eliminate #1 immediately (and #4 as well, because there are no autocensors on Planet X).

[epepke]

It's simple if 2 and 3 are subsets of 1 and it's a question of numbers, not probability, which is epepke's example.

Yup.

[case#46cw39]

OK, so let's go by probability. I suck at math so pls help me out. Let's say there are 200 elevator rides as described in option A. Let's say there are 100 in option B and 100 in option C.

Let's say the biggest likelihood is that someone just attacks w/o proposition (option C), and the probability is .15

Let's say the smaller likelihood is a proposition is first delivered and then attack launched, (Option B) and the probability is .05.

A. A woman is on an elevator with a man. (200 times @ ??)
B. A woman is on an elevator with a man and he propositions her. (100 times @ .05)
C. A woman is on an elevator with a man and no proposition made. (100 times @ .15)

Is there an automatic answer for option A which is simply a calculation based on your belief about B and C? Is it .10?

[Pup]

OK, so let's go by probability. I suck at math so pls help me out. Let's say there are 200 elevator rides as described in option A. Let's say there are 100 in option B and 100 in option C.

Let's say the biggest likelihood is that someone just attacks w/o proposition (option C), and the probability is .15

Let's say the smaller likelihood is a proposition is first delivered and then attack launched, (Option B) and the probability is .05.

A. A woman is on an elevator with a man. (200 times @ ??)
B. A woman is on an elevator with a man and he propositions her. (100 times @ .05)
C. A woman is on an elevator with a man and no proposition made. (100 times @ .15)

Is there an automatic answer for option A which is simply a calculation based on your belief about B and C?

Yes. B would produce 5 rapes and C would produce 15 rapes, so there would be a total of 20 rapes out of a total of 200 rides. Thus the probability of A would 20/200 or .10.

This is a good example of how one can always eliminate A as the most probable of the three, since its probability falls between B and C.

[fls]

OK, so let's go by probability. I suck at math so pls help me out. Let's say there are 200 elevator rides as described in option A. Let's say there are 100 in option B and 100 in option C.

Let's say the biggest likelihood is that someone just attacks w/o proposition (option C), and the probability is .15

Let's say the smaller likelihood is a proposition is first delivered and then attack launched, (Option B) and the probability is .05.

A. A woman is on an elevator with a man. (200 times @ ??)
B. A woman is on an elevator with a man and he propositions her. (100 times @ .05)
C. A woman is on an elevator with a man and no proposition made. (100 times @ .15)

Is there an automatic answer for option A which is simply a calculation based on your belief about B and C? Is it 1.0?

The woman takes takes 200 elevator rides, some of which will turn out to be under option B and some under option C. The number of rapes which will happen to occur under option B will be the number of times option B occurs during those 200 rides (100) multiplied by the probability of rape in this subset (0.05), for a total of 5 rapes out of 200. The number of rapes which happen to occur under option C will be the number of times option C occurs during those 200 rides (100) multiplied by the probability of rape in this subset (0.15), for a total of 15 rapes out of 200. Option A will include all rapes (5+15), for a total of 20 rapes out of 200. Of course, the number of times option B or C occurs within those 200 elevator rides can change, which will alter the total number of rapes.

Probabilities for each:

A. 0.10
B. 0.025
C. 0.075

Linda

[AvalonXQ]

The woman takes takes 200 elevator rides, some of which will turn out to be under option B and some under option C. The number of rapes which will happen to occur under option B will be the number of times option B occurs during those 200 rides (100) multiplied by the probability of rape in this subset (0.05), for a total of 5 rapes out of 200. The number of rapes which happen to occur under option C will be the number of times option C occurs during those 200 rides (100) multiplied by the probability of rape in this subset (0.15), for a total of 15 rapes out of 200. Option A will include all rapes (5+15), for a total of 20 rapes out of 200. Of course, the number of times option B or C occurs within those 200 elevator rides can change, which will alter the total number of rapes.

Probabilities for each:

A. 0.10
B. 0.025
C. 0.075

Linda

Your calculations are for the following question:

"A man and a woman take an elevator ride. What is the probability that:
A. He will rape her?
B. He will proposition her and rape her?
C. He will rape her without propositioning her?"

Of course, the point of the OP question was to determine whether a woman should be more or less scared once a man propositions her -- which is properly dealt with by the conditional probabilities of the OP question, not the answer to this new question.

[fls]

Your calculations are for the following question:

"A man and a woman take an elevator ride. What is the probability that:
A. He will rape her?
B. He will proposition her and rape her?
C. He will rape her without propositioning her?"

Of course, the point of the OP question was to determine whether a woman should be more or less scared once a man propositions her -- which is properly dealt with by the conditional probabilities of the OP question, not the answer to this new question.

Thank you for addressing what we are trying to determine.

I agree that one can add questions which are applied only to one subset. Garrette's example illustrated that. He was considering whether a fired employee would become violent in order to decide whether he took a particular action (staying with his friend). I didn't say that we were trying to determine whether she should be more or less scared. If you recognize that this matters, then it is a question which should be asked before answering the poll. I was asked this by others (indirectly) and said (indirectly) that we weren't also asking a question which would be applied only to a subset (I was a little worried about being too obvious (not as a trick, just to see what assumptions other people would use)).

Linda

[Garrette]

I've been reading the thread and mostly following the discussion. In my case, any confusion is due to misunderstanding of what is being asked, not to an inability to answer once understood.

I now understand Linda's original intent and am happy to conclude that my original answer was correct. I think AvalonXQ's restatement in post 123 is accurate and more clear, though not the additional bit regarding when a woman should be scared.

[Skeptic Ginger]

So can we say what we already knew, the fact that EG asked RW back to his room for coffee was less worrisome than had he simply gotten on the elevator with her and said nothing? So should RW expect no man to share an elevator with her at 4am in any nice hotel, lest they creep her out?

[Beth]

So can we say what we already knew, the fact that EG asked RW back to his room for coffee was less worrisome than had he simply gotten on the elevator with her and said nothing? No. We didn't know that to begin with nor does the discussion in this thread indicate anything one way or the other because the actual conditional probabilities are unknown.

[Skeptic Ginger]

No. We didn't know that to begin with nor does the discussion in this thread indicate anything one way or the other because the actual conditional probabilities are unknown.I thought the odds were based on logic, not data.

But if we are going by data, the Dr BuzzO blog entry noted that in the actual incidents, hitting on the woman before attempting rape was less likely.

[Roboramma]

Your calculations are for the following question:

"A man and a woman take an elevator ride. What is the probability that:
A. He will rape her?
B. He will proposition her and rape her?
C. He will rape her without propositioning her?"

Of course, the point of the OP question was to determine whether a woman should be more or less scared once a man propositions her -- which is properly dealt with by the conditional probabilities of the OP question, not the answer to this new question.
This.

[fls]

So can we say what we already knew, the fact that EG asked RW back to his room for coffee was less worrisome than had he simply gotten on the elevator with her and said nothing?

I think I know what you are trying to say. If women are to be worried about rape on an elevator, then why only worry for those times you also happen to get on with a propositioner? But when you add "and said nothing", it has the effect of comparing one subset to another subset, rather than a subset to the whole (i.e. it brings us back around to the point of contention).

I suspect that this scenario introduces an availability bias which leads to a focus on subsets rather than the set. We don't think much of the risk of rape for ordinary scenarios (man and woman alone on an elevator), but once given an example of creepy behavior, it makes it easier for us to think about rape in that particular scenario. Which leads us to think about comparisons between subsets without first wondering whether it's the appropriate comparison.

Linda

[Beth]

I thought the odds were based on logic, not data. Depends on the question. In this case, logic is not sufficient to answer your question.

But if we are going by data, the Dr BuzzO blog entry noted that in the actual incidents, hitting on the woman before attempting rape was less likely.
I haven't seen that. Where did he find such statistics?