help with Statistical question

[dms1972]

Just looking a little help with a matter of statistics from anyone who understands, to clarify if a statisical conclusion I heard online in valid.

Does a percentage of a larger population group (say that of a country) map directly onto a smaller group, say that of a community within that country. Eg. if 10% of a country can be said to be xyz, does it follow that 10% of a small group within that population are also xyz?

Any help appreciated.

 

[Danielwright2311]

Need better information, and more clear understanding what you need the information for.

In a grid of x y and z =y being up and z being to the right and x being the the left.

It would be the same unless answer is =to what your problem is.

Do you see what i mean?

With out the problem, there is no answer.

 

[hedrick]

No. There are two reasons.

1) Subpopulations can be different. Let's look at the % of the US population that is Republican. Would you expect the same % to be Republican in New York and Arkansas?

2) Even if the population is completely uniform, there can be variations as a matter of chance. These tend to even out more the larger the population. So small subpopulations will tend to vary more. This depends upon the size, obviously. It's possible that the numbers we're talking about could be so large that the random variation won't be present in even the subpopulations.

But I question how often we see different subgroups that are the same. I guess if you were looking at two classes in the same school you might assume that the populations are on average the same. In that case only the second factor would apply.

 

[klutedavid]

Just looking a little help with a matter of statistics from anyone who understands, to clarify if a statisical conclusion I heard online in valid.

Does a percentage of a larger population group (say that of a country) map directly onto a smaller group, say that of a community within that country. Eg. if 10% of a country can be said to be xyz, does it follow that 10% of a small group within that population are also xyz?

Any help appreciated.

It will depend on the sample size that you test.

For example, lets say that 10% of the British population are Christian and the other 90% are atheists. If we then examine one town in Britain with a sample size of just one person, then it's obvious that we have a 10% chance of that one person being a christian.

The larger the sample size of any population will mean you are more confident of the result. The smaller the sample size then the less confident you will be.

 

[basilbear76]

Just looking a little help with a matter of statistics from anyone who understands, to clarify if a statisical conclusion I heard online in valid.

Does a percentage of a larger population group (say that of a country) map directly onto a smaller group, say that of a community within that country. Eg. if 10% of a country can be said to be xyz, does it follow that 10% of a small group within that population are also xyz?

Any help appreciated.

Someone said, "There are lies, there are damned lies, and there are statistics."

 

[dms1972]

Need better information, and more clear understanding what you need the information for.

In a grid of x y and z =y being up and z being to the right and x being the the left.

It would be the same unless answer is =to what your problem is.

Do you see what i mean?

With out the problem, there is no answer.

Sorry my use of "xyz" was just a way of saying can be classed in a particular way. Was not meaning in terms of x and y and z axis in geometry.

 

[dms1972]

Someone said, "There are lies, there are damned lies, and there are statistics."

Yes it was Mark Twain, I know cause I was thnking of that saying when I posted and was looking it up, haha! Twain attributed it to Disraeli:

Full quote (if anyone is interested)

"Figures often beguile me, particularly when I have the arranging of them myself; in which case the remark attributed to Disraeli would often apply with justice and force: 'There are three kinds of lies: lies, damned lies, and statistics.'

 

[dms1972]

No. There are two reasons.

1) Subpopulations can be different. Let's look at the % of the US population that is Republican. Would you expect the same % to be Republican in New York and Arkansas?

2) Even if the population is completely uniform, there can be variations as a matter of chance. These tend to even out more the larger the population. So small subpopulations will tend to vary more. This depends upon the size, obviously. It's possible that the numbers we're talking about could be so large that the random variation won't be present in even the subpopulations.

But I question how often we see different subgroups that are the same. I guess if you were looking at two classes in the same school you might assume that the populations are on average the same. In that case only the second factor would apply.

I presume that those states tend to vote Democrat. But people can vote either way, which may not be exactly the same comparison as the equation I am thinking about which was a correlation made by Mermaids (a UK transgender charity.) of 10% of the population (in UK) are LGBT, therefore 10% of the pupils at a COE primary school are LGBT. Is that a valid conclusion in your view?

BTW on the basis of studies online, I think the 10% is an inflated figure, but whatever that percentage is, does it follow that that is going to be the percentage of LGBT pupils at any particular school.

I find the conclusion dubious, because density of LGBT people, varies across the UK: London amongst under 35s is the highest density, but other regions are going to be lower.

 

[hedrick]

I am thinking about which was a correlation made by Mermaids (a UK transgender charity.) of 10% of the population (in UK) are LGBT, therefore 10% of the pupils at a COE primary school are LGBT. Is that a valid conclusion in your view?

No, I don't think that's quite accurate (nor is the 10% number to begin with). It's more accurate than saying that X% of the US population is Republican so the population of a specific is also X% Republican.

Why? The difference between that and which party you support is that there's at least a reasonable biological influence. (I'm not saying that it's all biological, and note that I'm using biological and not genetic because there are other aspects to biology.) But that doesn't mean the number is always going to be the same.

Most of the time what you see is how people report themselves. That is going to vary, and might well be different in a school than in the UK population. 100 years ago someone who would identify as gay today probably wouldn't have. They would just have been unhappy in their sexual relationships, and possibly would have had sex secretly with one or more people of the same sex. But they wouldn't have called themselves gay. Today, you can bet that not many people in Saudi Arabia would report themselves as gay. I'd assume (but haven't found data) that the number would be smaller among evangelicals; even if the same fraction might be oriented towards others of the same sex, many likely wouldn't regard themselves as gay. Self-reports have grown slowly, though I doubt that actual sexual attraction has changed.

In the UK, the overall number is 2%, but among age 16-24 it's 4%. It's also higher for males. (See Sexual identity, UK - Office for National Statistics) But you can bet there would be schools where almost no one would report themselves as homosexual (e.g. a Muslim school). Still, at an Anglican school, you'd assume it would be similar to the population of that age group, i.e. most likely it would be around 4%, while the overall population is 2%. If it's all-boys it would likely be a bit higher. But there could be things about the school that would affect it.

Not sure where a 10% number would come from. I heard that number used decades ago, but I've not seen it in any reliable data recently. I've seen things closer to 3 - 4%. Remember, this is self-reports. I've seen some suggestions that there's a much larger fraction with some attraction to the same sex. Remember that there's a spectrum, with various shades of bisexual in the middle. But I haven't seen 10% reporting themselves as gay. This site claims 4.5% for the US. (The Williams Institute) Gallup says 3.8% (Americans Greatly Overestimate Percent Gay, Lesbian in U.S.), but 7.3% of younger people.

[hmm.... my comment about Muslims may be unfair. In the US at least, Muslims are about evenly split between saying homosexuals should be accepted and they shouldn't. Evangelicals are the main religious group where a majority rejects them, although with JW's and Mormons, of course.]

 

[dms1972]

But you can bet there would be schools where almost no one would report themselves as homosexual (e.g. a Muslim school). Still, at an Anglican school, you'd assume it would be similar to the population of that age group, i.e. most likely it would be around 4%, while the overall population is 2%. If it's all-boys it would likely be a bit higher. But there could be things about the school that would affect it.

Thanks for your comments.

I thought it might be higher overall, still not as high as 10%.

Although these matters are being talked about in schools earlier than they used to be, and some kids can be precocious in sexual matters, I wonder how many children would even be thinking much about it except for teachers bringing the matter to their attention earlier? Are not most LGB only becoming aware of any same-sex sexual attraction as they enter adolescence?

 

[hedrick]

In my opinion there isn't entirely clear evidence on this, but there are some pretty good indications.

I think the answer is that it depends upon the person. Lots of gay people say that they always knew. You see articles about 8 year olds saying they're gay. This article points to some actual studies and some assertions without evidence: Is There an Age by Which People Realize Their Sexual Orientation? - Born Gay? - ProCon.org. The most interesting one claims that 11% of people knew in grade school. Note however that this isn't the most common, with many saying they realized in high school. However all of this seems to be based on asking adults what they remember. This is not considered a highly reliable approach. I think the other study is more likely to be reliable. It says around age 10. But even if that's true, people differ enough that it's certainly possible that a few 8 year olds know.

There's also some evidence that you can sometimes see differences in behavior even for young children who are going to end up gay. https://www.sciencemag.org/news/2017/03/toddler-play-may-give-clues-sexual-orientation.

Note that full sexual development isn't necessary. Kids have crushes on people before they're ready to have sex. Hopefully relationships are more than just sex, so it's perfectly possible that attractions could develop earlier. Here's a discussion for parents about "crushes" with young kids: First Crushes: How to Handle Your Child's First Crush. They aren't thinking of gay relationships, but I think this shows how it could be possible to have a gay crush before puberty.

 

[hedrick]

I teach Sunday school for grades 5 to 7. They're well aware of what it means to be gay, and controversies over it. They also report that at one point they had a transgender kid in school. In their school (and our church), nobody seems to care.

Of course I don't know whether any of my kids are gay. That's not a question I'd ask, and so far none has volunteered. I know there are gay kids in our youth program, but I don't know who.