Jump to content

A revisit to Pattay after 12 years


Torneyboy

Recommended Posts

You keep grasping for comebacks by looking for spelling errors, yet you write things like:

 

" I assumed you knew the proper way to sample to answer a question like ............"

 

That's "proper" English?

 

"If I ask you what is the net worth of a typical first class flyer, you don't take everyone who's ever flown first class over x time period and average. You'd sample multiple times over x time period. "

 

First of all, no one with any methodological background would be asking what the net worth of a TYPICAL......flyer is. They would ask:

What is the net worth of first class flyers.

If you knew anything about statistical studies,

you'd know that they would NOT sample multiple times over x time period. They would collect data on EVERYONE (over the time period of interest) with no statistical sampling, if they had the resouces (and, of course, if they could get people to provide their "net worth", which obviously would never happen, and is out of the question). Sampling is only done when researchers do not have the resources (i.e, money, time, etc.) to collect data on the full population (which is called "a census, as opposed to a "sample").

 

I learned a while ago that you don't have a statistical background when you used the phrase "random questions" (i.e., "I thought I'd ask a random question" is what you said, in a post where you proclaimed to be a "number cruncher, etc). That's fine for most writers, but someone with a statistial background would never say that...because such people use the word "random", all the time, and it means "equal probability of selection" to them (I mean, to people like me).

 

Now, go look for some spelling errors, and I'm sure you'll find some...because obviously I couldn't care less about spelling.

 

 

 

 

Link to comment
Share on other sites

  • Replies 53
  • Created
  • Last Reply

Dr. koow, please share your background. Mine is in stochastic calculus so I'm aware of how different fields of study use the term 'random'. Even in your context (I'm guessing you are a sociology, psych or poli sci major), 'random' only means "equal probability of selection" if the distribution is uniform. I'm sure you've heard of a few other probability distributions. Bell curve aka Gaussian maybe?

 

And I admit I use a lot of math terms loosely when talking to non-mathematicians or about things like whores and squirting. You should see the way I use the terms surely and definitely. I don't think they'll take back my degree for that. Just in case, please don't tell my professors.

Link to comment
Share on other sites

First of all, no one with any methodological background would be asking what the net worth of a TYPICAL......flyer is. They would ask:

What is the net worth of first class flyers.

If you knew anything about statistical studies,

you'd know that they would NOT sample multiple times over x time period. They would collect data on EVERYONE (over the time period of interest) with no statistical sampling, if they had the resouces (and, of course, if they could get people to provide their "net worth", which obviously would never happen, and is out of the question). Sampling is only done when researchers do not have the resources (i.e, money, time, etc.) to collect data on the full population (which is called "a census, as opposed to a "sample").

 

Ok, I'm procrastinating so...

 

Your statements show you don't know the difference between inferential and descriptive statistics. I posed an inferential question, which could have been formalized as 'What's the expected value of net worth for a person selected randomly from a first class flight?' Apparently, you're only familiar with descriptive statistics and can only answer questions like 'What's the average net worth for all people who have ever taken first class flights?'

 

The first question is asking you to make an inference about someone drawn from a population (inferential statistics). The second question is asking you to summarize information about the population itself(descriptive statistics). I assure you that statistics has tools to answer both types of questions and people with "methodological backgrounds" ask inferential questions all the time.

 

Collecting information on all people over a time period and then taking an average would answer question #2 but not #1.

 

To calculate an exact answer to #1, you could collect all information, create an empirical probability distribution, and then calculate an exact answer. Instead most statisticians would prefer to take samples to answer question #1. Why? Because of a little thing called the Central Limit Theorem. Please google and get back to me.

Link to comment
Share on other sites

What? Since when is it off topic to discuss the central tendency of net worth for a black guy flying first class since 1970?!

 

I'm pretty sure I said I was glad we stayed ON topic.

 

You wanna argue about it? You wanna piece of me? C'mon big boy ... differential equations, white boards, and dry-erase markers at 10 paces!

Link to comment
Share on other sites

What's really funny is that the rubbish you've been writing is probably impressing the uneducated readers of this forum. I assume that's

your intended audience (people who don't know any better). Some of them might think you know what you're talking about, since they have no methodological background at all. But the truth is, you don't know what you're talking about..and what you're writing would be easily seen, on a forum populated by mathematically inclined participants, as pure gibberish (starting with your curve fitting exercise on 3 crudely-measured data points). Isn't it great pretending to know what you're talking about on a forum where no one has any math/stats background? Now, with the latest post, you are demonstrating your lack of knowledge about the field of statistics (but of course, trying to impress the uneducated readers). Keep up the jibberish!! Its entertaining....and keep looking for those spelling errors.

 

 

 

 

04-06-09 11:35 AM - Post#785476

 

In response to koow - A revisit to Pattay after 12 years

 

* koow Said:

 

First of all, no one with any methodological background would be asking what the net worth of a TYPICAL......flyer is. They would ask:

What is the net worth of first class flyers.

If you knew anything about statistical studies,

you'd know that they would NOT sample multiple times over x time period. They would collect data on EVERYONE (over the time period of interest) with no statistical sampling, if they had the resouces (and, of course, if they could get people to provide their "net worth", which obviously would never happen, and is out of the question). Sampling is only done when researchers do not have the resources (i.e, money, time, etc.) to collect data on the full population (which is called "a census, as opposed to a "sample").

 

 

 

Ok, I'm procrastinating so...

 

Your statements show you don't know the difference between inferential and descriptive statistics. I posed an inferential question, which could have been formalized as 'What's the expected value of net worth for a person selected randomly from a first class flight?' Apparently, you're only familiar with descriptive statistics and can only answer questions like 'What's the average net worth for all people who have ever taken first class flights?'

 

The first question is asking you to make an inference about someone drawn from a population (inferential statistics). The second question is asking you to summarize information about the population itself(descriptive statistics). I assure you that statistics has tools to answer both types of questions and people with "methodological backgrounds" ask inferential questions all the time.

 

Collecting information on all people over a time period and then taking an average would answer question #2 but not #1.

 

To calculate an exact answer to #1, you could collect all information, create an empirical probability distribution, and then calculate an exact answer. Instead most statisticians would prefer to take samples to answer question #1. Why? Because of a little thing called the Central Limit Theorem. Please google and get back to me.

Link to comment
Share on other sites

Ok, since you insist. :) Why don't you post this on a math forum and tell us the results.

 

Here's a free lesson on interpolation vs. regression.

 

You keep bringing up interpolation with three points (actually I used two. you can't count eh?) as if there is some number of data points where interpolation is valid. You are confusing interpolation with regression. Regression is fitting a function to data points according to some criteria, usually minimizing the squared differences from the predicted value. In general, in a regression, the more data points, the more accurate your function fitting.

 

In interpolation, you start with an assumed function and then try to calculate a data mid-point using it. As I've said, most interpolation is piecewise and only requires two data points. Interpolation has very little to do with the number of data points because you've assumed a function. It has everything to do with how closely your assumed function matches the underlying data.

 

If I give you the coordinates of a bingo ball as it is being rolled constantly around inside of a cage at 1:00 and 2:00 and ask you to interpolate the coordinates at 1:30, you can't. The underlying process is too wild and no function you can come up with can describe it. No matter how many data points you give me, interpolation is virtually useless.

 

If I tell you the temperature of a cup of water inside an oven is 98 degrees at 1:00 and 20 degrees at 2:00, you still can't. But once I tell you that the temperature of the oven was unchanged during that time, you can now interpolate with great accuracy. What changed? Not the number of data points but by adding a condition, a known function, the heat diffusion equation, can be used.

 

So you see, your focus on the number of data points in interpolation shows that you have no idea what interpolation is or that you confused interpolation with regression.

 

When it comes to sex ratio, I wasn't interpolating because I wasn't coming up with a function. But what I did say was that ANY function has two properties. It only moves in one direction and it averages out to .98. You could also add a common sense assumption that it is fairly smooth as Thailand isn't at war or something else that might cause massive numbers of boys to die suddenly at age x. I said under these conditions, whatever function you decide to fit, you can infer that the first 25% of the curve is very close to or above 1.0.

 

Ok, end of lessons. After this I start charging.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.


×
×
  • Create New...