Things you like that most other people don't like that you're not proud of

[KrazyHorse]

Omniscience. Duh.

[Japher]

You'd think omniscient people would know that I can't read minds and whether or not they look fat and that

[chequita guevara]

But one would think that sampling involves samples, not the whole, i.e., I take a number of random people to make observations about the population as a whole, not try to physically count each and every single person in the USA.

That's simply because you have absolutely no training in probability or statistics.

That may be true, but every definition of statistical sampling I look up seems to indicate that sampling is done because counting every individual in a population is either impracticable or impossible and involves smaller numbers than the whole, i.e., you are trying to infer things about the whole by examining a sample. If you can count the whole, then sampling is pointless, as you have the data about the whole right in front of you.

[KrazyHorse]

Che, the set of all possible states is called the "sample space". A sampling method which gives you perfect info on the state is still a sampling method.

You can also have samples with lower information value.

Also, some quick searches seem to imply that the political interference, as it were, has to do with which methodologies the bureau may or may not use in reporting figures for reapportionment or redistricting purposes, not with which figures may be reported to the public.

[Kuciwalker]

KH, che is right here. Last census there was a big controversy over a proposal to let the census bureau use "sampling" (as it was called by the media), i.e. a particular scheme for adjusting counts based on a more restricted followup survey. The controversy was partly constitutional but largely political, because there would have been electoral considerations. When che says "the census bureau could not use sampling", it was immediately obvious to me (and probably to anyone else who lived through the issue) what he was talking about.

Also note that the Wikipedia article for census explicitly distinguishes it from "sampling".

You owe him an apology, dude.

[RGBVideo]

summary of this epic battle?

[KrazyHorse]

KH, che is right here. Last census there was a big controversy over a proposal to let the census bureau use "sampling" (as it was called by the media), i.e. a particular scheme for adjusting counts based on a more restricted followup survey. The controversy was partly constitutional but largely political, because there would have been electoral considerations. When che says "the census bureau could not use sampling", it was immediately obvious to me (and probably to anyone else who lived through the issue) what he was talking about.

Also note that the Wikipedia article for census explicitly distinguishes it from "sampling".

You owe him an apology, dude.

a) I don't particularly care what the media called ANYTHING. "sampling" is any process by which information is drawn from the sample space. No ****ing statistician would call what I'm assuming he's referring to as "sampling" "sampling". He would call it "statistical adjustment" or "statistical imputation" or some more technical term. I asked for more specifics as to what, precisely, he meant by saying that there was Congressional interference and he said that the agency was prevented from using "sampling". It was at this point that I decided Che didn't know what the **** he was talking about and was simply parroting claims he'd read somewhere.

b) The wiki on the 2000 american census itself refers to the controversy as "adjustment". http://en.wikipedia.org/wiki/2000_Un...nt_controversy. It also states that THE INTERFERENCE CAME FROM THE SUPREME COURT, NOT CONGRESS.

c) Both the ACS (which I'm pretty sure is where Drake's numbers come from) as well as the 2000 Census long form used "sampling" techniques whereby only a small number (relative to the whole) of individuals were questioned. The short form did not include ethnic origin questions which would reveal Lebanese ancestry specifically

[Drake Tungsten]

Both the ACS (which I'm pretty sure is where Drake's numbers come from)

The 45k number came from the ACS. The earlier 44k number was from the 2000 Census.

[Kuciwalker]

a) I don't particularly care what the media called ANYTHING. "sampling" is any process by which information is drawn from the sample space. No ****ing statistician would call what I'm assuming he's referring to as "sampling" "sampling". He would call it "statistical adjustment" or "statistical imputation" or some more technical term. I asked for more specifics as to what, precisely, he meant by saying that there was Congressional interference and he said that the agency was prevented from using "sampling". It was at this point that I decided Che didn't know what the **** he was talking about and was simply parroting claims he'd read somewhere.

He was using the term as it has been used by most people in relation to the issue at hand. You weren't familiar with the issue and interpreted the term from a strictly statistical context. Your bad.

Moreover, you didn't address my other point; that when talking about censuses in general, apparently "sampling" refers to exactly what che is talking about. (inference from wiki's use of the term in http://en.wikipedia.org/wiki/Census)

And I believe che did, in fact, know what he was talking about; the essence of the controversy was whether the Census Bureau could only use the exact tally of responses as their official count (a census according to wiki's usage) or if they could use statistical inference from the responses to compute the official count (sampling according to wiki's usage), and his comments were all consistent with that usage.

b) The wiki on the 2000 american census itself refers to the controversy as "adjustment". http://en.wikipedia.org/wiki/2000_Un...nt_controversy. It also states that THE INTERFERENCE CAME FROM THE SUPREME COURT, NOT CONGRESS.

His claim that the Republicans sabotaged the 2000 census has nothing to do with you ignorantly jumping on his correct terminology.

[KrazyHorse]

He was using the term as it has been used by most people in relation to the issue at hand. You weren't familiar with the issue and interpreted the term from a strictly statistical context. Your bad.

No, not my bad, you twit.

When I ask somebody for information on a subject (in good faith) and they provide me with a buzzword they picked up from some retarded reporter instead of actually providing me with information then it indicates that they have no real knowledge of the subject.

Moreover, you didn't address my other point; that when talking about censuses in general, apparently "sampling" refers to exactly what che is talking about. (inference from wiki's use of the term in http://en.wikipedia.org/wiki/Census)

Actually, the "sampling" that che was talking about appears to be a FOLLOWUP SAMPLING OF A SMALL NUMBER OF REPRESENTATIVE BLOCKS FROM WHICH RESPONSE RATES ARE IMPUTED AND STATE POPS ADJUSTED. Calling this "sampling" and expecting it to convey useful information is ****ing ridiculous.

[KrazyHorse]

Here's the back-and-forth:

Could you please be more specific as to what "modern statistical methods" were forbidden?

Among the methods they were not allowed to use was sampling.

"Sampling" describes EVERY evidence-based statistical method. It's like saying that scientists aren't allowed to use experiments.

The GOP Congress only allowed the Census Bureau to count those who responded to the questionnaire or an interviewer. No other methods were allowed to be used.

(This is completely wrong, by the way; the Census Bureau imputes the residency of houses where people do not answer the door but in which there are additional signs of residency such as mail being picked up, newspapers delivered or neighbours' statements. Secondly, the long-form questionnaire is only sent out to a small number of households, and the ACS to an even smaller number)

Che, what do you think it's called when you send out questionnaires and count the responses?

I'm just going by what both the Bureau and the GOP leadership said. If counting on your fingers is sampling, then the word has no meaning.

At this point it has become obvious that Che has no idea what the meaning of the "sampling" buzzword as it was used by the media at the time (I'm supposing, by your statements) meant. He might as well have told me that the Census Bureau was prevented from using asdw;ofd for all the information that provided. In fact, if that had been the buzzword then I could at least have done a search on it, whereas the term "sampling" is so generic as to be even LESS meaningful.

The fact that "sampling" was abused as a term for a few weeks during a tempest in a teapot a decade ago is NOT a basis to call it "correct terminology" in the context of somebody genuinely asking for information as to what happened.

[ricketyclik]

I understood what you were saying Che, and your version is what I recall reading in the press in Australia.

Moreover, our Bureau of Statistics routinely refers to the practice of interviewing a few thousand and inferring stats for our whole nation as "sampling".

And

sam⋅ple  /ˈsæmpəl, ˈsɑm-/ noun, adjective, verb, -pled, -pling.
–noun 1. a small part of anything or one of a number, intended to show the quality, style, or nature of the whole; specimen.
2. Statistics. a subset of a population: to study a sample of the total population.
3. a sound of short duration, as a musical tone or a drumbeat, digitally stored in a synthesizer for playback.

–adjective 4. serving as a specimen: a sample piece of cloth.

–verb (used with object) 5. to take a sample or samples of; test or judge by a sample.

http://dictionary.reference.com/browse/sampling?o=0

[KrazyHorse]

Moreover, our Bureau of Statistics routinely refers to the practice of interviewing a few thousand and inferring stats for our whole nation as "sampling".

That is, in fact sampling. That is not what Che was referring to originally. Nor does it imply that census-taking is NOT "sampling"

Statistics. a subset of a population: to study a sample of the total population.

a) "The whole population" IS A SUBSET OF THE WHOLE POPULATION

b) The census does not reach the whole population anyway. So it even studies a proper subset

Care to continue to provide comic relief?

[ricketyclik]

KH, you are technically correct, but not according to common usage.

The comic relief is usually supplied by the pedants, if in a slightly exasperating form.

[KrazyHorse]

You're out of your element, Donny.

This entire discussion stems from the fact that Che used the general term "sampling" to represent a SPECIFIC FORM OF STATISTICAL ADJUSTMENT which the Census Bureau was prevented from using (by the SC in reality, note).

The problem is that "sampling" means anything up to and including the counting of noses (which is NOT all the Census Bureau does). Thus saying that they were prevented from using "sampling" is ridiculous.

In order to determine what Che ACTUALLY meant took a few minutes of internet search in which the term "sampling" provided absolutely NO HELP WHATSOEVER because it was so mindbogglingly vague.