Messing With The Facts

My Father used to tell me growing up that "figures do not lie but, liar's figure". I didn't understand the statement until I became a Statistician and saw first hand how the numbers could be altered, (skewed), to arrive at any conclusion you wanted.

Let's suppose that we wanted the predominant hair color of children to be Brown. To give the "study" more credibility, we'll say that we "randomly" polled 10,000 children. What people do not realize is that the word "randomly" is subjective and is used flippantly by those who wish to make a survey come out a certain way. Of those 10,000 children "polled", 6,000 of them had Brown hair prior to the "random survey". Those children who do not have brown hair for the most part, we're excluded in order to meet or exceed the ratio that we desired. Obviously, we could then claim that a majority of children have Brown hair or whatever color we wanted the majority to have.

Political pollers do the exact same thing. They pre-screen the pool prior to the "random survey" to ensure that the majority of those "polled" will give the "correct" answers to the "random" survey questions. In this way, you can poll any number of people and claim that you got any statistic that you wanted. Then too, those being polled may not truthfully answer the question(s), which may skew the results even further.

What method of testing do you employ?

HawkTheSlayer wrote:What method of testing do you employ?

The method of testing is SUPPOSED to be a "random sample" however, a true random sample will not give the answer that the liberal Democrats want the outcome to be. A true random sample means that you sample the dataset based upon a formula (EVERY 10th household, EVERY hundredth household, or whatever), NOT the pool of data that will reveal the outcome that you want, (i.e. randomly selecting Democrat households as a majority).

As I said, one can skew the dataset to reveal whatever number one wants if one wishes to skew the data for a political agenda.

RV wrote:

HawkTheSlayer wrote:What method of testing do you employ?

The method of testing is SUPPOSED to be a "random sample" however, a true random sample will not give the answer that the liberal Democrats want the outcome to be. A true random sample means that you sample the dataset based upon a formula (EVERY 10th household, EVERY hundredth household, or whatever), NOT the pool of data that will reveal the outcome that you want, (i.e. randomly selecting Democrat households as a majority).

As I said, one can skew the dataset to reveal whatever number one wants if one wishes to skew the data for a political agenda.

Well I realize that. You posted you were a statistician.
Z-test, T-test, chi-square, etc?

Some years ago I had to organize and supervise polls for the left before a federal election and the list of respondents to call was supposed to be 'anonymous' but every detail about their lives was on record including who they vote for and, as such, the outcomes were skewed to the result they paid for and wanted.

Incidentally it's something both sides do and each for their own political agenda. The left just does it more.

HawkTheSlayer wrote:

RV wrote:

HawkTheSlayer wrote:What method of testing do you employ?

The method of testing is SUPPOSED to be a "random sample" however, a true random sample will not give the answer that the liberal Democrats want the outcome to be. A true random sample means that you sample the dataset based upon a formula (EVERY 10th household, EVERY hundredth household, or whatever), NOT the pool of data that will reveal the outcome that you want, (i.e. randomly selecting Democrat households as a majority).

As I said, one can skew the dataset to reveal whatever number one wants if one wishes to skew the data for a political agenda.

Well I realize that. You posted you were a statistician.
Z-test, T-test, chi-square,  etc?

I used to be a statistician. That was a long time ago and no, I wasn't ISO 9000 certified. When I was a statistician, ISO 9000 didn't exist for the masses. Even if it had existed, I doubt very much that they would have paid for a Supervisor to go through the certification process the Union would have had a fit!