Sunday, July 31, 2005
Test of Hypothesis and Law and Order
My back has been killing me all weekend. Not much fun. Just sitting around waiting for the pain to subside. Catch up on reading. Post a few thoughts. And watch mutliple episodes of Law and Order-SVU. What a great show! I'm hooked. (I'm not a big TV watcher).
On a daily basis, it appears (after 4PM) either on USA Network or TNT. Today, I watched three back to back episodes on USA.
So how does Law and Order relate to Test of Hypothesis? For those of you who have taken my CQE course, you probably know how. Test of Hypothesis is a statistical technique used to make inferences about populations. (Single or more than one) The tests can be used to test assumptions about population means and/or variation.
At first, the tool escapes people. The concept takes a while to sink in. Several years ago, I heard the following analogy presented. This helped bring clarity to me.
The test is comprised of a null hypothesis and a test hypothesis. The former is what you assume to be true about the population parameter. The latter is what you are trying to prove. Compare this to a court case. Before a case is tried, the defendant is assumed to be innocent. Until sufficient evidence is collected and presented, we must hold on to this assumption. This is beautifully analagous to the null hypothesis.
The prosecution is trying to get a guilty verdict. Their job is to present enough evidence to reach the verdict. So this equates to the test hypothesis.
In court, the jury can convict a defendant or acquit. An acquittal does not mean that the defendant is "truly innocent". It just means that based on the evidence presented in the case, the decision was "not guilty". Perhaps if additional time was spent and extra witnesses called, more incriminating evidence could lead to a different verdict. The same applies to tests of hypothesis. You state your assumptions, collect the data (evidence) and reach a conclusion. The conclusion could be to:
1. Reject what is assumed to be true (guilty verdict)
2. Fail to reject the assumed truth. You can't unequivocally say the assumption is true because more evidence could prove otherwise. (Acquit)
Minitab is great for conducting the tests. Once you know the math, Excel can also be used. You can find test templates and data sets (for practice) at http://www.qualitymindsinc.com
Please help yourself!
Subscribe to:
Post Comments (Atom)
4 comments:
Whew, this particular subject can be a bit confusing. I tend to stand on the prosecutor's side. I have the tendency to dig to prove the hypothesis to be false. Then there are times when even I can get a bit swayed. So no matter which result you end up with, there are always opportunities that can lead you to believe that the data may have been skewed to some degree. So try to to look at the hypothesis from several different angles and get other inputs because what you thought may have seemed correct could result in a decision that may not be the true answer.
There is always the risk of data being skewed. This is why sampling is such an important topic. It is one, however, that is mostly taken for granted. "Just go grab some parts" is not a sample.
Without understanding the population, the sample can't be pulled correctly. Without a correct sample, the data will probably be skewed.
Hello Stephen,
After reading your blog "Test of Hypothesis and Law and Order", I believe you will find the wages information on my site about Jobs With Best Wages to be very helpful.
To give you an idea of our wide range, some of the recent searches that found our site included ...
Electronics Engineers Wages
Engineering Technicians Wages
Engineers Wages
Environmental Engineers Wages
Environmental Technicians Wages
Health & Safety Engineers Wages
Industrial Engineering Technicians Wages
Industrial Engineers Wages
Landscape Architects Wages.
We have hundreds of "must read" articles on wages and careers and many other topics in our Top Career Wages site.
Kind Regards
Emily
Hello Stephen,
After reading your blog "Test of Hypothesis and Law and Order", I believe you will find the wages information on my site about Jobs With Best Wages to be very helpful.
To give you an idea of our wide range, some of the recent searches that found our site included ...
Public Address Announcers Wages
Public Relations Specialists Wages
Radio & Television Announcers Wages
Technical Writers Wages
Radio Operators Wages
Reporters & Correspondents Wages
Set & Exhibit Designers Wages
Writers & Authors Wages
Sound Engineering Technicians Wages.
We have hundreds of "must read" articles on wages and careers and many other topics in our Top Career Wages site.
Kind Regards
Emily
Post a Comment