The Null Hypothesis

The book mentions the Null hypothesis and we would like to know when we might come across something like it. How could one have populations whose means are equal and how is it due to random error?
The book also mentions degrees of freedom. What is this and how does one determine the degree?

McDonnell, Nieswiadomy, Painter

The null and alternate hypotheses represent potential situations that might occur when evaulation the strength of a relationship between variables in an experiment that usually invovles looking at the sample or population means. The null situation states that the population means of the two groups being tested are equal and there is no relationship between the two gropus, and any difference observed may be to randomized error. The alternate condition suggests that there is in fact a difference between the the population means for the two groups and that there is a relationship between the two groups that is not subject to a randomized error as determined by the degree of statistical significance...So in attempting to answer your question, Yes you will come across situations where the null hypothesis prevails as the population means are in fact equal between the two groups you are testing and thus no relationship exists between the populations you are testing.

Clark, Caroll

I read the above messages about null hypothesis, but I
still do not understand exactly what null hypothesis is
or does. Can someone please explain it to me?

All populations are link in somes ways. The American population all do have common experiences,throught the T.V., standardized education practices, or just the psychology of being American. So the Null hypothesis is blown out of the water. It's foolish to think that throughout the Behavior Research tools of dividing up the sample populations and simple writing off the common experience that all people have in common. If you breath the planet's air. and take the sun's warmth, than we all are the same. So no study is without some fault, no matter how random the assignment is or how double blind the people invold in the study are. I must admit that some theories of Behavior Research are unclear, which some of you will endeavor to see. But I see that sometimes Research go out of their way to elimate the vast bonds that tie and bond an society together. All statistical, all those pretty little number do bleed to together.

I still don't understand the purpose of the null hypothesis
or when and how it's used. I think I understand type I and
type II error from the answer given by Schepis. Could that
same person please give me an example of the null hypothesis
and the decision matrix? Also, on page 206 in the book
figure 11-2 (decision matrix)can someone explain that
example with numbers.

Altough I understand what the null hypothesis is, I do not understand the value of proving it false, as opposed to proving the research hypothesis true. I think that it maybe because science does not accept anything as "true". So instead of saying something is true, we say the opposite is false. Is this correct?

In reference to Pryors last statement, I feel that many researchers think that it would be easier to attempt to prove something as false rather than to disprove something that has already been accepted as "true". Most researchers would rather save their reputation than to attempt and persuade others into believing something that may seem outrageous to them.

Also, I was wondering why researchers "traditionally" use a significance level of .05 to .01 to reject the null hypothesis? The book explains it, but the answer is cloudy to me.

OK, Kimberly, put me on the spot, huh? Sorry that I haven't responded until now, but I just looked at it again. I can give you an example of the null hypothesis off the top of my head, but as for the decision matrix stuff, either someone else needs to tell you that, or I need a day or so to look it all over.
The Null Hypothesis states that whatever effect that you found in your experiment is just due to random error, and not due to your manipulation of the independant variable. Here's an example. Let's say that you organize an experiment that takes two groups that are as equal as possible, with one as the control and the other as the experimental group. On the control group you just run a test for verbal ability twice, with no true manipulation in between. In the experimental group, you run the test, and then give them a class on improving their verbal abilities - kind of like that Word Power crap. Then you test the experimental group again.
If the null hypothesis is true, then whatever improvement that you found in your experimental group is just due to random things, but not due to the class. This error could be that some people in the experimental group had a better day than the control group. Thus, the some people in the control group couldn't have cared less about the experiment, and as such they did worse. But, this is due to chance, NOT the manipulation. Since only chance affected the experiment we say that we have failed to disprove the Null Hypothesis.
IF the experiment showed a signifigant change in the verbal scores on the experimental group, then we would have disproven the Null Hypothesis. To figure out which to do, statistics must be used. Hope that that helps!!! If it doesn't let me know.

Ty Schepis