I need somebody who really know about statistics

How to do a Goodness-of-Fit Test (GOF) and Read p-values Class:

You will notice that I have a lot of pictures of normal distributions in the Power Point presentation and for good reason. If your data is normally distributed (sometimes called parametric) or not normally distributed (sometimes called non-parametric or distribution free), the researcher must state this in every research paper. The reason is that the type of distribution determines what type of test (method or formula) will be used to analyze the data. Parametric data may use t-tests, z-test, and ANOVA (analysis of variance). Non-parametric data requires Mann-Whitney U tests and Kurskal-Wallace tests. Regression may use either but still the distribution must be declared.

So, determining if your data is parametric (normal) or non-parametric is important.

The easiest way to do that is to use Megastat, Descriptive Statistics and check the last box that says “Normal curve goodness of fit.”

That is a Chi-Square Goodness-of-Fit test developed by Karl Pearson. It will tell you if your data is drawn from a normal distribution.

The answer is presented to you as a p-value, a very important number in the statistics world. (The p-value is not the results of a urinalysis test). It holds the answer to if your data is normal and is also used in a t & z tests to answer your research question.

Normal does not mean good and non-normal bad. It is just the shape of the data.

So you don’t have to guess if data is normal.

The hypothesis is set up thus:

Null Hypothesis Ho: The data is parametric (normal).

Alternative Hypothesis H1: The data is non-parametric (not normal).

The way you read the p-value is to compare the p-value to your alpha level of significance. Alpha is set at the beginning of the research as .05 (important) or .01 (very important).

If the p-value is smaller than your alpha level you reject the null hypothesis (your data is non-parametric). If the p-value is greater than your alpha level you retain the null hypothesis (your data is normal).

Megastats makes it easy with colors. Bright yellow means the p-value is smaller than the alpha (reject). Black and white means the p-value is larger than the alpha (retain). Dull yellow means the p-value is between .011 AND .049, so you would retain at the .01 alpha level or reject at the .05 alpha level.

If the p-value = 1.56 E-5 that is Megastat saying the p-value is .00000156. It is 1.56 with 5 zero’s. A very small number. It is also bright yellow. Reject. There is a significant difference.

I hope this helps.

Dr. Loro