# Excel 2007 : Choice of statistical method

1. ## Choice of statistical method

I want to find whether there is a significant difference between 9 groups with regard to one dependent factor.

The groups are divided based on how many times they have reserved a service (independent variable X). The dependent factor is a score which goes from 0 to 100. My hypothesis (H1) is that the higher the number of received services (X) is, the higher the score will be (dependent variable Y).

Hence, both the X and the Y are continous variables.

My question is which statistical method I should use to figure this out? And also I will probably need help with the method in excel and the interpretation of the results...

The data are attached.

The groups are highly variable in size, where the majority have received a low number of services.

2. ## Re: Choice of statistical method

I thought the method was to ASSUME there was NO difference between the two groups and see what the probability is that your assumption was true. This lead to the .05 and .01 stuff, which meant the probability that there is really no difference between the two groups is only 5% chance or stronger the 1% chance that they are really the same.
http://www.qimacros.com/qiwizard/hyp...ing-excel.html may help showing 3 types of tests
I like examples so http://www.mathematics-statistics-so...om/Sample7.doc might help.

I haven't even looked at your data because you have to select the correct method based on your data and what you want to accomplish. I hope the above links help.

3. ## Re: Choice of statistical method

Right, I could have formulated it better but as you may see I named this hypothesis H1. Statistically, I am aware of that the H0 hypothesis is that there is no difference. But of course I have a theory that there actually is a difference (or else I would not bother to investigate it, would I?)

However, what I am asking for is exactly which method I should choose to investigate this question - whether or not there is a significant difference...

Originally Posted by MarvinP
I thought the method was to ASSUME there was NO difference between the two groups and see what the probability is that your assumption was true. This lead to the .05 and .01 stuff, which meant the probability that there is really no difference between the two groups is only 5% chance or stronger the 1% chance that they are really the same.
http://www.qimacros.com/qiwizard/hyp...ing-excel.html may help showing 3 types of tests
I like examples so http://www.mathematics-statistics-so...om/Sample7.doc might help.

I haven't even looked at your data because you have to select the correct method based on your data and what you want to accomplish. I hope the above links help.

4. ## Re: Choice of statistical method

That was the one thing I took from my Stat class 40 years ago. If you want to show there is a difference, you need to assume there is no difference and then give a probability that you are wrong. This was a real stretch for my brain's logic meter. It was like proving Root(2) is irrational. You needs to assume it is rational and then show your assumption is wrong. In this probability stuff you need to assume there is no difference and do the bell curves, correlations coefficients and standard deviations to decide the probability that your data is really the same. If the probability is so remote that they are the same, you conclude there is only 1% chance they are the same based on the samples you extracted from both.

After a little study at http://office.microsoft.com/en-us/ex...005203066.aspx and remembering my class, I remember the method we used to solve this weeks problem was to use the method in the chapter we had last worked on.

I want to find whether there is a significant difference between 9 groups with regard to one dependent factor.
Is like asking which method of statistics should I use to convince myself (and others) that my conclusion is right.

Look at the attached with a variety of stats using pivot tables and charts. Do any of these convince you there is a difference? If not then what process (in statistics) do you need to use to show what you are talking about? Lots of formulas. Which one to use based on your data is the problem.

I'm sorry that I don't have THE answer. It is like asking a person - "What method do I need to show you my conclusion is true?". Look at the attached and see if you need to set up your data in a different format to extract what you are seeking. Also look at the discription of all the Excel Stat functions to see all the different methods people have concocted to show how numbers relate to each other will solve your problem.

5. ## Re: Choice of statistical method

=CORREL(C2:C6245,D2:D6245) returns 0.042, very weak positive correlation.

If you plot col B vs col A, it's visually apparent.

There are currently 1 users browsing this thread. (0 members and 1 guests)