# Thread: An assignment using G*Power

1. ## An assignment using G*Power

A student I am tutoring in precalc asked me to help him on the following assignment from his Intro to Prob & Stat class. I am suppose to "explain" the assignment to him tomorrow, but I do not understand it myself.
Here's the kicker... I read the email from the instructor and what I have typed below is the instructor's project/assignment description.
Could someone shed some light on what the instructor might be looking for my student to submit?
Thanks!
Mike Bayville
(1) Select an individual population such as an organization of your choice.
(2) Determine the appropriate sample size.
(3) Use the G*Power statistical software to determine the power analysis.

2. ## Re: An assignment using G*Power

Hey fran1942.

The question by itself is meaningless: you need a properly framed question to be able to know what exactly is required.

3. ## Re: An assignment using G*Power

That is a great approach! I will tell my 15 year-old student to boycott the assignment since it is not properly framed and when his teacher asks why he did not submit the assignment, he will simply observe that the question itself is meaningless but will also mention his enthusiasm to do those assignments that state exactly what is required. Why didn't I think of that??? Oh that's write, I read the teachers email in which he presented the project word-for-word as I transcribed it in my first post on this subject. The reality is the #\$%^& question is so poorly framed, so lacking the most basic information student's need for a successful effort, every time I role play my initiating an inquiry with the teacher, my contempt and total lack of respect can be seen in my eyes and other not so subtle body language indicators.
Let me ask you this... How would you approach a high school teacher so that you leave the conversation in possession of that information which in necessary for a successful submission?

4. ## Re: An assignment using G*Power

Well I guess in a statistical problem, you would want to know the following:

1) Assumptions - What models are you using to represent the population, sample, and descriptive models of relationships (like say y = 2x + error)?
2) What are you ultimately trying to do? Are you trying to make an inference based on a parameter for your underlying assumed distribution? What you are trying to compare and how are you both comparing the information and obtaining the information in the first place?
3) Have you used the above plus extra domain information (i.e. specific to the problem and not general stuff) to not only select the above but to check and make sure the results you get make sense intuitively?

The short answer to your problem is that you need enough information to solve a problem and if you don't, then you need to make some sort of assumption.

Its like having a relationship x + y + z = 2 and you are asked to solve this system. If you tried to solve this as it is it would be relatively useless since you have missing information to get a specific answer. If you added a condition x + 2y = 3 then you are narrowing it down but still can't get a specific answer. If y added however z + 2x = 5 then you can use all of the results to get a specific answer for x, y, and z.

If you can in simple terms show that you need minimum said information before you can get the analagous x, y, z and also take into context the stuff I mentioned above, then it should be extremely clear to the teacher not only what your argument is, but why its valid.

5. ## Re: An assignment using G*Power

First, Please accept my apologies for my inconsiderate and uncalled for response to you first reply to my post.

Second, You are obviously a more thoughtful and considerate person than I am. I am humbled by your willingness to not only look past my inappropriate behaviour, but to then provide me with additional assistance, a kindness I certainly do not deserve.