Thanks @Carl Love.

I found that the array that I was trying to use had different dimensions than the one in your procedure (it looks like the Parts and Appraisers numbers are flipped). That is nor a big deal; I can change indices as needed. But even when I define an array that I think should work (i.e., I use the Array command and order things properly), I get "Error, invalid input: ANOVA_3D expects its 1st argument, X, to be of type And(rtable,3 &under rtable_num_dims), but received TestA" and when I try to check the dimensions with ArrayDims(), I get "Error, invalid input: ArrayDims expects its 1st argument, A, to be of type rtable, but received TestA" (TestA is the test array I made).

Any suggestions? Thanks.

j.
 

Hi @Carl Love.

My plan was to insert your code into my worksheet and then call your ANOVA_3D procedure using an array I assembled in the specified format. It seemed like a reasonable plan to me, but when I try to call your procedure using my data matrix, I get the following error: Error, invalid input: ANOVA_3D expects its 1st argument, X, to be of type And(rtable,3 &under rtable_num_dims), but received nm1 [nm1 is my data, by the way]. If I try to use your original function call and set AP equal to nm1, I get an error that nm1 does not exist.

What do I need to do so that your procedure can see my matrix. Is it a format issue? Would you be kind enough to explain the following code: (X::And(rtable, 3 &under rtable_num_dims)) ?

Thanks.

j,

Thanks @Carl Love. I am very impressed. I don't understand all of your code (yet), but I can't argue with the results.

Thanks also for the worksheet. You were correct about the input mode; I did use Ctrl-J or Ctrl-k, but it seems that the input mode was still 2D math. 

j.

Thanks @Carl Love for your efforts in tightening up your code. I tried putting it in a blank Maple worksheet, but I get an error: "Error, invalid arrow procedure." 

Here is some sample data from the MSA manual. In practice, each parameter (appraiser, part, and test) can range from 2 to 10. The number of test values is the same for each appraiser/part combination.

The values below the table are the "right" numbers from the MSA manual,although I can also get them using a rather roundabout approach.

Thanks.

j.

Thanks @Carl Love. Wow. I am very impressed. Your solution is much more elegant than my brute-force approach. It will take me a while to figure everything out, and I will investigate 1D Input.

Thanks!

j.

Thanks @Carl Love.

Yes, I found out how to create a 3D array, but I am wondering now if that is really the best way. It might be good for TSS, but I might have to change what I do depending on the meaning of each parameter. And I think that the notation is probably adding to my confusion. I am an engineer not a mathematician. The dots in place of the indices are shorthand for a prior summation in that dimension. For instance, I should only need to do a summation in the i dimension ultimately for SS_P. And the final term in each equation is the average of all my measurements. 

By the way, A stands for Appraiser, and P stands for Part. I have n meaurements of each part from each appraiser. I suspect that the struggles I am experiencing explain why people just buy Minitab or some other statistical software instead of coding the math themselves. However, I would like to figure out the math.

j.

Thanks @Carl Love for your explanation. Things like defining out were already nagging at me, but most of the things you mentioned were completely new to me.

j.

Thanks @Joe Riel for your suggestion. It is efficient and much more elegant than what I was trying. And you gave me at least three new commands that I will be using in the future.

j.