Select certain cells in matrix to assign a 0-1 or 1-0 dyad 10,000 times

Hello everyone and thank you for reading my plea for help!
I am a research student trying to calculate the linearity of a certain number of individuals (the subjects of my study).

In my matrix of individuals I have the row individuals dominant to column individuals given values of 1, and where individuals in the column are dominant to those in the row, a value of 0 is given. Where the relationships are "tied" a value of 0.5 is entered. I have more than 30 unknown relationships, which I have filled with "x".

What I have to do, is select these x-cells (Haha, Excels :P) and randomly insert a 1-0 or 0-1 dyad between the corresponding individuals. So if (in the data file) Garth and Nala are to have a 1-0 or 0-1 inserted to their corresponding cells, but they must be opposite; like if row-Garth gets a 1 on column-Nala, then column-Garth gets a 0 on row-Nala.
I have highlighted (in the second matrix) all of the non-x cells. I have to make 10,000 matrices (or at least replicate the application of these random 1-0/0-1 dyads 10,000 times).

I am attaching my matrix file, and also a word file describing what I am trying to be doing after this is managed, as I have to calculate a particular equation after as well...
And I have to do all of these steps 10,000 in order to satisfy probability requirements.

I hope that someone will be able to help me!

Thank you so much everyone for your time taken to read this, and time taken in attempting to help me solve this problem

I think Excel is well suited to solving this, but am not sure I understand the problem.

In the upper matrix of the attachment, your original data is in the UR triangle. The LL triangle uses a simple UDF to convert the corresponding value in the UR, changing 1 to 0 and 0 to 1, and leave everything else ("x" and 0.5) as is. A formula at right computes h for the matrix (2.032)

In the lower matrix, the UR triangle gets the values from the upper matrix, except for values which are x; for those, it randomly choose 0 or 1. The LL triangle uses the same function as the upper matrix. The formula again computes h for this matrix.

The problem is, h is much lower for the lower matrix when all the x's of the upper matrix are replaced with 1s or 0s; it's in the neigborhood of 0.1.

If you can explain what's going wrong, I'll try to help fix it.

EDIT: This isn't an attempt to solve the iteration thing, just a step to understanding the problem.

Ok, I took a look at the file, thanks for your work so far!

What I am trying to do, is change the 'x' values, where they are, into either 1's or 0's.
So, for the pair Garth and Lara, for example, they have not interacted once during my study (hence the x - for unknown relationship).
This procedure (once repeated thousands of times, apparently) is a rough attempt to determine what their winning:losing relationship would be. So, by randomly assigning 1 or 0 to the first cell (rowGarth, colLara, for example) then the corresponding opposite would have to be entered into colGarth, rowLara - to express the winning-losing aspect of this dyad...
In the next round (of the 10,000 repetitions) it could be a 1 or 0 in either spot again..
And this has to be repeated for each x-pair (and there are many :S)

The 1s and 0s, and 0.5s already present in the excel file are known winning-losing relationships (based on my data); the 1s indicate that the row individual is dominant to the column individual. The 0s mean that the column individual was dominant to the row individual.

So, the x's need to be replaced with these numbers (0s or 1s) and then the formula calculated.
and I need to be able to automate it so that it can be done 10,000 times (for probability purposes).

Does that help a bit?

EDIT: I should mention that the second matrix in the file I originally attached to this message (with the yellow highlighting) was for my own ease of understanding just how many x's I had going on...and an attempt at excel functions of cell-colours, etc... But they are otherwise identical.

Second EDIT: Alright, upon reviewing the literature again, I've found that I have to calculate the h-value for the matrix I am observing, then randomly change all the dyad values again (to 1-0, 0-1s) and then calculate the new h, to compare to the old h (from the previous matrix).
I need to calculate how many times the new h is greater than or equal to the old h...
I am going to include the URL for the article (page 9/15 is where it starts talking about what I am trying to do). Find the step-by-step instructions under "An Improved Test of Linearity".

http://www.biol.ruu.nl/behaviour/deV...eVries1995.pdf


Thank you so much!!!