# Not sure how to analyse this Rank-Order Data

#### MsfStl

##### New Member
Good Day to All,

I am not sure where to begin in analyzing some data I have.

DATA: Sample population N=100
Looking at factors that helped or hindered participation in a previous study.
Ex. Reasons that made it hard for respondent to participate. Lists 10 reasons numbered 1-10. Respondent is asked to rank the reasons in order of importance to the respondent from 1 = most important to 10 = least important. Also, the respondent was allowed to rank ties, i.e., having two or more reasons as the same importance, AND the respondent was also allowed to say that a reason had no relevance, and that reason was coded a 95.

I hope that explained the situation for how the data was collected well enough.

OBJECTIVES: What I would like to do is: 1) to determine the order that the reasons were ranked by the entire sample, 2) split the sample into two groups based upon an independent variable and compare the two groups for differences.

WHAT HAS BEEN DONE THUS FAR: Originally I ranked the questions by frequencies (the number of times a particular question was ranked 1st, 2nd, and so on); and also by a rank sum procedure where I summed the total ranks both with and without the 95 codes. This gave me a rank sum total whereby, I determined that the lowest Rank Sum question was deemed the most important (i.e. say a five question example ranked 1 to 5, [Q1:{5(1), 2(2), 1(3), 2(4), 0(5)} would sum to 20, Q2:{4(1), 3(2), 0(3), 2(4), 1(5)} would sum to 23, etc] so that Q1 would have a higher level of importance that Q2 and so on), additionally I performed the same rank sum while including the 95’s.

Now that I have done that, I am thinking that my methods for analysis are incorrect, but honestly, I am at a loss as to say why or understand what I should do either next, or differently.

What is really throwing me off are the 95s (no relevance). If I do not include them then the particular question that may have several 95s may actually be counted as having a higher importance level because the total would be lower, I in effect create a 0 ranking which would show higher than a 1. If I include the 95 I have the opposite effect, I penalize the question heavily.

When looking for Rank Order examples I find the Spearman Rank Order Correlation Coefficient, or some things I just can’t seem to understand. The problem I run into with the Spearman Coefficient is the examples are made for two variables (x,y), I can see that as a test once I have my populations separated dichotomously, but I do not see how to translate that across the entire sample of 100. Basically, I am unsure that my foundation is correct and wish to understand first things first.

What steps should I take? What would be a good resource for information on understanding this/these analysis procedure(s)?

Also, I use SAS 9.1 for my frequencies and can utilize the package for other procedures if someone can walk me through.

Any help is appreciated and I thank you for your time.

Scot

#### MsfStl

##### New Member
Ok, well can someone tell me if I have the right idea or not?