Extract meaning from data ordered according to the frequency Let’s suppose I got the following answers for the question "list the social networks, in order of importance, you generally use":
            Most Used   Second Most Used   Third Most Used
Person 1    Facebook    Instagram          Linkedin
Person 2    Facebook    Instagram          Linkedin
Person 3    Tumbrl      Facebook           Flickr
Person 4    ...           ... 
...
Person 90   Facebook    Linkedin

What kind of information can I extract? What is the best way to visualize it? 
Let's say that only one person listed Tumblr, as the most used, and all other people listed Instagram as the second most used. Does it mean that Tumbrl is more important than Instagram?
 A: *

*A multi-level Sankey is one nice way to visualize this information, where each vertical level represents most-used, second-most used..

*Ideally, one would have preferred time spent on each social network by each person within a day. This would give you a clean set of numbers. In categorizing them as most-used, second-most used, we lose this information, and may have to make a few assumptions

*You can begin by assuming that each individual spends about the same time in a day on social networks, and assume some proportion of time distributed between the three usage types (e.g. 60% most used, 30% second most used, 10% third most used - does not have to add up to 100%). Adding this up across all people by social network and calculating the relative proportions will give you a sense for how they are ordered relative to each other (and may partly address the scenario with Tumblr that you have described)

*You can now change the % numbers assigned to most used, second-most used, while maintaining their ordering (i.e. most used will have a higher %usage value than second-most used), and see how robust your final ordering of social networks is to changes to your assumptions. You can always present/interpret the results as applicable under a given set of assumptions about usage

*If you now want to take this to its logical extreme, you can simulate random numerical assignments for usage time assigned to each of most used, second-most used, under the constraints that they are ordered, and total usage is within a threshold (for e.g. if the above reflects daily total usage per person, then say <17 hours), get the usage values for each social network, and average across simulations (along with standard deviations) and see how they are ordered relative to each other. 

A: A plan should be


*

*consider an answer as a 'match' between social networks

*randomize matches' order (of insert them in time order, if any) and insert them in a rating system engine. 


If you're interested in use more than in development, rankade, our free ranking system for sports, games, and more, allows matches with both 2 and 3+ factions, like TrueSkill and opposite to Elo and Glicko, that work just for one-on-one (here's a comparison).
In addition, rankade has a weight feature (all answers have same impact?) that might refine your work.
