How to compare ranked list items

New Member
I have compiled a survey asking respondents to rank social media by the amount of time they spend on each of 10 platforms. So if, for example, I followed Facebook mainly and spent a little time on Twitter I would rank Facebook 1, Twitter 2 and leave the other 8 options unranked.

The results are compiled into a spreadsheet showing ten tables - one for each ranking. So in the above example if I were the only person to complete the survey table one would show Facebook 1 and all the rest 0; table two would show Twitter 1 and all the rest 0.

My question is how can I manipulate the results to get a realistic feel for which two or three social media platforms I should support. I don't have time to support more than that but want to have the biggest possible reach.

My initial thought is to multiply the scores in table one by 10, then table two by 9 etc. Then total up for each platform. So in the example Facebook would score 10 and Twitter 9 as I am the only person to complete the survey. If I then report on the ratio of the 10 total scores to the grand total, would I get a fair representation of where to apply my future efforts ?

If you've at least read this far, thanks.

fitz0r

New Member
Are you considering any other variables such as the cost or investment required to support the platform? If your only consideration is the survey response then how about prioritizing based on the number of users who indicated spending any time at all (as evidenced by providing any number other than blank). Some simple tests might be what proportion of users who ranked both Twitter and Facebook ranked Twitter higher and is that proportion significantly different from the proportion that ranked Facebook higher? I don't know if it would be informative to do that for every pair of 10 choose 2 = 45 but it might make sense for the top few based on number of users who gave it a non-blank ranking. It might be interesting for you to share your data if you did not have a reason to keep it private.

New Member
I don't think I fully understand your suggestion fitz0r but perhaps my attaching the actual data will help. This is a spreadsheet with 10 blocks, the first block shows how many respondents selected each of the 10 social media as the one they followed most. Block two shows the media they followed second most. Note that some of the 120 respondents will have selected none of the social media. Only true die hard mouse potatoes will follow all 10.

My aim is to know where best to channel resources in maintaining a social media presence. Incidentally, this relates to a non for profit film club I am setting up - I want to know how to communicate which films are to be shown and when to those that signed up.

fitz0r

New Member
Looks interesting Dumbasadog (it feels weird to type that sorry!) Thank you for sharing the data. To try to put it in concrete terms my thought was around adding up the total marks for each social network like so:
Flickr (8) 4
Instagram (7) 13
Pinterest (4) 14
StreetLife (9) 19
Tumblr (6) 3
WhatsApp (10) 40

Now you can see Facebook and WhatsApp are looked at by the most people. The idea being if someone thought it was worth enough of their time to give it a number that is more significant than the actual number they gave it --I don't know anything about survey design but I would guess there is an appreciable amount of variability in these types of rankings being self-reported.

Another thing to look at is the amount of people who indicated a first choice (86), second choice (65), third choice (42), etc. If I support platform A over platform B, how many people who ranked B more highly than A could miss the message? Is it good enough to pick the top 2-3 platforms -- that is my gut feeling anyway.

I would be interested to hear your thoughts and what you decide.

New Member
Well my original thinking was that weighting the preferences was worthwhile. Say for example 150 people rated Twitter as 6th choice and 50 chose Facebook as 1st preference; well I would get more bang for my buck by concentrating on Twitter rather than Facebook. But if I weighted them as previously described, Twitter would score 5 * 150 = 750 whereas Facebook would score 10 * 50 = 500.

Your point about self reporting inaccuracies is well made. I should also perhaps factor in the amount of time I would need to get up to speed with each as well as time to maintain it.

Thanks for your thoughts. I'll see if anyone else cares to comment before deciding how to handle this.