I’ve got a data set that ask respondent to indicate where they work most, as a percentage – e.g.

Site 1 – 20 – 10 – etc
Site 2 – 30 – 10
Site 3 – 0 – 10
Site 4 – 50 – 10
Site 5 – 0 – 20
Site 6 – 0 – 40

n = 140

To get the average should i use mean, median or mode? It has been a long time since i’ve dealt with numbers

Depends what you’re looking for out of the data? Stats will tell you anything you want to hear D:

edit: “Average” is a very vague term.

Mean

Why mean?

Seriously, i don’t get stats.

Mean is the term for the statistical thingy that us layfolk think of as average. ( I think)

Mode is the most common

Median is the one in the middle in a sort of geometrical way

I think

The mean is what most people refer to when they say ‘average’.

Median is the middle value in an ordered sequence

Mode is the most common value in a set.

You probably want the mean = sum up all values and divide by n
but would help if you gave more info

Thanks. I get the definitions (even i remember back to my GCSEs), i was just wondering what was most appropriate. More detail:

Each of the sites represents part of the UK (English regions plus the ‘nations). The idea is to use the percentage data to map where this group of people work most. If i get averages the total of the means is 150. Can i then convert the means in % of the total of the means (150), or is that just stupid?

I am pretty syre you will find the only relevant measure is the mode.

Site 1, Site 2 etc are ordinal measures so they can not be used to form an arithmetic mean. Similarly the order 1 , 2 , 3 etc is irrelevant as the sites could have been asigned numbers in any old fashion.

The answer is pretty evident from the question IMHO ” … where this group of people work most …” directs you to the mode.

Simply add up the numbers for Site 1, Site 2 etc respectively the one with the biggest value (assuming no shared results) is the one (ie. Site) that is used most.

Sounds like common sense has gone out of window with people getting hung up on descriptive statistics defintions.

Trust me I’m a Doctor 😆

Thanks GJP, that makes sense. These are the totals for each place:

386.5 5051 311.5 674.5 87 127.5 89 193 1791 47

Can i turn these into percentages to map without insulting the Gods of statistics?

Or do i just rank them?

Agree with GJP, dredging up from my uni statistics course, mode seems the most appropriate and was the one I was about to recommend before I saw your post.

All I can see from your numbers is that your respondents are spending, relatively, nearly all their time at just two sites, with the rest pretty negligible in comparison. If you’re looking for a easy way to represent that, consider a pie chart. This is roughly what it looks like with your numbers:

Cheers for that. What i’m thinking is arrows of different sizez from base region to sites of work. Either that or circles over each place. I’m a geographer, so got to have a map innit.

There was a small error in my post above it should have read nominal and not ordinal – but does not seem to have confused anyone.

Flaperon’s post seems like a sensible approach to me.

Transcript analysis is so much easier than this.

Thanks for the pointers guys.