0

My first question. I have searched "effect size" and followed the links on "similar questions". There may well be a wealth of relevant info there, or even the answer; my problem is that the discussion is way over my level of understanding. Apologies if that means this is in some way duplicate. I'll try to describe what I'm trying to work out. If the answer is already floating around, it'd be wonderful if someone could point me at it and dumb it down into language that a stats layman can follow!

I've got data about the population of England, carved up into Local Authority areas (i.e. geographically) - there are 127 rows of info. For each row, I have:

  1. the size of the population of interest (over 60s)
  2. a score 1-5 for deprivation
  3. a categorization of policy 1, 2 or 3. (1 is "no policy"; 2 is "accessible for free"; 3 is "pay for access")
  4. a frequency rate of an occurrence/100,000 of the population of interest.

So a record might be: Bristol; 2; 1; 1,300

Happy to post all 127 rows if helpful!

Some excellent help elsewhere has got me to the point where I know that the groups by policy are significantly different (p<0.01). So policy makes a difference. This used ANOVA (but not by me!). I also know that deprivation correlates somewhat with frequency (R^2=0.25 apparently)

What I'm trying to get is how much of the total variation in the whole dataset is accounted for by policy, how much by deprivation, and how much by other factors as yet unspecified.

I want to be able to say "Group 1 is different to group 2 and group 3, and the policy accounts for X% of the overall variation, whilst deprivation accounts for Y%".

My knowledge is very limited. I can calculate weighted means...but I'm not even clear if my standard deviation calculations are right, because the populations are all different sizes.

I'd be very grateful for any help you might be able to offer, but it will need to be at a "add/subtract" level of complexity - I don't have the foundation tools to follow anything sophisticated unless it's broken down to basics.

Thanks again

Bilberry
  • 1
  • 2
  • Hi Robert. Thanks for the pointer. The title looks like it might be relevant..but I don't understand most of it, and especially the section of code (I don't code), so not sure I've explained well enough how basic my skills are! – Bilberry Feb 20 '21 at 11:19
  • 1
    Try the suggestion in the third answer by @StephanKolassa from the question linked in the first comment. You will have to fit 3 models. One with both variables and one for each of the single variables. You said one was already done that shows policy explains 25% of the variability. – John L Feb 20 '21 at 12:30
  • Thanks John. Is there a worked example of "fitting a model"? I just have the rows of data described and am unclear how to proceed – Bilberry Feb 20 '21 at 15:03

0 Answers0