0
$\begingroup$

I have a set of cities ranked 1-10 based on several criteria. I also have the percentage of the total population living in these cities. I want the cities with smaller populations to be able to compete with larger cities, so I want to give the population percentages a weight that will compensate for smaller population size. How do I do this?

$\endgroup$
  • $\begingroup$ I suggest you give a better description of the data and analysis you are doing. One wonders if such over-weighting is really the right thing to do. $\endgroup$ – ndoogan Apr 29 '13 at 11:16
  • $\begingroup$ It's for a school project, determining which city is most focused on the music industry, based on the number of establishments. But cities like L.A. that are much bigger with more businesses in general, are pushing out the smaller cities like Nashville, who may have more record stores, labels, etc for its size than a bigger city. $\endgroup$ – amber Apr 29 '13 at 14:12
0
$\begingroup$

It's still not perfectly clear exactly what your methods are, but you can normalize on city population. Say B (for big) has 200 stores and 100,000 people while S (for small) has 14 stores and 10,000 people. A simple normalization is to divide store count by population size. focus(B) = 200 / 100,000 and focus(S) = 14 / 10,000. Then you can say the focus score for some city is the number of stores per person. Or you can reverse the fraction to get the number of persons per store. But this effectively reverses your measure. So be careful which you choose.

$\endgroup$
  • $\begingroup$ Perfect! Absolutely perfect! Thank you so much. It was just a matter of going back to the raw numbers and reassigning the ranks, not working with the ranks that I had. $\endgroup$ – amber Apr 29 '13 at 18:07
1
$\begingroup$

One would usually weight according to population size to give smaller cities less importance. In your case you don't want to give them less importance.

So don't weight your values.

If you don't put any weights into the aggregates you're computing, that is effectively treating them equally.

Did I understand your question correctly?

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.