request for help for newbie

#1
Hi,

I'm a psychology undergraduate student but have some challenges at work which have got me thinking about the best way to tackle them.

my knowledge of statistics is pretty basic - comparison of means using t-test, 95% confidence intervals, testing of null hypotheses and chi-square.

anyway, my problem is this:

for a given English county, a certain number of complaints comes into my organisation each week/month. I have the data for a period of time, lets say a year.

I need to answer the question "is there a statistical difference between the number of complaints submitted from different counties?"

Different counties have different populations, so it makes sense to me to modify the number of complaints to a value "number of complaints per capita"

When this has been done, I can see the values across 2007 for each month, for each county.

disregarding confounding factors such as the awareness of the complaints tel. line in different counties (I'd expect this to be higher in predominantly urban counties but could be wrong), how can I answer the question?

I have SPSS at home to use.

Also, does it matter that the counties are in themselves arbritrary divisions? would I be better off trying to compare, say, counties which are predominantly urban aginst ones which are predominantly rural?

any help greatly appreciated.

Ed :confused:
 
#2
Not sure if this the perfect solution but if your organization serves the entire population of a county you can use a Chi Square. Set up like this

County A County B County C

Complainers

Non - Complainers

and test if there are differences.
 

Keo

New Member
#4
Beware of using the "per capita" complaints, Chi square assumes the numbers in the contingency table are actual observations. In the table suggested by airmont, the "non complainers" have to be the total number of persons who COULD complain, but didn't.
If you only take complaints from customers, this number should be much lower than the whole population.