Correlation and categorical variables I have a series of categorical variables I have, up until now, been undertaking correlation analysis with. The bulb has now lit up and I'm realising that I've used completely the wrong 
analysis if I want to explore the associations between the collected variables. I'm assuming I'd need to scrap everything I've written up to this point and start again? 
If I were to scrap everything what would be the best course of action for exploring the association between variables such as pest control actions, types of pests experienced, pest control preferences and views of both private and authority run pest control services alongside the standard demographic variables of ethnicity, gender, region, income and housing tenure?
I have searched the site, but being a complete newbie to stats and having a mentor who didn't flag the issue I'm finding myself getting confused.
 A: Chi-square is one measure of association between categorical variables, but not the only one and perhaps not the best one for your purposes; it measures how far the actual results are from the expected results if the variables were unrelated. It really depends on what exactly you are trying to do. For a lot of details, see Alan Agresti's books; perhaps best for a newbie is Introduction to Categorical Data Analysis.
For now, you might want to investigate the $\phi$ coefficient which is $\frac{\chi^2}{n}$ for pairs of binary variables; $\gamma$ or $\tau$ (there are a couple variations of this one) or the polychoric correlation  for pairs of ordinal variables, $\lambda$ for general cases and other measures as well.
A: Chi-square is used to test the association, under the hypothesis
Attributes are independent vs Attributes are associated.
In SPSS you input your data. Than go to analyze -> Descriptive Statistics -> CrossTabs
a window will be open like that  
enter your variable into rows and column, than go to statistics and check chi-square and press ok and again ok. you get the desire results.
