0
$\begingroup$

I have a table of counts of consumers, and I would like to run a 2-way ANOVA on them. There are two factors, year and age-group. Please see the following table of data.

Although I usually have access to each consumer's data in separate rows (each row as an individual case), and then run a 2-way ANOVA in SPSS, this time I don't have access to that individual-level consumer data. So, since each row entered in SPSS isn't a different case, but rather a summary count, after running the test, the results don't look right.

Is it easier to do this in Excel?

How can I handle this dilemma?

Another issue has to do with the best test to run given my data. I realize I can run chi-squares (or maybe a cochrane-armitage test) on it, but an ANOVA will allow me to look at main effects and interactions among the factors which will better show me where the action is in the data, and therefore what conclusions to draw and charts/graphs to show. Also, chi-squares are so sensitive to large sample/population sizes that even the smallest difference in proportions of age group counts of one year vs another year will be highly significant.

Thank you in advance for help with this.

enter image description here

Improve this question
$\endgroup$
4
  • 1
    $\begingroup$ You could use a (Poisson) generalized linear model, which is an extension of the Chi-square test for count data allowing for interactions and main effects. $\endgroup$
    – user2974951
    Commented Sep 9, 2019 at 6:29
  • 1
    $\begingroup$ I suspect with kinds of counts you have (large and varied), it may not be bad to use a general linear model (anova). Since both of your independent variables are ordered, you may want to consider this, perhaps treating them as continuous predictors or using a linear contrast along with treating them as categorical in the model. $\endgroup$
    – Sal Mangiafico
    Commented Sep 9, 2019 at 20:11
  • $\begingroup$ Thank you for your suggestions user2974951 and Sal--I will look into them. Sal, when you say treat the independent variables as continuous predictors, can you clarify? When I hear predictors, I think of a correlation rather than an ANOVA. $\endgroup$
    – LeeZee
    Commented Sep 11, 2019 at 18:37
  • $\begingroup$ How can I deal with not having standard deviations or means if this is summary count data? It's being treated like each of the age groups is a separate level of the factor age group, and each level has an N of 1. Does anyone have clarity on the best way to run the 2-way ANOVA test in SPSS given my situation? $\endgroup$
    – LeeZee
    Commented Sep 13, 2019 at 14:38

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.