# Statistical tests for aggregate data

#### s793365

##### New Member
Hi everyone,

I am looking at the snack products that are advertised online over time. I have two variables, snack category (which includes 4 levels - chips, candy, drinks, meals) and time (3 levels - 2010, 2015, 2020). I have the average number of online ads seen by my sample for each year, by food category (i.e., aggregate data). I am wondering if there is a statistical test I could use test the differences between ads viewed by food category over time for aggregate data? Is it possible to use chi square of independence or z test for proportions?

Thank you.

Last edited:

#### Karabiner

##### TS Contributor
So you have 3 independent samples (2010, 2015, 2020), and each participant indicated
how many of the e.g. 47 chips-related ads s/he has seen? If you have the standard deviation
of "proportion seen", you can test whether the means differ. By the way, how large are your
sample sizes?

With kind regards

Karabiner

#### s793365

##### New Member
So you have 3 independent samples (2010, 2015, 2020), and each participant indicated
how many of the e.g. 47 chips-related ads s/he has seen? If you have the standard deviation
of "proportion seen", you can test whether the means differ. By the way, how large are your
sample sizes?

With kind regards

Karabiner
So the data that I have was collected by a third party company and the data that the company gives me is expressed as the average percentage of online ads seen in their population sample for each year and snack category. I then divide these numbers by 100 to get the average number of ads seen in each year for each snack. Technically the samples may not be independent (I think) as the same people may have been included in each of the 3 years (but it's not paired data). I don't have access to how large the sample sizes either.

#### Karabiner

##### TS Contributor
Well, then I am afraid that I have no idea how you can perform a test here.

With kind regards

Karabiner