What population am I studying here?

[treefingers]

I'm conducting a monthly study for flea/tick medication and am using US canine households as my population. Sample data are obtained through surveys.

The basic methodology goes like this:

-I take the number of canine households that purchased in the last month and divide that by the total number of canine households that participated in the survey (let's say 1,000/10,000) to uncover an incidence rate (10%).

-I take that 10% and apply it to the entire canine household population (40mil) to get 4 mil households estimated to purchase that month.

-I divide 4 mil by 1,000 to obtain a projection multiplier (4000), which I will apply to doses purchased etc.

Suppose, however, that 20% of canine households never purchase meds. Which should I use for my population: canine households (40 mil) or canine households that purchase meds (32 mil)?

Apologize in advance if this is too elementary, but I appreciate any insight.

-A

[noetsi]

I dont understand why your incidence rate tied to the percent who participated in the survey tells you anything about what percent purchased something. Choice to participate in a survey, or how many you decide to sample does not tell you anything about what percent chose another action (to purchase). You need to look for actual data on what percent purchased and owned pets.

The answer to your question depends on your research question. is the population you are interested in total canine households or those that purchased meds? Statistics won't tell you that, or research design. It depends what you want to know