Someone explained a few days ago that the margin of error depended on the sample size, or something like that. They said that if you could poll everyone, there would be no margin of error, and the smaller the sample, the larger the margin of error.
> Someone explained a few days ago that the margin
> of error depended on the sample size, ...
It does when:
- you have access to all the data
- the accessible data represents all of the data
- the samples taken are reliable data, and
- the samples aren't lying to you
None of which is true for political polling.
The MOE is a mathematical function. It has to do with sample size. The larger the sample, the smaller the margin. The margin drops quickly as the sample size grows--until you reach a point. Then it drops very slowly. The cost of increasing the sample size is expensive.
So, when you are doing a survey/poll you have to ask yourself where the tradeoff is between MOE and the expense of sample size. For example, to get 300 RV in a proper distribution of the population (R/D/I) you probably have to make at least 1,500 phone calls. Not a simple thing to do in a short time frame--every day with a new sample--for a campaign. My expereince was about ten years ago--so the calling population has declined while the sample size stayed the same. THIS is what will lead to MOE problems.
Since most polls are about the same size, they are targeting the same MOE. This means they are using a manageable sample size, which can be repeated throughout the campaign.
It does not mean tha the poll IS off by that much, it just means that it COULD be off by that much.
When you are asking pretty basic questions, the MOE is manageable. When there polls are this close, the MOE is more critical.