What is Weighting?
It is frequently the case that the people who answered your poll are not fully representative of the region you were polling over. Weighting is a technique to adjust answers to account for over- and under-represented groups.
For example, suppose the population in an area is 55% female; but 70% of my survey answers come from women. When reporting my survey results, I'd want to to boost the importance of men's answers a bit, and reduce women's answers. Weighting is the process to do this.
Each respondent is given a "weight" number which represents the relative importance of that response compared to other responses. We calculate these values for you, and make them available when you download the raw result data. Perhaps most important, we show you how we calculated the weights.
Compare Weighted/Unweighted Results
You can view weighted results alongside unweighted results to see what effect weighting has on the numbers.
Weighted results include both bar charts of the aggregate weighted results, and also weighted crosstabs.
How Does it Work?
Precision Polling makes possible to weight across multiple factors (e.g. gender, age, ethnicity) in one click. We use your uploaded phone list as the baseline demographic makeup; we then compare this to the demographic makeup of your respondents.
To use this feature, you only need to ensure that your phone list contains the demographics you wish to weight over, and that the demographic makeup of this sample matches the population you are polling over. Then, you simply click the "Weighting" button, and we compute results.
For the Polling Nerds...
Our weighting technique is easy-to-use, and quite accurate given a representantive phone list. It has the additional benefit that if there are multiple dimensions, they are counted as dependent variables (e.g. we'd count a grouping by "female AND 35-49").
We realize that some poll nerds will want to twist the knobs themselves. We are working on an advanced option where you may define the expected demographic breakdown for each dimension yourself (as independant variables).