With dozens of new polls out every week, it's tempting to use polling averages -- such as those used by Real Clear Politics -- to smooth out the many results and hopefully make them more understandable.

But as I noted several months ago on Political Wire, there are many reasons poll averaging doesn't work:

• Surveys are in the field at different times -- often over several days -- so respondents are not always aware of the same recent events.
  
• Each poll typically uses different methodologies for forecasting turnout. 
• Polls have different sample sizes, yet when they're averaged they are usually are weighted the same.
  
• Pollsters use different screens to determine likely voters.
  
• Some pollsters use telephone surveys while others use robocall surveys.
  
• The wording of questions on surveys can differ which can wildly skew results.
• Some of the pollsters included in the polling averages are just not reliable.
  

While doing a simple average can eliminate some of the biases noted above, the technique may also magnify them dramatically. As a result, the polling average can be more unreliable than the original polls themselves.