What Kind of Sample is Best for a Voter Opinion Survey?

Random Digit Dialing vs. List Based Voter Sample

It’s not an uncommon question, exactly how do pollsters and obtain telephone numbers or sample for their voter opinion surveys?  We are glad you asked.

For a telephone survey, there are two primary methods that are widely used. One method is called random digit dialing, or “RDD”.  Random Digit Dialing, or “random sampling” starts with an active telephone area code and prefix combination often associated with a county. After finding the correct active area code and county code, for example “303-555”, the next 4 numbers of the telephone number are created and dialed at random. An example of an RDD “block” would be the 1,000 numbers in a sequence from 303-555-0001 to 303-555-9999. The random dialing of the last 4 digits bring varied results. Since the phone numbers are not associated with a list, sometimes unlisted working residential phone numbers are found.

However, the amount of unlisted numbers found do not outweigh the amount of phone numbers that have to be dialed to find a listed working residential number; “Only about one quarter of the possible numbers formed by adding random endings to active area code/prefix combinations yield working residential numbers” (Roger Tourangeau 2004). The significance of unlisted phone numbers is small in comparison to the end result. Instead of focusing all of the calling time on polling, the RDD method takes up time by finding working phone numbers before the polling process can begin.

Another drawback to the RDD approach is the “block” of phone numbers does not follow political boundaries, so it will call voters outside of a legislative, Congressional of special voting district. RDD is fine to use for a survey of a population at the state, county, or multi-county level. However, for districts that do not following county boundaries, RDD can be messy.  Using RDD sample will call voters outside of the district you are attempting to interview, and attempts to screen out voters by asking what district they vote in is not a reliable remedy to the problem. Primarily because very few voter know the number of their state house, state senate or Congressional district.

List Based Sampling Using a Database of Registered Voters

A list-based sampling approach uses a database of registered voters that has landline and cellphone numbers appended to it.  This is the approach we use  for all of our surveys here at Magellan Strategies. Rather than generating phone numbers randomly via RDD, a list based sampling approach selects telephone numbers randomly from the database of registered voters. This approach eliminates the problems of RDD such as interviewing voters outside of a district, and wasting time by calling telephone numbers that are not working resident numbers.  We believe list-based sampling is superior to RDD for these reasons, and for the simple fact that we are calling households that we know are tied to a registered voter. Our team believes so strongly in list based sampling for our surveys we build our own state databases of registered voters in-house. To ensure that our list-based sample of registered voters is accurate and current, we update our files on a quarterly basis and run then through an address standardization and householding process. Finally, we identify movers using the US Postal Service movers database.

The bottom line is list based sampling is superior to RDD sampling for voter opinion surveys.

By | 2017-05-24T03:25:53+00:00 April 28th, 2016|