Making Telephone Surveys Effective: Quality Management in Polling and Telephone Survey Research

Kevin Stay June 12, 2019

Live interview telephone surveys are still one of the most accurate ways to conduct a public opinion poll. This is especially true in geographic areas that are too small for an adequate sample size of online respondents. According to Pew Research, about 11% of Americans don’t use the internet at all, including 15% of rural Americans.

However, Pew also reports that telephone survey response rates have dropped by over 30% between 1997 and 2019, and currently sit at around 7%. This decline in response rates is due to a number of factors, including a recent uptick in spam robocalls, and subsequent growing use of technologies that block auto-dialers, as well as the increasing popularity of cellphones over landlines.

While these declining response rates don’t necessarily impact data accuracy, they do mean that pollsters and survey researchers have to spend more time and other resources to reach and contact individuals by phone. Costs for a single research project can thus be in the tens of thousands.

So, it is critical to hire contact centers with high quality interviewers capable of correctly executing your survey instruments.

Even with the best contact centers, calling has to be monitored to ensure the quality of a project and accuracy of the survey. Traditionally, this means that, for each day the survey is in the field, you need to spend hours listening in to a random selection of calls.

The problem here is that random sampling means you will miss the opportunity to review interviews in progress because you are listening to an agent try and fail (repeatedly) to get someone on the line. Despite conducting up to a hundred interviews a night and listening for 3-4 hours each time, chances are you will only hear one or two interviews from start to finish.

Any good pollster knows that this is not an acceptable sample size.

The odds of spotting a problem that could impact data quality, such as an agent failing to use the pronunciation guide correctly and mispronouncing a candidate’s name, or misreading question wording, are incredibly low. Accurate, effective surveys require a solution that can analyze as much call data as possible, to avoid missing errors in data that could waste your survey investment.

Speech to text technology is one such solution. With a good STT system in place, interviews can be transcribed in real-time (or near real-time). This allows for transcriptions to be quickly reviewed, either manually or with automated analytics tools, thus identifying areas of strength and areas of concern. STT also provides other benefits, such as measuring respondent emotion and sentiment, which can help the survey researcher.

Rather than spending hours sifting through calls and hoping to find a relevant recording, STT enables a pollster to focus in on agents who are successfully making calls, and employ effective quality management.

Kevin Stay

Kevin Stay is an Account Executive at Voci Technologies. He previously worked as a pollster and campaign consultant on campaigns from city-wide ballot initiatives to campaigns for governor. In his free time, Kevin hosts pub-trivia and regularly performs with an improv comedy troupe.

Access our ASR API

With up to 1000 hours of audio at no charge