You call cell phones? Yes, and if you are conducting telephone surveys, here are some things to consider

By: Amanda Barna

When cell phones were first gaining popularity, several research studies comparing the population with landlines to the population using only cell phones were conducted and found that respondents answered questions similarly and therefore concluded that not including cell phones in a sampling frame was acceptable because, at the time, there was no evidence that there were differences in the populations. Times have changed: 25% of US households no longer have landlines. When you add in households that are use cell phones primarily and keep landlines for purposes other than engaging in conversations, the percentage jumps to 35%. This means that only 65% of households can be reached using traditional telephone survey methods- developing a sample frame based on residential landlines. For certain populations, such as younger adults as well as Hispanics or African Americans, the percentage reached traditionally is even lower. In order to obtain a representative sample of the general public, it is now important to include a cell phone sample- if at all possible. While this may increase the cost of data collection, the data will be better quality and will yield more accurate results.

Including a cell-phone sample is not always easy and is not always appropriate. Here are several things that you need to consider:

  1. What is the geographic area that you are surveying? Where land-line samples can be drawn down to the neighborhood block level, the smallest area you can sample for cell phone samples is county. So, if you are conducting a telephone survey of any given city, for example, including a cell phone sample may be cost prohibitive due to having to screen for the right geographic area. Even if you are conducting a county-wide or state survey, it is still important to verify the location of the respondent- because numbers can be ported, people can move out of the area and take their phone number with them. So, you may think you are calling someone from Summit County, and they are actually in California.
  2. Who are you surveying? Do you want to talk with the general public or do you need a more specific population such as females with children under 18 or registered voters who have voted in two of the last three elections? Cell phone samples can only be ordered based on geography (at the county level or higher). They cannot be ordered based on demographic characteristics such as gender, age, race, etc.
  3. How many phones do they have and what are they doing when you call? You need to add a short series of questions to the beginning of a survey that includes a cell phone sample to ensure that (1) you are reaching a respondent that only has a cell phone and not a landline (to ensure that a person is not included twice in the final sample) and (2) that the respondent is not driving and is safe to complete the survey.
  4. Are you using auto dialers? The law prohibits the use of predictive dialing/auto dialers when calling cell phones so you have to make sure you are ready to get those digits working.
  5. Will you compensate respondents for their cell time? The jury is out on this. While some do compensate, others don’t and the research does not show an impact on cooperation or data quality either way.

Developing a strong sampling frame that takes these things into consideration is critical to collecting data from which you can truly draw conclusions.

Additional resources on cell phone surveys:
Assessing cell phone challenge in public opinion surveys
Collecting survey data - Cell phone surveys