An observant member of today’s society might notice that just about every time they finish their shopping at the mall or grocery store, they receive not only a receipt, but also the opportunity to take a survey. In fact, it seems that everywhere you turn you are asked to take a survey of some sort.

With all these surveys and data collected, you would think companies have all the right information they need to satisfy every type of customer, right? Don’t be so sure.

Anyone can get on the Internet and send out a survey to a list of emails, and anyone could stand at the entrance of a supermarket and ask people questions. Just because you are able to collect the opinions of people does not mean that you have the ability to project inferences to a larger population.

Data collection goes beyond just collecting the opinions of people. In order for research to be effective, the sample must be representative of the target population.

One of the most famous mistakes in data collection comes from the 1948 presidential election. Pre-election polls of the day had Thomas Dewey defeating Harry Truman in the election. But when the election was over, it was Truman who carried the popular and electorate vote.

On his way back to Washington, Truman held up the Chicago Tribune announcing in large, bold print, “Dewey Defeats Truman,” adding insult to injury for those early pollsters. Although the polls had a fairly substantial sample size, an investigation into the polls discovered that only people with telephones were surveyed. The polling was biased, due to the fact that those with telephones were systematically different than the population without a telephone, which favored Truman.

In other words, the sample was not representative of the target population.

Advancements in technology have changed the means of collecting survey data throughout the last several decades, and will continue to change. Throughout the 1970s, 1980s and 1990s, random-digit dialing, which uses the phone number a the sampling unit, dominated as the golden rule of data collection due to landline penetration.

In fact, landline penetration into homes reached 97% in 1998, but has been declining every year since. The decline in landline penetration is due to technological advancements and rise in popularity in cellular phones.

Recent studies such as the National Health Interview Study have begun to provide evidence into the demographic differences of wireless-only households and their landline counterparts. According to the Pew Research Center, one in four households today are wireless only. This means that landline penetration is roughly the same as the early 1960s, and random-digit dialing was not even considered viable then.

So why is this important? As research organizations fail to account for wireless telephone numbers in the sampling frame, under-representation of the wireless-only households can potentially lead to biases in the survey estimates. This was the same mistake pollsters made in 1948 by not including people without a telephone in their sample.

Consider this: If a researcher in Oklahoma were to pull a random sample of addresses throughout the state and then run a phone match, about 65% would match with a phone number. This allows the researcher the ability to contact approximately 65% of the sample by phone and mail.

The remaining 35% of addresses can be contacted by mail only. This allows the researcher to contact approximately every person sampled, whereas, the traditional random-digit dial will have coverage of only approximately 75% of the general population. Utilizing such tools allows the researcher, as well as the client, to have full confidence generalizing the sample statistics to the larger population.

Research in the 21st century must continue to address the coverage concerns of traditional modes of data collection by leveraging the benefits of different collection methods, while simultaneously minimizing their errors.

In a world where technology has challenged the conventional wisdom that has guided researchers for many years, researchers must move to mixed-mode surveying to make sure the sample is representative of the population. As you develop your research strategies for the rest of the year, make sure your researcher can provide you with the kind of data it takes to see that the correct headline gets printed.