When conducting research on a topic of interest or, in our case, a PR survey, researchers simply cannot feasibly collect data from every individual. Instead, the sample of a PR Survey includes a smaller group of individuals that represents a larger population. Once the sample is truly representative of the population in question, researchers can take their results and generalize them to the larger group such as the entirety of the U.S. population.

What is a sample

A sample is a subset of a population that is used to represent an entire group as a whole. The participant sample will be representative of an overall population if the collective demographic characteristics within the sample align with the same percentage spread as the population it aims to represent.

In order to acquire a statistically valid sample, researchers use several sampling methods for their data collection process.

Probability sampling:
Probability sampling means that every individual in a population stands an equal chance of being selected. Because probability sampling involves random selection, it assures that different subsets of a population have an equal chance of being represented in the survey results. This makes probability samples more diverse, and researchers are better able to reduce any sampling biases that may skew data.

Probability Sampling Consists of 3 Subgroups:

Simple random sample – uses software to randomly select participants – the researcher selects random individuals throughout a population to participate in a survey study

Stratified random sample – Separates the sample into defined subgroups (such as gender, age, race, etc.) and collects data from random participants within each group – researchers typically use this method to ensure greater accuracy as each subset is represented

Cluster sampling – targets naturally occurring clusters (e.x., geographic clusters such as grouping people from specific cities together)

Non-probability sampling:
Non-probability sampling, on the other hand, involves selecting participants using methods that do not give every individual in a population an equal chance of being chosen to participate in a survey study. This method of sampling is more expensive and labor intensive – as it requires availability of respondents that match specific criteria.

Here too we can identify several types of non-probability samples:

Convenience sampling – uses respondents that are available and convenient (or cost effective)

Purposive sample – seeks out respondents that meet specific criteria (e.g., men aged 18-35 that play football)

Quota sampling – limits number of respondents per groups/attributes by use of quotas – one of the most common methods in online research

Use of Targeting:
Targeting is used in market research to reach specific groups of people. Targets can be set up using existing questions included in the PR survey.

As such, if researchers are interested in sampling only those Americans working full-time, researchers could ask respondents their current employment status, only continuing the survey for those who report being employed full-time.

Cost is a significant implication of excessive targeting. Feasibility of a research project plummets and costs increase as the sample becomes more niche.

Due to the potential impact on feasibility and costs, targeting should be the first topic of discussion when planning your survey – this helps researchers carry out a plan for sampling accordingly.

Calculating the Margin of Error:

While it is impossible to know the precise variance in generalizability of data between the overall population and the survey sample, researchers are able to statistically estimate the size of the sampling error. In political polls, for example, you might often hear the margin of error expressed within a particular confidence interval: “The margin of error fell within +/- two percentage points, with a confidence interval of 95 percent.”

What exactly does that mean? Let’s say a survey study found that 25% of Americans prefer the color red over green, and the margin of error is +/- two percentage points with a confidence interval of 95 percent. This tells the reader that if you were to field the same survey 100 times over, 95 of the reiterations would yield data points that indicate anywhere from 23% to 27% of Americans prefer the color red over green. Thus, researchers are 95% certain that the percentage of Americans who prefer red over green falls between 23% and 27%.

In general, the margin of error decreases if sample size increases. As the sample becomes closer to reaching the size of the total population, the chances of accurately capturing all characteristics of the population become more likely.

When formulizing PR surveys, it is important to carefully consider samples and targeting to genuinely represent your target consumer or business market. We can help you evaluate your next PR survey research project!