What is a representative sample?

When conducting market research, reaching the full spectrum of your target audience is vital to ensure the results will benefit your company. Poorly conducted market research could do more harm to your company and its products or services than no research at all. 

But asking for the opinions of every person in your target audience is neither practical nor affordable. So how do you gain the most accurate insights into your target audience in a cost-effective and efficient manner? Easy, turn to a representative sample.

Market analysis template

Save time, highlight crucial insights, and drive strategic decision-making

Use template

A representative sample is a smaller subset of your target audience, but it’s one that represents the characteristics of the entire population. These characteristics are typically designated by demographics, such as age, ethnicity, sex, geography, socioeconomic status, etc. 

For example, if you were determining the representative sample for the potential audience of a new reality TV show, your target audience would be extremely different than if you were marketing a new tennis racket. The reality show would have a much larger potential audience, so you would need a larger sample, whereas the group that plays tennis would be smaller and would tend to skew differently in terms of socioeconomic makeup. 

The purpose of selecting a representative sample is that the results should accurately reflect the answers you would get if you were able to survey the entire target audience. 

What is a sample and a representative sample?

A sample is a random selection from a population, whereas a representative sample selects from the population based on certain criteria. 

For example, if you want to survey a city of 100,000 people and you choose 500 random telephone numbers from an online phone directory, that would be a sample. 

A representative sample would look at age, race, socioeconomic status, geography, etc., and then choose participants to match the demographics of the entire city. 

Market research helps companies better understand their potential customers, what their pain points are, how they can be engaged, and where they search for solutions to their problems. 

This understanding allows the company to develop a product or service that meets their customers' needs. It teaches them how to set a price and how to develop the product or service within those constraints. 

Working from a representative sample that accurately gauges the full target audience's needs provides the following advantages:

Cost-effective

Surveying a few hundred people is much simpler and more efficient than trying to survey thousands or millions within a target audience.

Timely

You can conduct a smaller, representative sample-based survey quickly to get results back to the team driving development. 

Accurate

Survey results from this type of project can mean an innovation team, for example, is able to work from strong data to steer product development in the right direction and not go off on tangents that won't appeal to a target audience.

Bias reduced 

Researchers might accidentally introduce bias if they merely survey the easiest people to reach, but ensuring they reach the representative sample prevents bias from creeping in. 

A population would represent the entirety of a body that a researcher wants to learn about. However, surveying an entire population can be cost-prohibitive and inefficient. 

By crafting a representative sample of the population, the researcher can survey a small subset of the population with a degree of confidence that the answers obtained would be accurate for the entire population. 

Researchers use two primary methods in developing representative samples:

Probability sampling

Probability sampling means researchers make random selections from the entire target audience. This means each person in the target audience has the same probability of being selected.

Probability sampling can lead to the over-sampling of one demographic in a target audience while another is under-represented or not represented at all.

For example, if you were surveying for a target audience of 10,000 equally divided between men and women and chose to pick 100 to survey, you could end up with slightly more men than women, or vice versa. If, in that same audience, one age group was heavily over-represented because of geography, probability sampling might not balance that age group differential. 

Non-probability sampling

In non-probability sampling, the researcher tries to avoid the problem of over-representing or under-representing by specifically choosing the people to survey based on the total audience demographic. 

In the example above, the researcher would factor the varying age groups for a particular geographic region to ensure the survey samples on age based on the population in each age group. 

As you may suspect, this method requires more administrative effort by the researcher since they will choose their survey targets, and if someone declines the survey, the researcher must continue to look for a person in the same demographic. 

Non-probability sampling also increases the risk of bias being inserted into the research because the researcher selects each survey participant. 

In reality, researchers aren’t going to build a representative sample based on one criterion. They will use multiple criteria that fit within the target audience. Researchers will often build a grid based on those multiple criteria to determine where members of the target audience fall on the grid. 

Again, turning to the sample above, let's assume this target audience all live in the same city that can be divided into five quadrants that reflect socioeconomic status. Now if we want to base the survey on gender, household income, and geography, we would build a grid based on four different income tiers, two genders, and five locations, giving us 40 grids where the target population would be divided. 

In this case, the researcher would want to increase the sample size to 120 to be able to select, either by choice or randomly, three people from each grid. The researcher also could double the previous number to 200, still a manageable survey size, to select five from each area, increasing the survey's accuracy. 

What is the minimum sample size for a representative sample?

The minimum size for a representative sample will depend on how accurate the researcher wants the final product to be. For example, if the researcher would be content with a 90% confidence level with a 5% margin of error, the sample size would be smaller than if the researcher wanted a 95% confidence level. A sample size calculator can help with determining the minimum sample size. 

Two steps are key to ensuring you have a representative sample:

1. Understand your target audience

You need to clearly define your target audience and draft a set of characteristics that can be accurately sampled from the general population.

2. Choose an adequate data set

You want to have a large enough data set to get an accurate sample of your entire target audience, but you don’t want to push the set size so large as to be unmanageable. 

Once you’ve accomplished these steps, you could further enhance your outcome by taking a weighted approach to choosing your representative sample. For example, if you have five characteristics for your target audience but know age is the most important factor, you could weight the age choices to ensure you’re getting a truly representative sample. 

What is a representative sample and a non-representative sample?

A representative sample reflects the general population or target audience. A non-representative sample would introduce errors into the research because certain people or groups would be under-represented and others over-represented.

The best way to avoid sampling bias is to adopt a probability sampling method where participants are chosen randomly so that each person in your target audience has an equal chance of being selected for the representative sample. 

You might still have some gaps in your research if random sampling accidentally omits or under-represents certain sectors of your target audience. 

All statistical tools come with some downsides, including representative sampling. A representative sample will never give you a 100% picture of your target audience. Most researchers aim for around 90 or 95% accuracy when constructing the formula to determine their representative sample. 

Errors also can creep into the research. The main sources of error are:

Systemic error

These errors are caused by a flawed approach from the researcher, either not clearly defining the target audience or not fully understanding the characteristics that factor into their decisions.

Random error

When generating random samples, errors accidentally arise when certain characteristics are over- or under-represented. 

Bias

Conscious or subconscious bias arises when a researcher selects a representative sample.