Defining the Population Size

The first step in quantitative research is to identify the study population—the group of people, objects, or entities from which observations will be made and conclusions drawn. Although it may seem simple, the concept of population is sometimes not well-defined and is often confused with the "universe," which are two very different terms. The **universe** refers to the total set of all possible subjects, while the **population** corresponds to a specific subset of that universe that you intend to analyze to obtain valid conclusions. Let's look at some simple examples: If we want to conduct a job satisfaction survey to measure how happy employees are at a corporation, the population is clear: the total number of employees in the company. However, in market research—for example, measuring interest in a new product—the population is not as obvious, as it should only cover the set of potential buyers.

Ultimately, the population consists of the group of individuals, entities, or objects that possess the unique and specific characteristics you intend to analyze. Defining this group precisely is vital for the validity of your results.

Sample Selection Strategy

Once the target population is identified, you must determine how many people to study; that is, how many people should be interviewed or given a questionnaire. Ideally, a census—interviewing every member of the population—provides the most accurate and comprehensive data. However, in practice, this is rarely possible due to limitations such as high costs, logistical complexity, and time constraints.

Therefore, instead of a census, a representative sample of that population should be used. When we scientifically extract a small sample, we can project the behaviors of the entire population within a pre-established margin of error. It is similar to a blood test: it is not necessary to extract all the blood in the body; a few drops are enough to get reliable information about your general health. While larger samples increase the precision of observations, there is a point of diminishing returns where adding more participants becomes a waste of effort, time, and resources. Therefore, your goal as a researcher is to balance three factors:

• Available time
• Budget
• Required precision

To maintain data integrity and avoid bias, samples must be selected randomly. While "purposive sampling" can be used in specific, well-justified cases, you should always incorporate a degree of randomness to reduce the natural bias produced by order.

Calculating Sample Size

Although there are many ways to determine the ideal sample size, we provide a standard framework for our users. You can use the Sample Calculator shown in the left menu to automate these calculations.

Base Formula (Infinite Populations)

Use this formula when the total population is unknown or extremely large:

Variables:

• Z (Z-value): Corresponds to the desired confidence level (e.g., 1.96 for 95% confidence).
• p (Percentage): The expected distribution of the attribute (usually 0.5 is used to ensure the most conservative sample size).
• c (Confidence Interval): Your margin of error, expressed as a decimal (e.g., 0.04 for ±4%).

Correction for Finite Populations (*)

If you are working with a small and known population, apply this adjustment to avoid an oversized sample and unnecessary costs:

Where pop is the total size of the population.

Note: In many practical applications, both formulas yield similar results; many researchers prefer to simplify by using Formula I.

Balancing the Sample through Quotas

A quota is a target number for a specific subgroup within your sample. Implementing quotas ensures that your sample proportionally reflects the real diversity of the population within each subgroup.

For example, since opinions often vary by gender and age, you can set quotas to ensure that the proportion of men and women in different age ranges in your sample reflects the behavior of the real population. Similarly, if studying brand usage, you can set quotas so that buyers of Brands A, B, and C are represented according to their respective known market shares.