← BCM 323 Business Research Methods
Sampling Techniques and Sample Size Determination
Notes
Foundations of Business Sampling
In empirical commerce studies, measuring an entire population (a census) is often prohibitively expensive and time-consuming. Researchers select a representative sample from the target population to make valid statistical inferences. The target population comprises the complete set of individuals or business entities possessing the characteristics under investigation. Crucial to this step is establishing an accurate sampling frame, which is the operational list of population elements from which the sample is drawn, such as a company employee register or a municipal business licensing directory.
Probability versus Non-Probability Sampling
Sampling techniques are fundamentally divided into two methodological branches:
Probability Sampling: Every element in the population has a known, non-zero probability of selection, permitting statistical generalizability. Common techniques include Simple Random Sampling, Systematic Sampling (selecting every kth element), Stratified Sampling (dividing the population into homogeneous subgroups like departments or turnover tiers, then sampling proportionally), and Cluster Sampling (sampling naturally occurring heterogeneous groups like geographic zones).
Non-Probability Sampling: Selection depends on subjective researcher judgment rather than chance. Examples include Purposive Sampling (hand-picking subject matter experts), Convenience Sampling, and Snowball Sampling (referral-based selection for hidden populations).
Sample Size Calculation
Sample size determination requires balancing statistical precision, confidence level, and practical constraints. Yamane provides a recognized formula for finite populations: n = N divided by (1 + N multiplied by e squared), where n is the sample size, N is the population size, and e is the allowable margin of error (typically 0.05 at a 95 percent confidence level).
Practical example
A researcher investigates the financial resilience of smallholder tea farmers affiliated with the Kenya Tea Development Agency (KTDA) in Murang'a County. The total population consists of 12,000 registered tea growers spread across four ecological tea factory catchment zones: Zone A (4,000 farmers), Zone B (3,000 farmers), Zone C (3,000 farmers), and Zone D (2,000 farmers). Using Yamane formula with an error margin of e = 0.05: n = 12000 / (1 + 12000 * 0.0025) = 12000 / 31 = 387.09, yielding a required sample of 388 farmers. To ensure fair representation across geographical zones, the researcher applies proportionate stratified sampling. Zone A receives 129 questionnaires (4000/12000 * 388), Zone B receives 97, Zone C receives 97, and Zone D receives 65. Within each zone, systematic random sampling selects every 30th farmer from the factory payment register. This approach mitigates bias and ensures generalizability across the county.
Quiz
1. Why is non-probability sampling generally inappropriate for formal hypothesis testing in quantitative business research?
2. If a commercial researcher uses Yamane's formula on a target population of 2,500 SME owners with a 5 percent margin of error, what is the approximate sample size?
3. Which sampling technique guarantees that distinct subgroups within an organization, such as managerial levels, are proportionally represented?
Sign in to submit and securely grade your quiz.