Unit 5: Sampling techniques

I. Orientation

Sampling is the systematic selection of a smaller group from a defined population so that psychological researchers can make justified statements about the larger group. The central principle is representativeness: the selected participants should reflect important characteristics of the population, while the sampling procedure should reduce avoidable bias.

• Population: The complete group to which the researcher wants to generalize, such as all first-year university students in a country.
• Sample: The smaller group actually studied, such as 300 first-year students selected from several universities.
• Sampling frame: The operational list or source from which participants are selected, such as an enrolment register.
• Parameter and statistic: A parameter describes the population; a statistic describes the sample. For example, population mean ( \mu ) is estimated by sample mean ( \bar{x} ).
• Representativeness: Similarity between sample and population on relevant characteristics, such as age, gender, socioeconomic status, or clinical diagnosis.
• Sampling error: Difference between a sample statistic and the true population parameter caused by studying a sample rather than the whole population.
• Sampling bias: Systematic distortion caused by undercoverage, nonresponse, self-selection, or an unsuitable sampling frame.
• Ethical foundation: Participation must be voluntary, informed, confidential, and consistent with institutional ethical standards.

II. Sampling — Selecting a manageable part of a population

A. Sampling

Sampling is the planned process of choosing participants from a population for measurement and analysis. It is used because psychological populations are usually too large, dispersed, or costly to study completely.

• Purpose: Sampling makes research feasible. A study of depression among 100,000 adults may realistically collect data from 1,000 carefully selected adults.
• Target population: This is the theoretical group to which findings apply, for example, adolescents aged 13–18 with diagnosed anxiety.
• Accessible population: This is the portion practically available to the researcher, such as adolescents attending clinics in one city.
• Sampling unit: The basic element selected, such as an individual, household, classroom, school, or hospital.
• Sample size: Larger samples generally produce more precise estimates, but size alone cannot remove systematic bias.
• Sampling fraction: The proportion selected from the population.

Sampling fraction = n / N

Here, ( n ) is sample size and ( N ) is population size. Selecting 500 people from 10,000 gives a sampling fraction of ( 500/10{,}000 = 0.05 ), or 5%.

• Generalization: Researchers may generalize from sample to population only when the sampling method, sample quality, and study conditions support that inference.
• Planning sequence: Define the population, identify the sampling frame, choose a technique, determine sample size, recruit participants, and document refusals or exclusions.

III. Probability sampling and its types — Selection with known chances

A. Probability sampling and its types

Probability sampling gives every population element a known, usually non-zero, chance of selection. It is the preferred approach when the aim is statistical generalization and estimation of sampling error.

• Known selection probability: If each person has a stated chance of selection, randomization can be used to reduce investigator choice and selection bias.
• Random mechanism: Random-number generators, lottery procedures, or computerized selection—not convenience or researcher preference—determine inclusion.
• Statistical inference: Probability samples permit confidence intervals and design-based estimates because selection probabilities are known.
• Main condition: A reasonably complete sampling frame is needed. A missing group cannot be selected, even by a perfectly random procedure.
• Core limitation: Random selection does not guarantee a perfectly representative sample in every study; chance imbalance can still occur.

B. Probability sampling and its types: Simple random sampling

Simple random sampling selects units so that every unit, and every possible sample of a given size, has an equal chance of selection.

• Procedure: Number all 2,000 eligible students from 1 to 2,000, then use a random-number generator to select 200.
• Strength: It is conceptually straightforward and minimizes conscious selection by the researcher.
• Requirement: The sampling frame must list the entire population accurately and without duplicates.
• Limitation: It can be expensive when population members are widely scattered, and a complete list may be unavailable.
• Psychological application: A university may randomly select participants from its full student database for a study of academic stress.

C. Probability sampling and its types: Systematic sampling

Systematic sampling selects every ( k )-th unit after choosing a random starting point.

k = N / n

Here, ( N ) is population size, ( n ) is desired sample size, and ( k ) is the sampling interval.

• Procedure: From 10,000 records, select 500 participants; ( k = 10{,}000/500 = 20 ). Choose a random start from 1 to 20, then select every twentieth record.
• Strength: It is simpler to administer than repeated random-number selection and spreads the sample across the frame.
• Risk: Periodicity can create bias. If every twentieth record follows a repeating pattern, selected cases may be unusually similar.
• Practical safeguard: Inspect the ordering of the sampling frame before applying the interval.

D. Probability sampling and its types: Stratified random sampling

Stratified random sampling divides the population into relevant subgroups, or strata, and randomly samples within each stratum.

• Strata: Useful categories may include age group, sex, language, region, or diagnostic status.
• Proportionate selection: If a population is 60% women and 40% men, a sample of 500 may include 300 women and 200 men.
• Disproportionate selection: Researchers may deliberately oversample a small group, such as 100 participants with a rare disorder, and later apply weights.
• Strength: It improves subgroup representation and can increase precision when members within strata are similar.
• Limitation: Accurate information about stratum membership is needed before sampling, and analysis may require weighting.

E. Probability sampling and its types: Cluster and multistage sampling

Cluster sampling selects naturally occurring groups, such as schools or clinics, rather than directly selecting individuals. Multistage sampling combines several selection stages.

• Cluster procedure: Randomly select 20 schools, then survey eligible students within those schools.
• Multistage procedure: Select regions, then schools within regions, then classrooms, and finally students.
• Strength: It reduces travel, administrative cost, and the need for one nationwide list of individuals.
• Limitation: People in the same cluster often resemble one another, producing less independent information than the same number of individually selected participants.
• Design effect: Clustered data often require larger samples or adjusted analyses because observations within a school or clinic are correlated.

IV. Non-probability sampling and its types — Selection without known probabilities

A. Non-probability sampling and its types

Non-probability sampling selects participants without giving every population member a known chance of inclusion. It is common in exploratory, qualitative, clinical, and time-limited research.

• Selection basis: Inclusion may depend on accessibility, researcher judgment, referrals, volunteering, or predetermined characteristics.
• Main implication: Sampling error cannot be calculated in the usual probability-sampling sense, and population generalization is more limited.
• Appropriate use: It can help study hidden populations, rare experiences, pilot interventions, or participants with specialized expertise.
• Risk: Volunteers and easily reached participants may differ systematically from nonparticipants, especially in motivation or psychological distress.

B. Non-probability sampling and its types: Convenience sampling

Convenience sampling recruits people who are readily available to the researcher.

• Example: A researcher distributes a stress questionnaire to students in a nearby psychology class.
• Advantage: It is rapid, inexpensive, and practical for pilot studies or classroom demonstrations.
• Disadvantage: Students from one class may differ from working adults in age, education, income, and stress exposure.
• Bias mechanism: Availability replaces random selection, so the sample may overrepresent people who are nearby, cooperative, or connected to the institution.
• Use condition: Findings should normally be described as applying to the sampled group rather than automatically to all adults.

C. Non-probability sampling and its types: Purposive or judgmental sampling

Purposive sampling deliberately selects participants who possess characteristics relevant to the research question.

• Example: A study of trauma recovery recruits licensed clinical psychologists who treat survivors.
• Strength: It provides information-rich cases and is especially useful for qualitative interviews or expert research.
• Selection criterion: The researcher may require a diagnosis, professional role, lived experience, or minimum duration of treatment.
• Limitation: Researcher judgment determines inclusion, so another researcher might construct a different sample.
• Quality safeguard: State inclusion criteria clearly and explain why the selected participants are suitable for the phenomenon studied.

D. Non-probability sampling and its types: Quota sampling

Quota sampling fills specified subgroup quotas without randomly selecting participants within those groups.

• Example: A researcher recruits 100 participants—50 men and 50 women—from public locations, even though selection inside each category is based on availability.
• Advantage: It ensures visible representation of chosen categories more quickly than probability sampling.
• Difference from stratified sampling: Stratified sampling randomly selects within strata; quota sampling does not.
• Limitation: Participants inside a quota may share unmeasured characteristics, such as higher education or stronger interest in psychology.
• Use condition: Quotas should be based on variables demonstrably relevant to the study, not merely convenient categories.

E. Non-probability sampling and its types: Snowball or chain-referral sampling

Snowball sampling asks initial participants to refer other eligible people from their networks.

• Example: Individuals from a stigmatized minority group refer peers for interviews about discrimination.
• Strength: It can reach hidden or difficult-to-list populations where ordinary sampling frames do not exist.
• Network effect: Participants with large, connected, or similar social networks are more likely to appear in the sample.
• Limitation: Referrals may produce homogeneity and exclude isolated individuals who lack connections to initial participants.
• Ethical safeguard: Recruitment must protect privacy; participants should not be pressured to reveal another person’s identity or participation.

F. Non-probability sampling and its types: Volunteer or self-selection sampling

Volunteer sampling relies on people choosing to respond to an invitation, advertisement, online post, or research-panel notice.

• Example: People who click an online advertisement about loneliness complete a survey.
• Advantage: Recruitment is efficient and participants are often motivated to provide detailed responses.
• Disadvantage: Volunteers may have unusually strong opinions, personal experience, available time, or emotional investment.
• Interpretive caution: High response numbers do not prove representativeness because nonresponders may differ systematically.
• Safeguard: Report recruitment channels, eligibility screening, incentives, and response or completion rates.

V. Advantages and disadvantages of sampling techniques — Choosing a defensible design

A. Advantages and disadvantages of sampling techniques

The value of a sampling technique depends on the research aim, population accessibility, required generalization, resources, and ethical constraints.

• Probability techniques: They support stronger population inference because selection probabilities are known.
  • Advantages: Reduced selection bias, measurable sampling error, use of confidence intervals, and transparent replication.
  • Disadvantages: They may require complete sampling frames, extensive administration, greater cost, and contact with geographically dispersed participants.
• Non-probability techniques: They provide practical access when random sampling is impossible or when depth is more important than statistical representativeness.
  • Advantages: Lower cost, faster recruitment, access to hidden groups, and suitability for exploratory or qualitative designs.
  • Disadvantages: Unknown selection probabilities, greater risk of bias, weak statistical generalization, and difficulty estimating sampling error.
• Simple random versus systematic: Simple random sampling has strong randomization but may be administratively cumbersome; systematic sampling is efficient but vulnerable to periodic ordering.
• Stratified versus cluster: Stratification can improve subgroup precision; clustering reduces cost but usually increases sampling variance because participants within clusters resemble one another.
• Convenience versus purposive: Convenience prioritizes access; purposive prioritizes relevance and information richness. Neither automatically represents the wider population.
• Sample size and quality: A large biased sample can be less useful than a smaller, well-designed sample. Precision concerns random error; representativeness concerns systematic error.
• Reporting standard: A credible report identifies the target population, sampling frame, selection method, sample size, inclusion criteria, refusals, exclusions, and limits on generalization.
• Decision principle: Select probability sampling when population estimates are central; select a non-probability method when accessibility, specialized experience, or exploratory depth is the primary requirement.