Unit 5: Sampling techniques - Subjective Questions

Sampling is the process of selecting a smaller group of individuals from a larger population for the purpose of conducting research. The selected group is called the sample, while the complete group about which the researcher wants to draw conclusions is called the population.

• A population may consist of all university students, patients with depression, or adults in a country.
• A sample is a manageable subset selected from that population.
• A sampling frame is a list or operational representation of the population from which the sample is selected.

The main purpose of sampling is to obtain information about a population without studying every individual. A properly selected sample can save time, reduce costs, and make psychological research more practical. The quality of conclusions depends on how well the sample represents the population.

Sampling is important because psychological populations are often too large, geographically dispersed, or difficult to study completely. Researchers therefore select a sample and use its findings to estimate characteristics of the population.

Importance of sampling:

• It reduces the time and cost of data collection.
• It makes large-scale research manageable.
• It allows researchers to conduct detailed assessments.
• It supports the study of populations that cannot be completely examined.

Sampling directly affects generalizability, which refers to the extent to which findings can be applied to people beyond the sample. A representative sample improves generalizability, whereas a biased or very narrow sample limits it. Sampling also affects validity because systematic differences between the sample and population can produce misleading conclusions. For example, findings based only on psychology students may not apply to older adults or people from different cultural backgrounds.

Probability sampling is a sampling method in which every eligible member of the population has a known and usually non-zero probability of being selected. Selection is based on a random procedure rather than the researcher's personal judgment.

Essential features:

• A clearly defined target population is required.
• A sampling frame is usually needed.
• Random selection is used.
• The probability of selection can be estimated.
• Sampling error can be assessed statistically.

Probability sampling is useful because it generally produces samples that are more representative than convenience samples. It reduces selection bias and allows researchers to estimate how accurately sample findings reflect the population. It also supports statistical inference, including the calculation of confidence intervals and sampling error. However, probability sampling may require considerable time, money, and access to a complete population list.

Simple random sampling is a probability sampling method in which every member of the population has an equal chance of being selected.

Procedure:

1. Define the target population.
2. Prepare a complete sampling frame.
3. Assign a unique number to each member.
4. Select the required number using a lottery method, random-number table, or computer-generated random numbers.

Example: A researcher studying stress among 2,000 employees numbers all employees and randomly selects 200 participants.

Advantages:

• It is easy to understand and statistically straightforward.
• It minimizes personal selection bias.
• It permits the use of standard inferential statistics.

Disadvantages:

• A complete and accurate sampling frame is necessary.
• It may be expensive or difficult for widely dispersed populations.
• Small subgroups may be underrepresented by chance.
• Selected participants may still refuse to participate, causing non-response bias.

Systematic sampling involves selecting members from an ordered sampling frame at a regular interval. The researcher first chooses a random starting point and then selects every th member.

The sampling interval is calculated as:

where is the population size and is the desired sample size. For example, if and , then . After choosing a random starting number between 1 and 10, every tenth person is selected.

Difference from simple random sampling:

• Simple random sampling selects each participant independently through random procedures.
• Systematic sampling selects participants according to a fixed interval after a random start.

Advantages: It is simple, quick, and spreads the sample across the sampling frame. Disadvantages: It can be biased if the list contains a hidden periodic pattern related to the sampling interval.

Stratified random sampling divides the population into relatively homogeneous subgroups called strata, and then randomly selects participants from each stratum. Strata may be based on variables such as age, gender, educational level, location, or clinical status.

In proportionate stratified sampling, the sample reflects the population proportions. If a population contains 60% women and 40% men, a sample of 200 would include approximately 120 women and 80 men.

In disproportionate stratified sampling, the researcher selects different proportions from the strata. A small but important group may be oversampled so that it has enough participants for analysis. Statistical weights may later be used to represent the population accurately.

Advantages:

• Ensures representation of important subgroups.
• Can increase precision and reduce sampling error.
• Allows comparisons between strata.

Disadvantages:

• Requires accurate information about the population.
• Poorly chosen strata can reduce usefulness.
• It is more complex to organize and analyze.

Cluster sampling divides a population into naturally occurring groups called clusters, such as schools, hospitals, villages, or counseling centers. The researcher randomly selects clusters and studies all or some individuals within the selected clusters.

Types:

• In one-stage cluster sampling, all members of selected clusters are studied.
• In two-stage cluster sampling, individuals are randomly selected from the chosen clusters.
• Multistage sampling selects samples through several successive levels, such as states, districts, schools, and students.

Advantages:

• It is economical when participants are geographically scattered.
• It reduces travel and administrative costs.
• It is useful when a complete list of individuals is unavailable but a list of clusters exists.

Disadvantages:

• Individuals within the same cluster may be similar, reducing the diversity of the sample.
• Sampling error may be higher than in simple random sampling.
• Analysis may require adjustments for the cluster design.
• Findings can be biased if selected clusters are not representative.

Both methods divide the population into groups, but they use those groups differently.

Stratified sampling is usually preferred when subgroup comparisons are important. Cluster sampling is useful when individuals are widely dispersed or a complete list of individuals is unavailable. Stratification can reduce sampling error, whereas clustering may increase it because members of a cluster often share similar experiences.

Multistage sampling is a probability sampling procedure in which sampling is carried out in two or more stages. Different sampling methods may be combined at each stage.

Example of a nationwide study of adolescent mental health:

1. Randomly select states or regions.
2. Randomly select districts within those regions.
3. Randomly select schools within the districts.
4. Randomly select classes or students within the schools.

This design is useful when the population is large and geographically dispersed. It reduces travel, administrative effort, and cost. It also makes it possible to organize data collection through institutions such as schools and hospitals.

However, participants selected at later stages may be similar because they come from the same areas or institutions. Consequently, sampling error may be greater than in a simple random sample. Researchers should use appropriate sample-size calculations and statistical methods that account for clustering. Non-response at any stage can also reduce representativeness.

Non-probability sampling is a method in which not every member of the population has a known or equal chance of selection. Participants are selected through accessibility, researcher judgment, referrals, or predetermined characteristics rather than random procedures.

Characteristics:

• A complete sampling frame is usually unnecessary.
• Selection probabilities cannot usually be calculated.
• Sampling is often quicker and less expensive.
• The risk of selection bias is higher.
• Generalization to the wider population is limited.

Non-probability sampling may be appropriate when:

• The population is difficult to identify or access.
• The research is exploratory or qualitative.
• The study concerns rare or hidden populations.
• Time and resources are limited.
• The researcher needs participants with specific experiences.

Although these methods are less suitable for estimating population parameters, they can provide valuable information about attitudes, experiences, and psychological processes.

Convenience sampling involves selecting participants who are easiest for the researcher to access. Examples include recruiting students from a nearby college, patients available at a clinic, or volunteers responding to an online advertisement.

Advantages:

• It is inexpensive and quick.
• Recruitment and data collection are simple.
• It is useful for pilot studies, classroom projects, and initial exploration.
• It may help researchers test instruments before a larger study.

Limitations:

• Participants may differ systematically from the target population.
• Volunteers may be more motivated or psychologically aware than non-volunteers.
• It often produces a narrow demographic range.
• Sampling error cannot be estimated reliably.
• Generalizability is weak.

For example, a study based only on psychology students may overestimate knowledge about mental health. Researchers should clearly describe the sample and avoid making claims that extend beyond the group studied.

Purposive sampling, also called judgmental sampling, involves deliberately selecting participants who possess characteristics, experiences, or knowledge relevant to the research question. The researcher uses informed judgment rather than random selection.

Common forms include:

• Typical-case sampling: selecting participants who represent ordinary or usual cases.
• Extreme or deviant-case sampling: selecting unusual or highly distinctive cases.
• Criterion sampling: including only people who meet a specified criterion.
• Maximum-variation sampling: selecting participants with diverse backgrounds or experiences.
• Expert sampling: selecting individuals with specialized knowledge.

For example, a researcher studying recovery from trauma may purposively select adults who have completed a specified therapy program. The method is useful in qualitative research, case studies, and research involving specialized or rare populations.

Its main limitation is that the researcher’s judgment may introduce bias. Since selection probabilities are unknown, findings should be interpreted as context-specific rather than automatically representative of the broader population.

Quota sampling is a non-probability technique in which the researcher establishes quotas for particular categories and then recruits convenient participants until each quota is filled. For example, a researcher may require 50 men and 50 women but recruit them from easily accessible locations.

Similarities:

• Both methods divide the population into categories.
• Both can ensure that selected categories appear in the sample.
• Both may use variables such as age, gender, or location.

Differences:

• Stratified random sampling randomly selects participants within each stratum.
• Quota sampling selects participants non-randomly within each category.
• Stratified sampling permits stronger statistical inference.
• Quota sampling is generally quicker and less costly.

Quota sampling can be useful when researchers need a sample with particular demographic characteristics but cannot obtain a complete sampling frame. However, participants within a quota may not represent all members of that category because selection is based on convenience or interviewer judgment.

Snowball sampling is a non-probability method in which existing participants recruit or refer other people who meet the study criteria. The sample grows through successive waves of referrals.

Procedure:

1. The researcher identifies a few initial participants, called seeds.
2. These participants are asked to refer eligible individuals.
3. New participants may provide further referrals.
4. Recruitment continues until the sample size or information requirement is reached.

It is useful for hidden, stigmatized, or hard-to-reach populations, such as people with uncommon disorders, undocumented migrants, or survivors of sensitive experiences.

Advantages:

• Provides access to populations without a formal sampling frame.
• Uses trust and social networks to improve recruitment.
• Can be relatively inexpensive.

Disadvantages:

• Participants tend to recruit people similar to themselves.
• Social-network bias may reduce diversity.
• The probability of selection is unknown.
• Confidentiality and undue influence must be carefully managed.
• Findings have limited generalizability.

Voluntary response sampling occurs when individuals choose themselves to participate after seeing an invitation, advertisement, online post, or public appeal. The researcher does not randomly select participants from the population.

Volunteers may differ from non-volunteers in several ways. They may have stronger opinions, greater interest in psychology, more free time, higher motivation, or personal experience with the topic. For example, people with especially positive or negative experiences of counseling may be more likely to complete a survey about therapy.

Advantages:

• Recruitment is fast and relatively inexpensive.
• It is useful for online surveys and exploratory research.
• It may reach participants who are difficult to recruit through institutions.

Disadvantages:

• Self-selection bias can distort findings.
• The sample may not reflect the population.
• Individuals with limited internet access or low motivation may be excluded.
• Generalization should be cautious, even when the sample is large.

Probability sampling is preferable when the goal is to generalize findings to a clearly defined population. Non-probability sampling is useful when the population is inaccessible, the study is exploratory, or resources are limited. Neither method automatically guarantees a good sample; poor measurement, non-response, and inadequate sample size can affect both.

Sampling error is the difference between a sample statistic and the true population parameter that occurs because only a subset of the population is studied. It can occur even when the sample is selected randomly.

Random sampling error:

• Results from chance variation between samples.
• Can be reduced by increasing sample size and using an appropriate probability design.
• Does not necessarily favor one direction.

Systematic sampling bias:

• Occurs when the method consistently overrepresents or underrepresents certain members.
• Cannot be corrected simply by increasing the sample size.
• May result from convenience sampling, an incomplete sampling frame, or non-response.

For example, a random sample may contain slightly more highly anxious individuals than the population by chance. This is sampling error. If a depression survey recruits only people attending a mental-health clinic, it may systematically overestimate depression in the general population. This is sampling bias.

The choice of sampling technique should be guided by the research purpose and the characteristics of the target population.

Important factors include:

• Research objective: Generalization requires probability sampling, whereas exploratory or qualitative work may use purposive sampling.
• Nature of the population: Rare, hidden, or specialized populations may require snowball or purposive methods.
• Availability of a sampling frame: A complete list supports simple random, systematic, or stratified sampling.
• Geographical distribution: Widely dispersed populations may be studied using cluster or multistage sampling.
• Need to compare subgroups: Stratified sampling can ensure adequate representation.
• Time and budget: Convenience or quota sampling may be practical when resources are limited.
• Required precision: Probability methods are preferable when accurate population estimates are needed.
• Ethical and access issues: Recruitment must be fair, voluntary, confidential, and feasible.
• Expected non-response: Researchers may need additional recruitment to compensate for refusals or dropouts.

Advantages of probability sampling:

• Every eligible member has a known or estimable chance of selection.
• Selection bias is reduced compared with non-probability methods.
• Samples are more likely to represent the target population.
• Sampling error can be estimated.
• Findings can be generalized more defensibly.
• Statistical tests and confidence intervals can be applied appropriately.

Disadvantages of probability sampling:

• A complete and accurate sampling frame may be difficult to obtain.
• It can be expensive and time-consuming.
• Fieldwork may be complicated when participants are geographically dispersed.
• Selected individuals may refuse to participate.
• Non-response can create bias despite random selection.
• Some designs, such as cluster sampling, may have increased sampling error.

Thus, probability sampling provides stronger scientific justification for generalization, but it requires careful planning, adequate resources, and procedures for handling non-response and incomplete population lists.

Advantages of non-probability sampling:

• It is relatively inexpensive and quick.
• It does not usually require a complete sampling frame.
• It can provide access to hidden, rare, or specialized populations.
• It is useful for exploratory, qualitative, and pilot research.
• Researchers can deliberately include people with relevant experiences.
• Recruitment may be flexible in changing research settings.

Disadvantages:

• Selection probabilities are unknown.
• Selection bias and self-selection bias are common.
• Sampling error cannot be estimated reliably.
• Samples may be homogeneous or unrepresentative.
• Findings have limited generalizability.
• Researchers may unintentionally recruit participants who support their expectations.

Non-probability methods are not inherently useless. They are appropriate when access is difficult or when depth of understanding is more important than population estimation. Researchers should describe the sampling process transparently and limit conclusions to the population and context actually studied.