Sampling Techniques: Definitions and Concepts

• Sampling saves resources like time, money, and effort.
• Smaller, well-managed samples yield more reliable data, improving data quality.
• Sampling captures diverse subgroups within a population, addressing population heterogeneity.
• The target population is the group to which research results are generalized.
• Sampling error is the difference between sample estimates and actual population values, related to sample size and method.
• Lower sampling error corresponds to higher external validity.
• External validity is determined by representativeness and generalizability.
• Representativeness reflects how closely the sample mirrors the target population.
• Generalizability is the extent to which results apply to other people, settings, or times.
• Random selection ensures everyone in the population has an equal chance of being chosen, reducing bias and enhancing external validity.
• Confidence level is the percentage chance the population parameter lies within the confidence interval, commonly at 95%.
• Confidence interval is the range within which the population parameter is likely to fall.

Sampling Techniques

• Probability samples involve random selection.
• Simple random sampling gives each individual an equal chance of selection, is highly representative, but hard to implement.
• Stratified random sampling divides the population into subgroups and samples from each, which reduces sampling error and increases representation.
• Systematic sampling involves choosing every kth element after a random start as it's easy to implement and close to random.
• Multi-stage cluster sampling samples groups (clusters) and then individuals, eliminating the need for a full list.
• This is cost-effective for large populations.
• Non-probability samples do not involve random selection.
• Convenience sampling easily accesses participants, such as college students, but has low generalizability.
• Purposive sampling selects individuals with specific traits, which is useful for studying targeted subpopulations.
• Quota sampling ensures certain ratios across categories but is not random.
• Referral samples use existing participants to refer others (snowball sampling) and is useful for hidden populations.
• Network sampling focuses on social networks or connections.

Experimentation Assumptions

• Time order: cause leads to effect.
• Correlation exists between the independent variable (IV) and dependent variable (DV).
• Extraneous variables need to be controlled

Internal Validity and Random Assignment

• Internal validity assesses whether the research is truly measuring causality.
• Random assignment ensures each participant has an equal chance of being in any group, reducing bias.

Threats to Internal Validity

• Confounds are variables that mix with the IV and affect the DV.
• Differential attrition involves uneven dropout rates between groups.
• Research effects are unintended influences from researchers.
• Demand characteristics occur when participants guess the study purpose and change behavior.
• The double-blind technique keeps both researcher and participant unaware of conditions, while separating IV and DV administration maintains integrity.

Experimental Designs

• Pre-experimental designs lack random assignment or a control group, so they cannot test for causality.
• True experimental designs include random assignment and a control group, enabling tests for causality.
• Factorial designs include two or more independent variables (IVs), with each IV having at least two levels, resulting in a minimum of four total conditions.
• Main effects are the effect of each IV alone.
• Interactions are the combined effect of IVs.
• Quasi-experimental designs resemble experiments but lack full random assignment.
• They are often used in real-world settings with limited control.
• Quasi-experimental designs have less internal validity than true experiments but more than pre-experimental designs.
• Intervention effect depends on contingent variables.