Research & Research Design Notes

Research and Research Design

The presentation is by Dr. Megan Kelly from the University of Wollongong, Australia.
The main focus is on understanding what is essential for designing good research studies.

Population: The entire group that is of interest in a research study.
Sample: A subpopulation selected from the larger group that is accessible for the study.
Target Population: Refers to the population of interest from which the sample is derived.
Exclusion Criteria: Characteristics that disqualify individuals from being part of the sample population.
Sample Population: Those individuals who are actually included in the study following the inclusion criteria.
Inclusion Criteria: Specific characteristics that allow individuals to participate in the research study.

Bias: Refers to sources of interference that can skew study results and conclusions. Bias can occur at different stages:
- Selection Bias: Occurs when the researcher selects participants in a way that affects the results.
- Performance Bias: Results when participants modify their behavior because they know they are being studied.
- Recall Bias: Arises when individuals have difficulty accurately recalling past events affecting data reliability.
- Attrition Bias: Concerns the effects of a higher dropout rate in cohort studies, which can distort findings.
Confounding: Occurs when the relationship between an exposure and an outcome is influenced by other variables (confounders). A confounder must meet two criteria:
- It is associated with the exposure.
- It is independently related to the outcome of interest.
Diagrams can help visualize how exposure, outcome, and confounders interact in research studies.
Validity: Ensures that the study measures what it intends to measure.
- High Validity / High Reliability: Study accurately assesses intention and produces consistent results.
- High Validity / Low Reliability: Accurate but inconsistency in measurements.
- Low Validity / High Reliability: Consistent but inaccurate.
- Low Validity / Low Reliability: Neither accurate nor consistent measurements.

Placebo Effect: Positive physiological response by a patient who receives no active treatment (placebo).
Nocebo Effect: Negative response by a patient who experiences harm or discomfort due to a placebo, stemming from the expectation of negative outcomes.
Blinding: Critical methodology that involves keeping participants, researchers, and statisticians unaware of treatment assignments to minimize biases.

Benefits:
- Capable of gathering data from a large sample size.
- Allows for comparisons across different groups.
- Supports the use of statistical techniques for relations between variables.
- Enables generalization to broader populations.
- Generates numerical or rating information quickly and cost-effectively.
Limitations:
- May struggle to recognize new phenomena.
- Caution is needed when interpreting results without control groups.
- Requires participant literacy.
- Generalizability to the broader population may be limited.
- Difficulties can arise in applying statistical results to conclusions.

Definition: Involves asking questions aimed at understanding concepts, perceptions, feelings, and experiences.
Benefits:
- Helps in developing theories and exploratory insights.
- Provides a deeper understanding of mechanisms behind observed behaviors.
- Offers anecdotal insights into the human experiences that can complement quantitative findings.
- Data may come from open-ended surveys, focus groups, interviews, observations, and case studies.
Limitations:
- Challenges in quantifying findings due to the less structured format.
- Typically includes a smaller number of participants.

Case-Control Study: Involves comparing ‘cases’ (those who have a disease) versus ‘controls’ (those who do not have the disease). Key issues to consider include causation and the potential for recall bias.
Cohort Study: Observational study design that follows subjects over time to determine how exposure to certain factors impacts disease outcomes, accounting for attrition bias. This method can be expensive and resource-intensive.
Randomized Control Trial (RCT): Participants are randomly assigned to either a treatment or control group, which helps to reduce biases such as selection and performance bias. This is an experimental design often viewed as the gold standard in clinical research.

Methodical search through existing literature to identify all relevant studies related to a specific topic.
Each study's quality is appraised and the results are synthesized in a coherent manner:
- Include the objectives of the review and the eligibility criteria.
- Perform searches across databases, screen for eligibility, and tabulate data.
- Address reasons for exclusions and assess potential biases in included studies.
- If applicable, conduct a meta-analysis to statistically synthesize results from eligible studies and assess overall efficacy.

Summarized identification of studies:
- Consider tracking prior versions of reviews and identifying new studies through various databases and registers, citation searches, and other methods.
- Documenting the number of total records identified, screened, and excluded for various reasons is essential for transparency and reproducibility in research.
- Highlight the importance of maintaining rigorous standards to ensure the accuracy and completeness of systematic reviews, meta-analyses, and their methodologies.