Research lecture mod 4

Intervention
- Question: Is there an intervention?
- Yes: Likely using experimental or quasi-experimental design.
- No: Non-experimental or observational design.

Comparative Nature
- Questions to consider: Are comparisons made between the same participants or different groups?
- Product of comparisons:
- Within-Subject Comparison: Comparing the same participants at different times or conditions (e.g., participant responses as J1s versus S2s).
- Between-Group Comparison: Comparing different participants or groups (e.g., one group of J1s against another).

Control Measures
- Key question: How much control do researchers have over confounding variables?
- Control Options:
- Randomization: Assigning participants randomly to minimize bias.
- Crossover: Participants experience all conditions, helping control for individual differences.
- Homogeneity and Matching: Ensuring groups are comparable based on certain characteristics.
- Statistical Control: Utilizing statistical methods to account for confounders.
- Blinding: Withholding information to reduce bias (e.g., blinding participants or researchers).

Time Frame
- Are data collected at one time point or multiple points?
- Considerations:
- Pre-test/Post-test designs.
- Longitudinal data collection over time.
- Relative Timing:
- Retrospective vs. Prospective design (looking backward vs. forward).

Site Selection
- Determine if the study will take place at:
- A single site (e.g., one college or one hospital) or
- Multiple sites, which may offer broader generalizability.

Study of Cause and Effect
- Central research question: Is one thing causing another?
- Example Questions:
- Does exercise cause improvements in heart health?
- Does high-calorie intake cause weight gain?
- Acknowledgement of multiple causes:
- A complex interplay of factors often affects outcomes (e.g., weight gain factors like activity levels, metabolic conditions).

Definition of Counterfactual:
- Refers to the concept of what would have happened to individuals who received the intervention had they not received it (similar to a control group).
Related Concept - Effect:
- The difference between the actual outcomes resulting from exposure and the projected outcomes had there been no exposure.

Three Main Criteria:
1. Temporal:
  - The cause must precede the effect in time (e.g., smoking before lung cancer).
2. Relationship:
  - A statistical association must exist between cause and effect (e.g., higher percentages of smokers developing lung cancer compared to non-smokers).
3. No Confounding Variable:
  - The observed relationship cannot be explained by a third factor (confounder) that impacts both the cause and effect.
Additional Criterion (Biological Plausibility):
- Evaluation of whether the causal relationship is realistic based on known biological mechanisms (e.g., smoking damaging lung tissues leading to cancer).

Case of Ice Cream Sales and Drownings:
- Identifies an association between two variables (ice cream sales and drownings).
- Possible confounding variable: Weather conditions leading to higher ice cream sales and more people swimming (thus increasing drownings).
- Conclusion: Although an association exists, one does not cause the other; rather, they share a confounding variable (time of year associated with heat).

Therapeutic/Intervention Questions:
- Highest level of evidence through Randomized Controlled Trials (RCT).
Prognostic Questions: To assess potential outcomes based on current states/conditions.
Causal Questions: Addressing alleged causes of diseases or conditions.
Descriptive Questions: Not focused on causality but elucidating characteristics or phenomena without manipulating variables.

Components of True Experimental Design (RCT):
- Intervention: Manipulation of an independent variable to determine its effect on a dependent variable.
- Control: Comparison against a group that does not receive the experimental treatment.
- Randomization: Random assignment of participants to either intervention or control group to eliminate selection bias.

Function of Control Group:
- Used for comparison to evaluate the effects of the intervention group.
- Variations can include:
- No treatment, usual care, alternative interventions, placebo control, or attention control conditions.
- Example of the Snoop Dogg positive affirmations in a study on performance:
- Control group could either receive no affirmations or an alternative treatment such as meditation.

Concept:
- Participants may initially receive no treatment but will eventually be given the treatment later on, allowing researchers to evaluate the treatment's impact over time (also known as waitlisted controls).

Types of Data Collection Designs:
- Post-test Only: Data collected only after the intervention (without baseline measurement).
- Pretest/Post-test: Data collected at baseline and after the intervention to assess changes.
- Crossover Design: Participants exposed to multiple conditions in different orders, serving as their own control group.
- Challenges include potential carryover effects from one condition affecting outcomes in subsequent conditions, as noted in the example of Snoop Dogg's affirmations versus meditation across exams.

Implications for Research Design: Each aspect of design should be chosen carefully based on the research objective and the focus on causal inference requires rigorous criteria and ideas surrounding control and comparison for valid conclusions.