Notes on Experimental Design: Variables, Groups, Placebo, Blinding, and Ethics

Experiments aim to determine cause-and-effect relationships (why something happens).
Key variables:
- Independent Variable (IV): the factor that is manipulated or assigned by the experimenter. Examples include the type of beverage (coffee, Red Bull, Celsius, water) or level of activity (stairs, laps around the quad).
- Dependent Variable (DV): the outcome that is measured to assess the effect of the IV. Examples include heart rate, time for heart rate to return to baseline, or GPA outcomes.
Basic workflow:
- Identify an IV to manipulate.
- Randomly assign participants to different conditions if possible.
- Measure the DV to compare outcomes across groups.
- Consider the need for a control group to establish a baseline for comparison.

Independent Variable (IV): any variable that is manipulated or assigned. Examples include giving a beverage (coffee, Red Bull, Celsius, water) or assigning different activities.
Dependent Variable (DV): the measurement or outcome observed. Examples include heart rate changes, how long it takes for heart rate to come down, or GPA.
Non-manipulated measurements alone (no intervention): still a comparison of how different activities affect heart rate (e.g., stairs, laps around the quad) without giving a substance.
Relationship cue: the DV is what you measure to see the effect of the IV.

Experimental group: receives the manipulation or treatment (the IV is applied).
Control group: baseline or standard condition, used for comparison; does not receive the active treatment or receives a neutral condition.
Examples:
- Beverage study: experimental group gets coffee or other drinks; control group might receive water (baseline).
- 40-yard dash example: all participants perform the baseline test, then undergo six weeks of strength/speed training; comparison is made against the initial baseline (control condition).
- Involvement and GPA: control group is a diverse sample reflecting baseline levels before involving in activities; experimental group is matched similarly to assess impact of involvement.
Purpose: to quantify how the DV changes relative to the baseline provided by the control group.

Placebo: a treatment that has no active therapeutic ingredient (e.g., a sugar pill) given to participants in a way that they believe they might be receiving an active treatment.
Placebo effect: improvement that occurs because the participant believes they are receiving an effective treatment, not because of the active treatment itself.
Use in studies: often employed to determine if observed effects are due to the actual treatment or to participants’ expectations.

Random assignment: participants are allocated to IV conditions by chance, giving each participant an equal chance of receiving any condition.
Purpose: reduces selection bias and helps ensure groups are comparable on both observed and unobserved variables.
Typical implementation:
- If testing a depression medication, randomly assign participants so that roughly half receive the medicine and half receive placebo.
- Equal probability example: for 100 people, approximately 50 receive the medicine and 50 receive placebo.
  -Notation: for a random assignment with p = 1/2, the number assigned to treatment X follows a Binomial distribution: $X \sim \text{Binomial}(N, p)$ with $N = 100,\ p = \frac{1}{2}$ , so expected treatment group size is $E[X] = Np = 50$ .
Potential bias risk: sometimes the researcher or lab assistant knows which participants are in which group, which can bias measurements or evaluations.
Self-fulfilling bias: researchers may unintentionally influence results to confirm their hypothesis, shaping outcomes rather than letting the data speak for themselves.
Blinding helps mitigate bias (see below).

Confounding variable: an uncontrolled factor that can influence the DV and provide alternative explanations for observed effects.
Example: studying whether listening to music improves test performance by letting participants choose their own music. If grades improve, it’s unclear whether the improvement is due to the act of listening to music, the type of music chosen, or merely the presence of any noise.
Important to control or randomize to isolate the effect of the IV.

Single-blind: participants do not know which treatment they are receiving, reducing placebo effects and participant bias.
- Example: in a depression medication study, participants may not know whether they received the active drug or a placebo.
Double-blind: neither participants nor the experimenters know which participants receive the active treatment or placebo.
Purpose: minimize biases from both participants and researchers, especially when evaluating subjective outcomes or when researchers influence assessments.

Ethics provide the backbone of human-subject research: researchers must minimize risk and avoid harming participants.
Informed consent, clear communication of potential risks, and safeguards for participant safety are essential.
Researchers must design studies to maximize benefits while minimizing harm, and be transparent about limitations and potential biases.
Ethical considerations are ongoing and foundational topics that accompany methods like randomization, blinding, and placebo controls.

Coffee vs caffeine drinks and heart rate: IVs represent different beverages; DV is heart rate response; a control condition (e.g., water) helps determine if caffeine content specifically affects heart rate.
Activity-based heart rate measurement: IV is the level of physical activity; DV is heart rate metrics; no ingestion involved, illustrating a naturalistic IV.
GPA and school involvement: IV could be level of involvement; DV is GPA; control group should reflect baseline characteristics to compare effects.
40-yard dash example: baseline measurement, then intervention (training) over six weeks; comparison to initial performance demonstrates treatment effect versus baseline.
Placebo in medical trials: helps determine if improvements are due to actual pharmacological effect or participant expectations.
Random assignment and data integrity: ensuring equal distribution across groups helps reduce selection bias but must be managed to maintain blinding and minimize bias; data integrity requires careful tracking of who received what without introducing bias.

Variables:
- IV: manipulated variable (e.g., beverage type, activity level)
- DV: outcome measured (e.g., heart rate, recovery time, GPA)
Notation examples:
- IV values: $ext{IV} \in {\text{coffee}, \text{Red Bull}, \text{Celsius}, \text{water}}$
- Group sizes (example for N = 100): $N = 100,\ n<em>t = 50,\ n</em>c = 50$
- Random assignment probability: $p = \frac{1}{2}$
- Outcome distribution under random assignment: $X \sim \text{Binomial}(N, p)$
Key terms to remember:
- IV: manipulated variable
- DV: measured outcome
- Control group: baseline for comparison
- Experimental group: receives manipulation
- Placebo: inert treatment used to test for placebo effects
- Placebo effect: improvement due to belief in treatment
- Confounding variable: uncontrolled factor affecting DV
- Single-blind: subjects unaware of treatment
- Double-blind: neither subjects nor researchers aware of treatment
- Random assignment: equal chance of being in any group to reduce bias
- Ethics: ensure safety, informed consent, and risk minimization in human studies