Chapter 2 Notes: Participant and Experimenter Bias, Blinding, and Sampling

Definition: participant bias refers to when a participant's awareness of being studied changes their behavior.
Consequence: participants may behave differently in the lab than they would in the real world because they know they are being observed.
Related bias: social desirability bias, where participants try to help confirm the researchers' hypotheses by behaving in a 'better' or more acceptable way.
Core issue: we want to observe real behavior, not behavior influenced by the knowledge of being studied.

Goal: make participation and observation as unobvious as possible to participants.
Living-room laboratory example (graduate study):
- Room designed to feel like someone’s living room, not a sterile lab.
- Details: walls painted in non-stark colors, curtains, big comfortable sofas and chairs, coffee table with magazines, plants, pictures on the wall.
- Rationale: people feel more comfortable and behave more naturally when they feel they are in a familiar, non-clinical space.
Observation unobtrusiveness example:
- Historically: researchers would be in a different room than participants with a glass partition (one-way mirror) so participants could not see the observer.
- Modern approach: cameras in the room allow observation without the observer being visibly present in the room.
- Rationale: avoiding the sense of being watched reduces the likelihood of altered behavior.

Principle: participants should not know the hypothesis or what is expected, so they do not tailor their behavior to help or hinder the hypothesis.
Reason: knowing the hypothesis can bias participants to act in ways that confirm or disconfirm it, depending on their beliefs or motivations.
Outcome: when participants are unaware of the hypothesis, their best bet is to act as naturally as possible.

Definition: experimenter bias occurs when the researcher's behavior changes the participant's behavior or otherwise influences the study, biasing measurements.
Classic example: Clever Hans the horse.
- Wilhelm von Osten believed animals could do math and used Hans to perform arithmetic.
- Hans appeared to respond correctly to addition, subtraction, multiplication, division, and even square roots by stamping hooves the right number of times.
- A psychologist investigated and found the cue: von Osten subtly nodded or inclined head as Hans approached the correct answer; Hans picked up on these cues and matched the expected response.
- Result: the horse did not actually perform math; the experimenter’s cues led to the illusion of ability.
Lesson: researcher behavior can inadvertently cue participants or subjects and thus distort results.
Solution: to prevent experimenter bias, implement blinding so the experimenter does not know the hypothesis.

Definition: a double-blind study is one in which both the participant and the experimenter are unaware of the hypothesis.
Purpose: if the experimenter does not know the hypothesis, they cannot influence participants toward confirming or disconfirming it.
Practical note: in many psychology studies, the person who designs the hypothesis is a professor, while the person running the study is a graduate student, undergraduate, or staff who should not be told the hypothesis until after data collection is complete.
Outcome: reduces both participant and experimenter biases, improving internal validity.

Typically, professors at research universities come up with the hypotheses to test.
The actual data collection and participant interaction are often conducted by subordinates (graduate students, undergraduate students, or paid staff).
These individuals usually are not told the hypothesis until the study is over, supporting the integrity of the blinding process.

Population: the entire group to which researchers want their study results to generalize.
Sample: the subset of individuals actually tested in the study.
Practical constraint: studying an entire population is often infeasible; hence a sample is used.
Example: citizens of Westchester County.
- Population: all residents of Westchester County.
- Sample: a smaller group (e.g., 100–200 individuals) drawn from that population for the study.

Definition: a representative sample is one that resembles the population in all relevant characteristics.
Rationale: to generalize results from the sample to the population, the sample should reflect the population's composition.
Illustration with race/ethnicity: if the population distribution is 25% White, 25% Black, 25% Asian, 25% Latino, then the sample should mirror this distribution:
- $25\%$ White, $25\%$ Black, $25\%$ Asian, $25\%$ Latino.
This principle applies not only to race/ethnicity but to every demographic characteristic: age, gender, sexual orientation, social class, and so on.

Participant bias and experimenter bias are universal concerns in both experimental and correlational studies.
Unobtrusiveness and blinding (including the double-blind design) are foundational strategies to protect against these biases.
The origin of the hypothesis often lies with senior researchers, while the data collection is performed by others who should remain blind to the hypothesis during the study.
A representative sample is essential for generalizing findings beyond the study sample to the broader population.
Always consider how the sampling strategy and experimental design might influence participants’ natural behavior and the researchers’ potential influence on outcomes.