Statistical Reasoning, Research Design, and Ethics in Psychology

Non-Experimental Methods:
- Case Studies, Naturalistic Observations, and Surveys: Observe and record behavior without manipulating variables. Weaknesses include a lack of variable control and potentially misleading single-case findings.
- Correlational Studies: Collect data on $2$ or more variables to detect naturally occurring relationships and assess predictive power without variable manipulation. Weakness: cannot establish cause and effect.
Experimental Research: Explores cause-and-effect relationships by manipulating $1$ or more independent variables using random assignment. Weaknesses include potential generalization limits and ethical restrictions on variable manipulation.
Data Types:
- Quantitative: Uses numerical data to represent variable degrees (e.g., Likert scales from "strongly disagree" to "strongly agree").
- Qualitative: Uses in-depth narrative data (e.g., structured interviews with preset questions).
Predicting Everyday Behavior: Laboratory experiments test theoretical principles under controlled settings rather than directly replicating exact real-world scenarios.

Purpose: Measure and describe the specific characteristics of a studied group (e.g., unit exam scores).
Graphing & Scale Impact: Changing graph scale baselines (e.g., starting a y-axis at 95\text{\n%} versus 0\text{\n%}) drastically alters visual perception of the exact same data set.

Measures of Central Tendency:
- Mean: Arithmetic average of a data set. Note that skewed data reduces the reliability of the mean.
- Median: Exact midpoint score of a data set.
- Mode: Most frequently occurring score. A distribution with $2$ modes is bimodal.
- Percentile Rank: Percentage of scores falling below a given score.
Measures of Variation:
- Range: Difference calculated by subtracting the lowest score from the highest score.
- Standard Deviation: Computed statistic indicating how much individual scores vary around the mean. Low variability yields more reliable averages.
- Normal Curve: Symmetrical, bell-shaped distribution where most scores cluster near the mean.

Purpose: Uses numerical sample data to determine the probability that a finding is true for an entire population.
Rules for Generalizability:
1. Representative samples are superior to biased (unrepresentative) samples.
2. Larger samples yield better reliability than smaller samples.
3. More estimates or replicated studies (meta-analyses) provide stronger conclusions than fewer estimates.
Statistical Significance: A judgment of how likely an observed difference between control and experimental groups occurred by chance, assuming no actual difference exists in the underlying population.
P-values: Results are considered statistically significant when the probability of occurrence by chance is small ( $p < 0.05$ ).
Effect Size: Indicates the strength of the relationship between $2$ variables.

Animal Subject Guidelines:
- British Psychological Society (BPS): Mandates housing animals in reasonably natural living environments with companions for social species.
- American Psychological Association (APA): Requires humane care, healthful conditions, and minimization of discomfort during testing.
- Research proposals must receive prior Institutional Review Board (IRB) or animal care ethics committee approval.
Human Subject Guidelines:
- Informed Consent / Assent: Providing sufficient information for subjects to make rational decisions about participating.
- Protection: Safeguarding participants from potential physical or psychological harm and discomfort.
- Confidentiality: Keeping participant records and data strictly private.
- Debriefing: Fully disclosing the study's true purpose and explaining any necessary deception after participation ends.
- Avoid Coercion: Restricting excessive compensation (e.g., extreme financial rewards) that forces participation.
- Deception & Confederates: Deception and fake participants (confederates) are permitted only when essential to test hypotheses without bias.
- IRB Composition: Requires at least $5$ members: $1$ scientist, $1$ non-scientist, and $1$ community representative.
Research Integrity: Data falsification constitutes fraud and causes severe real-world harm (e.g., fraudulent claims linking vaccines and autism).

Question 1: What is the mean of the data set $33, 40, 12, 25, 80$ ?
- Response: $38$ (calculated by $\frac{33 + 40 + 12 + 25 + 80}{5} = \frac{190}{5} = 38$ ).
Question 2: What is the mode of the data set $33, 40, 12, 25, 80$ ?
- Response: There is no mode, as all numbers appear exactly once.
Question 3: How do descriptive statistics compare to inferential statistics?
- Response: Descriptive statistics summarize data, while inferential statistics assess if data can be generalized to broader populations.
Question 4: What must a researcher do to fulfill the ethical principle of informed consent?
- Response: Provide participants with enough information about a study to enable them to make a rational decision about whether to participate.
Question 5: Which ethical principle requires that participants be told about the true purpose of the research at the end of the study?
- Response: Debriefing.
Question 6: Which proposed animal study is most likely to meet ethical principles and receive IRB approval?
- Response: Studying whether dolphins can learn simple language.