revision cards

Domain 1: Data Types & Measurement

• Nominal Data: Categorical data with no inherent order or numerical value (e.g., gender, ethnicity, or "Yes/No").

• Ordinal Data: Data that can be ranked or ordered, but the mathematical distance between points is unequal (e.g., 1–10 pain scales or race finishing positions).

• Interval/Ratio Data: Continuous numerical data with equal distances between points. Ratio data has a "true zero" indicating the total absence of the trait (e.g., height, weight, or time).

• Mode, Median, Mean: Mode is the most frequent score; Median is the middle score (best for skewed data); Mean is the arithmetic average (sensitive to outliers). Example: In a dataset (2, 2, 5, 7, 34), the Mean (10) is pulled high by 34, while the Median (5) better represents the "middle."

• Independent Variable (IV): The variable that is manipulated, changed, or used to group participants. It is the hypothesized "cause." Example: In a study on drug effectiveness, the IV is the Dosage (0mg, 50mg, 100mg).

• Dependent Variable (DV): The variable that is measured or observed to see if it changes. It is the "effect" or outcome. Example: In the same drug study, the DV is the Symptom Severity Score measured after 4 weeks.

• Quasi-Independent Variable: An IV that cannot be randomly assigned or manipulated by the researcher, such as Gender, Age, or an existing Clinical Diagnosis.

• Confounding Variable: An "extra" variable that wasn't accounted for, which might influence both the IV and DV, potentially ruining the results. Example: In a study on exercise (IV) and weight loss (DV), Caloric Intake is a potential confounder.

• Covariate: A continuous variable that is not the main focus of the study but is "controlled for" statistically (usually in ANCOVA) to see the clear effect of the IV. Example: Controlling for Baseline Anxiety when testing a new therapy.

Domain 2: Distributions & Central Tendency

• Normal Distribution: A symmetrical, bell-shaped curve where the Mean, Median, and Mode are all equal and located at the center.

• Positive Skew: A distribution where the "tail" points toward the right (high numbers). The Mean is pulled higher than the Median by outliers.

• Negative Skew: A distribution where the "tail" points toward the left (low numbers). The Mean is pulled lower than the Median by outliers.

• Skewness: A measure of the asymmetry of a distribution. A value of 0 is perfectly symmetrical; positive values indicate a right-tail skew, and negative values indicate a left-tail skew. +1

• Kurtosis: A measure of the "peakedness" or "flatness" of a distribution. Leptokurtic curves are thin and peaked; Platykurtic curves are broad and flat.

• Floor Effect: Occurs when a test is too difficult, causing most scores to cluster at the very bottom of the scale. This masks differences between participants who might have different low-level abilities.

• Ceiling Effect: Occurs when a test is too easy, causing most scores to cluster at the very top of the scale. This makes it impossible to distinguish between high-performing participants.

• Standard Deviation: A measure of how much scores vary around the mean. High S indicates high spread; low S means scores cluster near the mean.

• Standard Error (SE): Estimates how much the sample mean would fluctuate if you took multiple samples

• z-score: A standardized score representing how many standard deviations a raw score is from the mean.

• z-Standardisation: The process of converting different scales (e.g., IQ and Anxiety scores) into z scores so they can be compared directly on the same metric.

Domain 3: Hypothesis Testing & Errors

• Null Hypothesis (H0): The assumption that there is no effect, no difference, or no relationship in the population.

• Type I Error: A "False Positive." Rejecting the Null Hypothesis when it is actually true (claiming an effect exists when it doesn't).

• Type II Error: A "False Negative." Failing to reject the Null Hypothesis when it is actually false (missing a real effect).

• Statistical Power: The probability of correctly rejecting a false Null Hypothesis. Formula: 1 - beta.

• p-value: The probability of obtaining the observed results if the Null Hypothesis were true. Significant if usually <0.05

Domain 4: Inferential Test Selection

• Independent Samples t-test: Parametric test comparing the means of two unrelated/independent groups (e.g., Control vs. Experimental).

• Mann-Whitney U Test: Non-Parametric version of the independent t-test; used for ordinal data or when normality is violated.

• Paired Samples t-test: Parametric test comparing means from the same group at two different times (e.g., Pre-test vs. Post-test).

• Wilcoxon Signed-Rank Test: Non-Parametric version of the paired t-test; used for related samples with ordinal or non-normal data.

• One-Way ANOVA: Parametric test used to compare the means of three or more independent groups.

• Kruskal-Wallis Test: Non-Parametric version of One-Way ANOVA; used to compare three or more independent groups using ranks.

• ANCOVA: An ANOVA that "controls" for a continuous nuisance variable (a covariate) to isolate the true effect of the IV.

Domain 5: Assumptions and Distribution Checking

• Parametric Assumptions: Data must be Interval/Ratio, Normally Distributed, and have Homogeneity of Variance.

• Non-Parametric Assumptions: Used when data is Ordinal/Nominal or fails parametric assumptions (no normality required).

• Levene’s Test: A test for Homogeneity of Variance. If p < .05, the variances are unequal, and the assumption is violated.

• Shapiro-Wilk & Kolmogorov-Smirnov (K-S): Tests for Normality. A result of p < .05 means the data is NOT normally distributed. Example: If Shapiro-Wilk p = .001$, you should use a Mann-Whitney U instead of a t-test.

Domain 6: Advanced Testing

• Bivariate Linear Regression: Using one independent variable (X) to predict one continuous dependent variable (Y). Example: Predicting Exam Score based on Hours Studied.

• Multiple Regression: Using several independent variables (e.g., Sleep, Stress, Diet) to predict one $Y$.

• Exploratory Factor Analysis (EFA): A method used to find underlying patterns (factors) in a large set of variables. Example: Taking 40 personality questions and finding they group into 5 main traits.

• MANOVA: Multivariate Analysis of Variance. Used when there are multiple Dependent Variables that are related to each other. Example: Testing if a drug affects both Anxiety and Depression levels simultaneously.

Domain 7: Effect Sizes and Relationships:

• Cohen’s d: Measures the magnitude of the difference between two groups in $SD$ units. Guidelines: 0.2 (Small), 0.5 (Medium), 0.8 (Large).

• Odds Ratio (OR): The ratio of the odds of an event happening in one group vs. another. Example: An OR of 2.0 means the treated group is twice as likely to recover.

• Log Odds (Logit): The natural logarithm of the odds; used in Logistic Regression to transform a probability curve into a straight line.

• Chi-Square : A test of association between two categorical variables. Example: Does Gender (M/F) predict Smoking Status (Yes/No)?

Domain 7: Probability and Interference:

• Probability Density Function (PDF): A function used to specify the probability of a continuous variable falling within a particular range of values (the area under the curve).

• Significance (p): The probability that the observed effect occurred by chance if the Null Hypothesis is true. Usually, p < .05 is the cutoff for "significance."

• Statistical Power (1 - beta$: The ability of a test to detect an effect that actually exists. Ideally set at 0.80 (80% chance).

• Bonferroni Correction: Dividing the alpha level (.05) by the number of tests performed to prevent "Type I Error Inflation." Example: If running 5 tests, the new required p-value is .01.

• Bayes' Theorem: A mathematical formula for determining conditional probability; it updates the "prior" probability of a hypothesis as new evidence becomes available.

Domain 8: Relationships & Reliability

• Pearson’s r: Parametric correlation coefficient measuring the strength and direction of a linear relationship between two continuous variables.

• Spearman’s Rho (): Non-Parametric correlation used for ordinal data or non-linear relationships.

• Chi-Square (): A test used to determine if there is a significant association between two categorical (nominal) variables.

• Cronbach’s Alpha (): A measure of Internal Reliability (consistency). It checks if all items in a scale are measuring the same construct (> 0.70 is ideal).

• Coefficient of Determination (): The percentage of variance in the dependent variable that is explained by the independent variable(s) in the model.

• 
  

Qualitative Methodology & Analysis

• Grounded Theory: A systematic methodology used to develop a theory "from the ground up" based on the data collected, rather than testing an existing hypothesis.

• - Open Coding: The first step in Grounded Theory where data is broken into discrete parts, examined, and labeled with initial concepts (e.g., labeling a quote about feeling judged as "Perceived Stigma").

• - Axial Coding: The second step in Grounded Theory where researchers identify relationships and links between the categories and sub-categories identified in open coding.

• - Selective Coding: The final step in Grounded Theory where one "core category" is chosen to unify the entire theory, and all other categories are integrated around it.

• Theoretical Saturation: The point in data collection where new interviews or observations no longer provide new information or insights, signaling that the theory is "full."

  
  

• Interpretative Phenomenological Analysis (IPA): A qualitative approach that explores how people make sense of their major life experiences (idiographic) and emphasizes the researcher's role in that sense-making.

• Double Hermeneutic: A hallmark of IPA where the researcher is trying to make sense of the participant, who is simultaneously trying to make sense of their own world.

• Thematic Analysis: A flexible method for identifying, analyzing, and reporting patterns (themes) within data across a whole dataset, rather than focusing on a single individual's "essence."

• Reflexivity: The process where a researcher reflects on their own biases, values, and presence, and how these might influence the collection and interpretation of qualitative data.

• Member Checking: A technique used to improve credibility where the researcher takes their findings back to the participants to see if the analysis accurately reflects their experiences.

• Inductive Reasoning: A "bottom-up" approach where the researcher moves from specific observations to broader generalizations and theories (the opposite of the "top-down" deductive approach).

• Narrative Analysis: An approach that focuses on the "story" as the unit of analysis. It examines how people use stories to make sense of their lives, looking at the sequence, plot, and character arc in a participant's account (e.g., analyzing a "recovery journey" story).

• Discourse Analysis: A method that examines how language is used in social contexts to create meaning and exercise power. It looks beyond the literal words to see how "discourses" (like "the medical discourse") shape our reality and social hierarchies.

• Conversation Analysis (CA): A highly technical study of "talk-in-interaction." It focuses on the fine-grained mechanics of conversation, such as turn-taking, pauses, overlaps, and repair (e.g., measuring exactly how many milliseconds of silence occur before a person "disagrees").

• Content Analysis: A systematic coding and categorizing of communication used to determine the frequency of certain words, themes, or concepts. It can be qualitative (interpreting meanings) or quantitative (counting occurrences).

• Ethnomethodology: The study of the "methods" people use to produce recognizable social order. It is the theoretical root of Conversation Analysis, focusing on how we "do" social life.