Source: Chapter 2, Pages 35–68, Dr. McGinty
Tags: scientific method, research methods, psychology, critical thinking, variables, experiments, correlation, descriptive statistics, inferential statistics, ethics
Difficulty: Introductory Prerequisites: None. This is foundational material for the entire course.
This chapter is the toolkit for the rest of PSYCH 1100. Before you study anything about the brain, perception, or consciousness, you need to understand how psychologists figure out what is true and what is not. The chapter walks through the scientific mindset, research design (descriptive, correlational, and experimental methods), basic statistics, and research ethics. If you can distinguish what each method can and cannot tell you, and explain why correlation does not equal causation, you are well positioned for the midterm.
Psychology uses the scientific method to move beyond casual observation. Different research designs answer different kinds of questions: descriptive methods describe, correlational methods measure relationships, and only experiments can establish cause and effect. Statistics help us summarise and interpret data, and ethical guidelines protect participants throughout the process.
Scientific mindset
A disciplined approach to understanding the world that relies on systematic observation, measurement, and testing rather than intuition or casual observation. Think of it as the difference between "I reckon people sleep badly before exams" and "Let me design a study to find out."
Critical thinking
The practice of evaluating claims by asking structured questions before accepting them. In this course, there are five core critical thinking questions you should be able to recall and apply to any claim.
Theory
A well-supported explanation of observed phenomena that generates testable predictions. In simple terms, a theory is a big-picture framework that organises what we already know and points toward what we should test next.
Hypothesis
A specific, testable prediction derived from a theory. Think of it as the theory's "if this is true, then we should see..." statement.
Peer review
The process by which other experts evaluate a researcher's work before it is published. It acts as a quality filter, catching errors and weak reasoning before findings enter the wider literature.
Replication
Repeating a study to see whether the same results emerge. If a finding cannot be replicated, its credibility drops sharply.
Case study
An in-depth investigation of a single individual, group, or event. Useful for rare phenomena, but findings may not generalise to the broader population.
Naturalistic observation
Watching and recording behaviour in its natural setting without interference. Good for describing what happens in the real world, but you cannot draw causal conclusions from it.
Survey
A method of collecting self-reported data from a sample, typically through questionnaires or interviews. Efficient for large samples, but vulnerable to response biases.
Correlation
A statistical measure of the relationship between two variables, expressed as a value between -1 and +1. A positive correlation means both variables move in the same direction; a negative correlation means they move in opposite directions; zero means no linear relationship.
Third variable problem
The possibility that an unmeasured variable is responsible for the apparent relationship between two other variables. In simple terms, just because ice cream sales and drowning rates both rise in summer does not mean ice cream causes drowning; temperature is the lurking third variable.
Independent variable (IV)
The variable the experimenter deliberately manipulates to observe its effect.
Dependent variable (DV)
The variable the experimenter measures as the outcome.
Random assignment
Allocating participants to experimental conditions by chance, so that pre-existing differences between individuals are spread evenly across groups. This is what allows experiments to support causal claims.
Confounding variable
An uncontrolled variable that co-varies with the independent variable, making it impossible to know which one caused the observed effect.
Meta-analysis
A statistical technique that combines results from multiple studies on the same topic to identify an overall pattern. Think of it as a study of studies.
Reliability
The consistency of a measure. A reliable measure gives similar results each time it is used under the same conditions. Think of a bathroom scale: if it reads 70 kg three times in a row, it is reliable (even if you weigh 72 kg, which would make it not valid).
Validity
The extent to which a measure actually captures what it claims to measure. A valid IQ test measures intelligence, not just vocabulary.
Descriptive statistics
Numbers that summarise a dataset: measures of central tendency (mean, median, mode) and measures of spread (variance, standard deviation).
Inferential statistics
Statistical methods that allow researchers to draw conclusions about a population from a sample, including whether results are likely due to chance.
Normal curve (normal distribution)
A symmetrical, bell-shaped distribution where most scores cluster near the mean. Also called a bell curve.
Informed consent
The ethical requirement that participants understand the nature of a study and agree to take part voluntarily before the research begins.
The scientific mindset is systematic, objective, and self-correcting. Casual observation is unsystematic, prone to bias, and rarely tested.
We need the scientific mindset because humans are naturally subject to confirmation bias, anecdotal reasoning, and overconfidence in personal experience.
Science demands that claims be falsifiable: you must be able to imagine evidence that would prove them wrong.
The five critical thinking questions provide a framework for evaluating any psychological claim before accepting it.
These questions push you to consider the source of a claim, whether the evidence supports it, whether alternative explanations exist, and whether the claim has been independently verified.
Scientific theories are broad explanations supported by a body of evidence. A good theory is parsimonious (simple), falsifiable, and generates testable hypotheses.
Generating good hypotheses means translating a theory into a specific, measurable prediction that can be tested through research.
Evaluating hypotheses requires collecting data and determining whether results support or contradict the prediction.
Peer review matters because it subjects findings to scrutiny by independent experts before publication. It catches methodological flaws, statistical errors, and unsupported conclusions.
Replication matters because a single study can produce a fluke result. When findings replicate across different samples, settings, and researchers, confidence in them grows.
Descriptive methods answer "Does this exist?" and "What does it look like?" They cannot establish cause and effect.
Case study: An in-depth look at one individual or event. Use it when the phenomenon is rare or when deep detail matters (e.g., studying a patient with unusual brain damage). It cannot tell you whether findings apply to people in general.
Naturalistic observation: Watching behaviour in its natural environment without intervening. Use it when you want ecological validity, to see what people do in the real world. It cannot tell you why they do it or what causes the behaviour.
The survey: Collecting self-report data from many people via questionnaires or interviews. Use it when you need data from a large sample quickly. It cannot establish causation, and it is vulnerable to social desirability bias and poorly worded questions.
Watch out for: poorly operationalised variables. If your measure of "happiness" is vague, your data will be vague.
Correlational studies measure the relationship between two or more variables. They answer "How do variables relate to each other?" but cannot answer "Does X cause Y?"
Positive correlation: Both variables increase together. Example: hours studied and exam scores.
Negative correlation: As one variable increases, the other decreases. Example: hours of sleep deprivation and cognitive performance.
Zero correlation: No systematic relationship. Example: shoe size and political preference.
The third variable problem: An unseen variable may be driving the apparent relationship. Ice cream sales and crime rates both rise in summer, but temperature (the third variable) explains both.
Watch out for: the temptation to infer causation from correlation. "Correlation does not imply causation" is one of the most frequently tested ideas on introductory psychology exams.
Experiments are the only method that can establish cause-and-effect relationships.
Necessary components: An independent variable (manipulated), a dependent variable (measured), at least two groups (experimental and control), and random assignment of participants to groups.
Random assignment ensures that any pre-existing differences between participants are distributed evenly, so group differences in the outcome can be attributed to the manipulation.
Confounding variables are uncontrolled factors that vary along with the IV. Researchers minimise them through random assignment, standardised procedures, single-blind and double-blind designs, and the use of control groups.
Meta-analyses combine results from many studies to identify overall patterns and effect sizes. They provide stronger evidence than any single experiment.
No single study or method is perfect. The strongest conclusions come from converging evidence across different methods (descriptive, correlational, experimental) and different laboratories.
Cross-sectional design: Compares different age groups at a single point in time. Quick, but differences may reflect generational effects rather than ageing itself.
Longitudinal design: Follows the same group over an extended period. Tracks real change over time, but expensive, time-consuming, and subject to participant dropout.
Reliability = consistency. Does the measure give the same result each time?
Validity = accuracy. Does the measure capture what it is supposed to capture?
A measure can be reliable without being valid. A scale that always reads 2 kg too heavy is consistent (reliable) but inaccurate (not valid). A measure cannot be valid without first being reliable.
Classic analogy: darts clustered tightly together but away from the bullseye = reliable, not valid. Darts scattered all over = neither reliable nor valid. Darts clustered on the bullseye = both reliable and valid.
Central tendency summarises where scores cluster:
Mean: The arithmetic average. Sensitive to extreme scores (outliers pull it up or down).
Median: The middle score when data are ranked. Resistant to outliers, so more useful with skewed data.
Mode: The most frequently occurring score. The only measure suitable for categorical data.
Variance and standard deviation describe the spread of scores around the mean. Higher variance = more spread.
The normal curve (bell curve): A symmetrical distribution where most scores fall near the mean, with progressively fewer scores at the extremes. Many psychological variables approximate this shape.
Descriptive statistics with two variables: Scatterplots and correlation coefficients show how two variables relate.
Descriptive statistics summarise the data you have. Inferential statistics let you generalise from a sample to a population.
They answer: "Is the difference between groups large enough that it probably did not happen by chance?"
Key concept: statistical significance. A result is statistically significant when the probability of it occurring by chance alone falls below a set threshold (typically p < .05).
Human participants:
Informed consent: Participants must be told what the study involves and agree voluntarily.
Right to withdraw: Participants can leave at any time without penalty.
Confidentiality: Personal data must be protected.
Debriefing: After the study, participants are told its true purpose, especially if any deception was used.
Minimal risk: The study should not expose participants to harm beyond what they encounter in daily life, unless the potential benefits clearly justify it and participants are fully informed.
Institutional Review Board (IRB): An independent committee that reviews research proposals to ensure they meet ethical standards before data collection begins.
Animal subjects:
Research with animals must be justified by the potential scientific value.
Animals must be treated humanely, with pain and distress minimised.
Ethical guidelines govern housing, care, and procedures.
Students often think correlation means causation. It does not. A correlational study can tell you that two variables are related, but it cannot tell you that one causes the other.
Students often confuse reliability with validity. A measure can be perfectly reliable (consistent) and still not measure what it claims to measure.
Students sometimes believe that a case study can be generalised to the whole population. A single case provides depth, not breadth.
Students often think random assignment and random sampling are the same thing. Random sampling is about who gets into the study (representativeness). Random assignment is about who goes into which condition (controlling for confounds).
⚠️ Be able to identify which research method (descriptive, correlational, experimental) a scenario describes, and state what conclusions it can and cannot support.
⚠️ "Correlation does not imply causation" appears in virtually every introductory psychology exam. Be ready to explain why, and to identify the third variable problem in examples.
⚠️ Know the difference between reliability and validity, and be able to use the dart-board analogy.
⚠️ Understand what each measure of central tendency tells you and how outliers affect the mean but not the median.
⚠️ Be able to list the components of an experiment: IV, DV, control group, experimental group, random assignment.
True or False: A correlational study can prove that watching violent TV causes aggression. False. Correlational studies cannot establish causation.
Fill in the blank: A measure that gives consistent results each time is called ________. Reliable.
True or False: The median is more affected by outliers than the mean. False. The mean is more affected by outliers.
Fill in the blank: The variable the experimenter manipulates is the ________ variable. Independent.
True or False: Random assignment helps control for confounding variables. True.
Q: A researcher finds that students who eat breakfast score higher on exams. Can the researcher conclude that eating breakfast causes better exam performance? Why or why not?
A: No. This is a correlational finding. A third variable (e.g., students who eat breakfast may also have more structured routines, more sleep, or higher socioeconomic status) could explain both behaviours. Only an experiment with random assignment could support a causal claim.
Q: What is the difference between the independent variable and the dependent variable?
A: The independent variable is what the researcher manipulates (the presumed cause). The dependent variable is what the researcher measures (the presumed effect).
Q: A researcher gives one group of participants a new study technique and another group their usual method, then compares test scores. What type of research design is this?
A: This is an experiment, provided participants were randomly assigned to the two groups.
Q: Explain how a measure can be reliable but not valid.
A: If a bathroom scale always reads 3 kg too high, it gives the same (wrong) result every time. The readings are consistent (reliable) but do not reflect the person's true weight (not valid).
Q: A study reports a correlation of -0.85 between stress and immune function. What does this tell you?
A: It tells you there is a strong negative correlation: as stress increases, immune function tends to decrease. It does not tell you that stress causes reduced immune function.
Q: Why is random assignment important in an experiment?
A: Random assignment distributes individual differences (personality, prior knowledge, health) evenly across conditions, so any observed difference in the dependent variable can be attributed to the independent variable rather than to pre-existing group differences.
Q: Name the three types of descriptive research methods discussed in this chapter and give one limitation of each.
A: Case study (limited generalisability), naturalistic observation (no control over variables, cannot determine cause), and survey (vulnerable to response bias and poorly worded questions).
This chapter connects to every subsequent topic in the course. When you study the brain (Ch. 4), perception (Ch. 5), or consciousness (Ch. 6), you will encounter references to specific experiments, correlational findings, and brain-imaging studies. Understanding what each method can and cannot tell you will help you evaluate those claims rather than just memorise them. The statistics concepts here also reappear when interpreting study results throughout the course.
Scientific method, hypothesis testing, falsifiability, peer review, replication crisis, case study, naturalistic observation, survey method, correlation coefficient, positive correlation, negative correlation, zero correlation, third variable problem, confound, independent variable, dependent variable, control group, experimental group, random assignment, random sampling, meta-analysis, reliability, validity, mean, median, mode, variance, standard deviation, normal distribution, bell curve, inferential statistics, descriptive statistics, statistical significance, p-value, informed consent, debriefing, IRB, ethical research, cross-sectional design, longitudinal design, operationalisation