Introduction to Statistics for Psychology
Learning Objectives
- Define statistics and distinguish descriptive from inferential statistics.
- Explain why psychology, a science of behavior, depends on statistical reasoning.
- Identify populations, samples, variables, and levels of measurement in a research scenario.
- Calculate basic descriptive measures (mean, median, mode, standard deviation, range) from raw data.
- Recognize how descriptive and inferential statistics connect to hypothesis testing and correlation.
Quick Answer
Statistics is the set of methods psychologists use to collect, organize, summarize, and draw conclusions from data about behavior and mental processes. It matters because psychology studies things you cannot observe directly — anxiety, memory, intelligence, attitudes — so researchers measure samples of people and use statistics to decide whether a pattern is real or just chance variation. Statistics splits into two branches: descriptive statistics, which summarize what the data show, and inferential statistics, which let researchers generalize from a sample to the wider population. Every psychology experiment you'll read about, from Milgram's obedience studies to modern drug trials for depression, relies on this two-step process to turn raw numbers into trustworthy conclusions.
Why Psychology Needs Statistics
Psychology can't put "happiness" or "aggression" on a scale the way you'd weigh a rock. Instead, researchers operationalize a construct (say, aggression) into something measurable (number of shocks a participant chooses to deliver, or a score on a rating scale), collect that measurement from many people, and then need a rigorous way to say "this difference is meaningful" rather than "this difference could easily have happened by luck." That rigorous way is statistics.
Four jobs statistics does for a psychologist:
- Design — deciding how many participants are needed and how to structure a study so it can actually detect an effect.
- Summary — condensing hundreds or thousands of scores into a few numbers (a mean, a standard deviation) that a reader can grasp instantly.
- Inference — using a sample to make a justified claim about a population the researcher could never fully test.
- Communication — giving researchers a shared, standardized language (p-values, effect sizes, confidence intervals) so findings can be compared across studies.
Populations, Samples, and Variables
Before any calculation happens, you need three building blocks straight:
- Population: the entire group a researcher wants to draw conclusions about (e.g., "all first-year university students").
- Sample: the smaller, actually-measured subset of that population (e.g., 200 first-years from one campus).
- Variable: anything that can take different values across people or occasions (exam score, reaction time, anxiety level).
Variables come in different levels of measurement, and this determines which statistics you're even allowed to use later:
| Level | What it means | Example | Statistics allowed |
|---|---|---|---|
| Nominal | Categories with no order | Gender, diagnosis type | Mode, frequencies |
| Ordinal | Ordered categories, unequal gaps | Class rank, pain scale (1-10) | Median, mode |
| Interval | Ordered, equal gaps, no true zero | IQ score, temperature (°C) | Mean, SD |
| Ratio | Ordered, equal gaps, true zero | Reaction time, number of errors | Mean, SD, ratios |
A common exam trap: students calculate a mean for ordinal data (like a 5-point Likert item) without acknowledging that strictly, only the median/mode are fully justified for ordinal data — though in practice psychologists often treat Likert data as interval-like. Knowing the distinction shows you understand why the rule exists, not just that it exists.
Descriptive Statistics: A Worked Example
Descriptive statistics summarize a data set using measures of central tendency (mean, median, mode) and variability (range, standard deviation).
Take 10 exam scores from a psychology class: 72, 85, 90, 78, 88, 92, 76, 84, 90, 81
Mean (μ): sum of scores ÷ number of scores = (72+85+90+78+88+92+76+84+90+81) / 10 = 836 / 10 = 83.6
Median: sort the data → 72, 76, 78, 81, 84, 85, 88, 90, 90, 92. With 10 values (even), the median is the average of the 5th and 6th values: (84+85)/2 = 84.5
Mode: the value appearing most often → 90 (appears twice)
Range: highest − lowest = 92 − 72 = 20
Standard deviation: measures how far, on average, scores stray from the mean. Steps: subtract the mean from each score, square each deviation, average the squared deviations (dividing by n−1 for a sample), then take the square root. For this data set, SD ≈ 6.9, meaning most scores fall within about 7 points of 83.6 — a fairly tightly clustered class.
Notice the mean (83.6) and median (84.5) are close, suggesting a roughly symmetric distribution with no extreme outliers pulling the mean away from the middle.
Inferential Statistics: The Other Half
Descriptive statistics stop at "here is what this sample looks like." Inferential statistics ask, "can I trust that this sample reflects the population?" This is where hypothesis testing, p-values, and confidence intervals come in — tools covered in depth in later chapters. For now, the key idea to hold onto is:
Descriptive statistics describe your data. Inferential statistics let you generalize beyond your data.
Real-World Application
Pharmaceutical trials for antidepressants follow exactly this pipeline: researchers measure depression scores (e.g., on the Beck Depression Inventory) before and after treatment in a sample, compute descriptive statistics for each group, then run an inferential test to see if the drug group's improvement is statistically greater than the placebo group's. Without statistics, "the drug seemed to help some people" could never become "the drug produces a reliable, generalizable reduction in symptoms" — the standard the FDA and journals demand.
Key Terms
| Term | Definition |
|---|---|
| Population | The complete group a researcher wants to draw conclusions about |
| Sample | A subset of the population that is actually measured |
| Variable | A measurable characteristic that can differ across people or occasions |
| Descriptive statistics | Methods for summarizing and describing a data set (mean, median, mode, SD, range) |
| Inferential statistics | Methods for drawing conclusions about a population from sample data |
| Level of measurement | The type of scale (nominal, ordinal, interval, ratio) a variable is measured on |
| Mean | The arithmetic average of a set of scores |
| Standard deviation | A measure of how spread out scores are around the mean |
Common Mistakes
-
Misconception: "Statistics just proves things are true." Why it's wrong: Statistics never proves anything with certainty — it deals in probabilities. A significant result means a pattern is unlikely to be due to chance, not that it's guaranteed true. Correct: Statistical results should be described as "providing evidence for" or "being consistent with" a hypothesis, always with an acknowledged margin of uncertainty.
-
Misconception: A larger mean difference between two groups always means a more "important" psychological effect. Why it's wrong: A raw difference doesn't account for variability or sample size. A huge mean gap with huge variability might be less statistically and practically meaningful than a smaller, consistent gap. Correct: Effect size and statistical significance, not the raw difference alone, determine how meaningful a result is.
-
Misconception: Descriptive and inferential statistics are interchangeable terms for "doing stats." Why it's wrong: They answer different questions — one describes the sample in hand, the other generalizes beyond it. Correct: Always ask "am I summarizing my data, or trying to say something about a population?" before choosing which type of statistic applies.
Comparison and Connections
| Feature | Descriptive Statistics | Inferential Statistics |
|---|---|---|
| Purpose | Summarize the sample | Generalize to the population |
| Example tools | Mean, median, mode, SD, range | t-tests, ANOVA, correlation, regression, confidence intervals |
| Question answered | "What does my data look like?" | "Can I trust this pattern applies more broadly?" |
| Certainty | Exact (calculated directly) | Probabilistic (based on likelihood) |
| Typical output | A single summary number | A p-value, test statistic, or confidence interval |
Practice Questions
Recall
- Define "population" and "sample," and explain how they differ. Answer guidance: Population = entire group of interest; sample = the subset actually measured. The sample should represent the population.
- Name the four levels of measurement and give one psychological example of each. Answer guidance: Nominal (diagnosis category), ordinal (pain rating), interval (IQ score), ratio (reaction time).
Understanding 3. Explain why psychologists cannot simply measure an entire population when studying something like "national stress levels." Answer guidance: Populations are usually too large/inaccessible to measure fully; time, cost, and practicality require sampling, which is why inferential statistics exist — to responsibly generalize from the sample. 4. Why might the mean be a misleading measure of central tendency for a skewed data set (e.g., household income)? Answer guidance: Extreme outliers pull the mean toward them; the median is more robust to skew and better represents a "typical" value.
Application 5. A researcher records the number of therapy sessions attended by 8 clients: 4, 6, 5, 20, 5, 6, 4, 6. Calculate the mean and median, and explain which better represents a "typical" client. Answer guidance: Mean = (4+6+5+20+5+6+4+6)/8 = 56/8 = 7; Median = order the data (4,4,5,5,6,6,6,20), average of 4th/5th = (5+6)/2 = 5.5. The median better represents typical attendance because the outlier (20) inflates the mean. 6. A study measures "satisfaction" on a 5-point scale (1 = very dissatisfied to 5 = very satisfied). What level of measurement is this, and what statistics are strictly appropriate? Answer guidance: Ordinal; strictly median/mode/frequencies, though means are commonly reported in practice with caveats.
Analysis 7. Compare descriptive and inferential statistics in terms of the kind of claim each allows a researcher to make. Answer guidance: Descriptive statistics support only claims about the specific sample measured; inferential statistics support probabilistic claims about the broader population, with an explicit margin of uncertainty. 8. A news headline claims "Study proves social media causes anxiety" based on a correlational study. Identify the statistical/logical error. Answer guidance: Correlation does not establish causation, and "proves" overstates what any statistical test can claim; inferential statistics estimate probability, not certainty.
FAQ
Do I need to be good at math to understand psychology statistics? No. Most psychology statistics courses focus on conceptual understanding and interpretation rather than complex computation — software does the calculating. What matters is knowing which test to use and what the output means.
What's the difference between a parameter and a statistic? A parameter describes a population (usually unknown, symbolized with Greek letters like μ), while a statistic describes a sample (symbolized with Roman letters like x̄). We use sample statistics to estimate unknown population parameters.
Why do psychologists care so much about sample size? Larger, well-selected samples produce more stable estimates and greater power to detect real effects, reducing the risk that a result is a fluke of a small, unrepresentative group.
Is standard deviation the same as variance? No — variance is the average of squared deviations from the mean; standard deviation is the square root of variance, which brings the units back to the original scale (e.g., points, seconds), making it easier to interpret.
Can descriptive statistics ever be "wrong"? Not in a mathematical sense — a correctly calculated mean is always accurate for that data set. But they can be misleading if the data are skewed, have outliers, or the wrong measure (mean vs. median) is chosen for the situation.
Quick Revision
- Statistics = collecting, organizing, analyzing, and interpreting data.
- Two branches: descriptive (summarize the sample) and inferential (generalize to the population).
- Population = entire group of interest; sample = the subset measured.
- Four levels of measurement: nominal, ordinal, interval, ratio — each allows different statistics.
- Mean = average; sensitive to outliers.
- Median = middle value; robust to outliers and skew.
- Mode = most frequent value; only measure usable for nominal data.
- Range = highest − lowest; standard deviation = typical distance of scores from the mean.
- Descriptive stats never "prove" anything — they only summarize what was observed.
- Inferential stats use probability, not certainty, to generalize findings.
- Statistics underpins research design, data analysis, and how findings are communicated across the field.
Related Topics
Prerequisites: Basic arithmetic and understanding of research methods (samples vs. populations, variables).
Related: Descriptive Statistics, Research Design and Measurement Scales.
Next: Descriptive Statistics (measures of central tendency and variability in depth), followed by Inferential Statistics and Hypothesis Testing.