Quantitative vs qualitative data
Quantitative: numerical (scores, times, counts). Easy to analyse statistically; can lose nuance. Examples: number of words recalled, reaction time, Likert score.
Qualitative: non-numerical (transcripts, descriptions). Rich detail; harder to compare. Examples: interview transcripts, open-question responses, observational notes.
Good research often combines both — e.g. a Likert scale (quantitative) plus an open follow-up question (qualitative).
Primary vs secondary data
Primary: collected directly by the researcher for this study. Tailored; current; expensive. Secondary: data collected previously by others (datasets, government statistics, prior studies). Cheap and broad; may not match current question; quality dependent on original collection.
Measures of central tendency
Three summary numbers that describe the "average" of a dataset:
- Mean: sum ÷ number of values. Uses every value; sensitive to outliers.
- Median: middle value when ranked. Robust to outliers.
- Mode: most common value. The only sensible "average" for categorical data.
Choice depends on the data. With outliers (e.g. one participant scoring 30 when others score 5–8), use the median. With nominal data (favourite subject, eye colour), use the mode.
Measure of dispersion
Range: highest minus lowest value. Quick but vulnerable to a single outlier.
A fuller picture combines a measure of central tendency with a measure of dispersion: "the average score was 12 (mean), with values ranging from 4 to 22 (range)."
✦Worked example
Dataset: 3, 5, 7, 7, 9, 11, 18.
- Sum = 60. n = 7. Mean = 60 ÷ 7 ≈ 8.57.
- Sorted: 3, 5, 7, 7, 9, 11, 18 → Median = 7.
- Mode = 7 (appears twice).
- Range = 18 − 3 = 15.
The outlier (18) pulls the mean above the median, suggesting the median is the safer summary for this dataset.
⚠Common mistakes— Common errors
- Calculating the mean of nominal categories.
- Reporting the mean without a measure of dispersion (numbers can mislead).
- Confusing the mode with "most" (mode is the most frequent value, not the largest).
AI-generated · claude-opus-4-7 · v3-deep-psychology