Internal vs. External Validity: A Student-Friendly Guide

Every psychology course eventually hands you a study and asks, "Is this good evidence?" Two ideas answer most of that question, and students mix them up constantly. This guide separates them for good.

Internal validity: did the cause really cause it?

A study has high internal validity when the change in the dependent variable can confidently be attributed to the independent variable rather than to something else. The "something else" is a confounding variable — anything that varies alongside your manipulation.

Three tools protect internal validity:

1. Random assignment. Assigning participants to conditions by chance spreads pre-existing differences evenly across groups. This is the single strongest protection, and it is what separates an experiment from a correlational study.
2. Control of procedures. Same instructions, same room, same time of day, same measures.
3. Blinding. When participants do not know their condition (single-blind) or neither participants nor researchers do (double-blind), expectation effects shrink.

Common threats worth naming on an exam: selection bias, attrition (people dropping out unevenly), history effects (something happens in the world mid-study), maturation, and demand characteristics.

External validity: would it happen anywhere else?

A study has high external validity when the result generalizes to other people, settings, and time periods. A tightly controlled experiment can be flawless internally and still tell you little about real life.

Three questions test it:

• Who was studied? A sample of undergraduates in one country is not humanity.
• Where was it studied? A quiet lab cubicle is not a classroom, a hospital, or a phone-in-hand commute.
• What exactly was measured? Reaction time on a screen is not the same construct as "attention at work."

The trade-off nobody escapes

Tighten control and internal validity rises while external validity often falls, because the setting becomes less like ordinary life. Loosen control to study behavior in the wild and you gain realism while losing certainty about cause.

Good research programs solve this across studies rather than inside one: an experiment establishes cause, a field study checks whether it survives outside. When you evaluate a single paper, the right verdict is usually "strong on one, limited on the other," not "good" or "bad."

A four-line checklist for any study

1. What is the IV and the DV, in the study's own words?
2. Was assignment random? If not, what could differ between groups?
3. Who was in the sample, and who is missing?
4. Is the measure the real-world behavior, or a stand-in for it?

Answer those four lines and you have written the core of a research-methods free-response answer — and you will read news coverage of psychology studies very differently.