Direct comparison
Validity vs reliability — what is the difference?
Validity vs reliability explained: the difference is measuring the right thing (accuracy) versus measuring consistently (repeatability).
Side-by-side comparison
| Dimension | Validity | Reliability |
|---|---|---|
| What it is | The extent to which a measure captures what it is intended to capture. | The extent to which a measure gives consistent, repeatable results. |
| Core question | Are we measuring the right thing? | Are we measuring consistently? |
| Focus | Accuracy and truthfulness of the inference. | Stability and repeatability of the measurement. |
| Dartboard analogy | Hitting the bullseye — on target. | Hitting the same spot each time — tightly grouped. |
| Common types | Construct, content, criterion and internal/external validity. | Test–retest, inter-rater and internal consistency. |
| How it is assessed | Against theory, expert judgement or an external criterion. | By repeated measurement and correlation or agreement statistics. |
| Can you have one without the other? | A measure cannot be fully valid without being reliable. | A measure can be reliable yet invalid — consistently measuring the wrong thing. |
| Effect of error | Systematic error (bias) threatens validity. | Random error (noise) threatens reliability. |
| Priority in design | Establishes that the conclusion is meaningful. | A necessary precondition that makes validity assessable. |
Common questions
FAQ
Can a measure be reliable but not valid?+
Yes — this is the classic distinction. A bathroom scale that always reads two kilograms heavy is perfectly reliable, because it is consistent, but invalid, because it does not report true weight. Reliability concerns consistency; validity concerns accuracy. Consistency alone does not guarantee you are measuring the right thing.
Does validity imply reliability?+
Generally yes. If a measure is valid — accurately capturing the intended construct — it must also produce reasonably consistent results, because wildly erratic readings could not be accurate. So reliability is treated as a necessary, but not sufficient, condition for validity.
Which should I prioritise in research design?+
You need both, but they are assessed in order: establish reliability first, because an unreliable measure cannot be valid, then demonstrate validity to show the measure captures the right construct. Random error undermines reliability, while systematic bias undermines validity.
