Every score has a wobble
No assessment measures capability perfectly. Take the same test twice with nothing changed in between and your score will still move a little — mood, the specific questions, familiarity with the format. Psychometricians call the typical size of that wobble the standard error of measurement.
From it you can work out the smallest detectable change: the difference between two scores that is bigger than error alone would usually produce. Anything smaller is not evidence of growth, however much we would like it to be.
Why this matters for development plans
Many development tools show a before-and-after chart and celebrate any upward tick. If the tick is within measurement error, that celebration is selling noise — and people eventually notice when "growth" appears regardless of what they did.
Honest reporting has three outcomes:
| Result | What it means |
|---|---|
| Improved | The change is larger than the detectable threshold for this instrument and interval |
| Held steady | The change is within measurement error — no credible evidence either way |
| Declined | The drop is larger than the threshold |
"Held steady" is a real result. Over 90 days, it is often the most likely one.
Short cycles are for doing; longer arcs are for measuring
Behavioural capabilities — negotiation, delegation, strategic thinking, match temperament — move slowly. A 90-day cycle is right for finishing a focused plan, and poor for measuring it. Over six to twelve months or longer, smaller real gains can clear the threshold.
That is why a good plan separates the two rhythms: short cycles to do the work, a longer horizon to see whether it changed anything.
Practical rules
- Know whether a number was measured or self-rated. They are not equivalent and should never be shown the same way.
- Do not reassess too often. Repeated testing also adds practice effects.
- Pair scores with outcomes. A signed renewal, a ranking, a certification or a promotion is evidence too.
- Expect "held steady" and do not be discouraged by it. Keep doing the work.
Jobulary applies interval-aware thresholds and says "held steady" when a change is within error. See how it works for the details.
Frequently asked questions
Why did my competency score only change a little?
Behavioural capability usually moves slowly, and every test has measurement error. A small change over a few months is often within that error, which is why it is reported as 'held steady' rather than growth.
What is the minimal detectable change?
The smallest difference between two measurements that is larger than what measurement error alone would typically produce. It depends on the instrument's reliability.
How often should I reassess?
Less often than you might think. Reassessing every few weeks mostly measures noise and practice effects; six to twelve months gives real change a chance to show.
Sources
- The applicability of standard error of measurement and minimal detectable change (Frontiers in Human Neuroscience, 2018)
- CASRAI — Test-retest reliability: the retest interval
Topics: Measurement · Individual Development Plan