Output figures for a growing research system are among the most cited and least interpretable numbers available.
They combine real growth, measurement change and coverage shifts into a single figure that cannot separate them.
What an output figure combines
- Genuine increase in research activity.
- Increase in the number of researchers.
- Change in which journals are indexed — measurement change, not research change.
- Change in publication practice, including unit size.
- Change in collaboration patterns, which inflates whole-counted totals.
- Reporting the total without the decomposition tells the reader very little.
Indicators that describe the system better
- Number of active researchers and its trend.
- Retention — how many remain active five years after training.
- Domestic first-authorship share in international work.
- Whether collaborations persist beyond single papers.
- Supervision capacity — who can train the next cohort.
- These are harder to measure and more informative.
The coverage problem
- Local-language publication is systematically underindexed.
- Applied and policy-facing work may be invisible to publication-based measures.
- New journals take years to be indexed regardless of quality.
- Undercounting is worst in the systems most in need of measurement.
- Any figure reported without this caveat understates what it measures.
Structural features worth understanding
- Where research is concentrated — institutions, regions, fields.
- How research is funded and how stable that funding is.
- Whether the training pipeline produces enough supervisors.
- What infrastructure exists and who can access it.
- These determine the trajectory more than current output does.
Reading comparisons carefully
- Compare with systems of similar size and stage, not with global leaders.
- Normalise for population or researcher numbers.
- Prefer trends over levels.
- Ask whether indexing coverage differs between the systems compared — it usually does.
What good reporting includes
- Data source and extraction date.
- Counting rule.
- Coverage limitations, stated plainly.
- Decomposition of growth where possible.
- Enough detail for someone to reproduce the figure.
One thing worth remembering
Before drawing any conclusion from a research output figure, ask whether the set of indexed journals changed over the period measured.
For emerging systems this frequently accounts for a substantial share of apparent growth — and it is the explanation least often checked, because checking it turns a good headline into a measurement artefact.
Câu hỏi thường gặp
What does an output figure combine?
Genuine increase in activity, more researchers, changes in which journals are indexed, changes in publication practice, and changed collaboration patterns — indistinguishable in a single total.
Which indicators describe a system better?
Number of active researchers, retention five years after training, domestic first-authorship share, whether collaborations persist, and supervision capacity.
What is the coverage problem?
Local-language publication is systematically underindexed, applied work may be invisible, and new journals take years to be indexed — with undercounting worst where measurement is most needed.
How should comparisons be read?
Compare with systems of similar size and stage, normalise for population or researcher numbers, prefer trends over levels, and ask whether indexing coverage differs.
What should be checked before any conclusion?
Whether the set of indexed journals changed over the period measured — for emerging systems this frequently accounts for much of apparent growth.
