National retention extends conventional incumbency measurement
Measuring power beyond the scheduled term captures retention that conventional next-election outcomes miss when elections are delayed or cancelled.
↳ Introduction; Section 4.1
Abstract not available — view at DOI
Key Claims
Primary findingTentative
On average, national election winners benefit from an incumbency advantage, but this effect is short-lived and differs markedly across contexts
Fragile — the claim's weakest requirement is that the effect exists at all. Based on how the study was designed and conducted, not on statistics extracted from the paper. Experimental. Every point below is traceable to a quoted passage, but this forecast has not yet been validated against known replication outcomes.
close-election identification is imperfect across regimes, while the marginal average estimate and later non-significance leave duration uncertain
Abstract; Sections 5.2–5.3
SecondaryTentative
In established democracies, the national incumbency advantage reflects gains in the subsequent electoral performance of election winners. In less democratic regimes, it mainly stems from manipulation of the fairness and the timing of elections
failed assignment checks in affected settings and an accounting decomposition do not identify manipulation as the causal mechanism
Abstract; Section 6
Derived from the full evaluation — not a separate score.
Strengths · Limitations
Strengths
Identification checks explicitly probe electoral manipulation
Pre-election placebo outcomes directly examine whether marginal winners were already more likely to hold power. Bandwidth, fixed-effect, and sample-restriction exercises further probe the results.
↳ Sections 5.1 and 5.3; Figure 4; Appendix E
Literature synthesis frames the national comparison
The synthesis of 94 estimates from 58 studies provides a structured comparison with national results and connects their heterogeneity to accountability and institutional context.
↳ Section 2; Sections 6.3–6.4
Limitations
Failed placebos challenge central causal interpretations
Autocracies and elections classified as not free and fair show pre-election discontinuities consistent with manipulation. Reporting these groups separately does not establish quasi-random assignment within them.
↳ Section 5.1; Figure 4; Sections 6–7
Average effect and duration remain uncertain
The headline party-level post-term estimate is 10 percentage points with p = 0.065. Later non-significant estimates do not establish that the effect has disappeared, and overlapping subgroup analyses add uncertainty.
↳ Sections 5.2–5.3; Figures 5–6
Mechanism language exceeds the decomposition's reach
The decomposition separates retention after fair elections, unfair elections, and no subsequent election. That accounting split does not by itself establish that manipulation causes the observed advantage, despite categorical wording in the abstract and conclusion.
↳ Abstract; Sections 6.2–6.3 and 7
Nabu Commentary
The national post-term retention estimand and global election dataset provide the paper's clearest advance. Sections 4–5 document a detailed close-election implementation, but Section 5.1 also reports placebo failures in autocracies and unfair elections that challenge causal claims for those settings. The aggregate 10-percentage-point estimate has p = 0.065, and later non-significance leaves the duration claim less certain than the headline suggests. The decomposition in Section 6.3 organizes retention pathways, but it does not justify the categorical manipulation explanation in the abstract and conclusion.
Check another paper
Author Note / Field Feedback
Nabu’s assessment, alongside the field’s view.
Are you an author of this paper?
Dimension breakdown
Study Quality
Sound3.2
Confidence mediumContribution
4.2
The paper adds a global national-election dataset and a post-term retention estimand that accommodates postponed or cancelled elections. The distinction from subnational reelection effects is substantive, although causal uncertainty in lower-democracy settings limits the breadth of the advance.
Introduction; Sections 2–4
Methodological Rigour
2.8
The RD implementation includes robust nonparametric inference, election-level clustering, placebo outcomes, and specification checks. Causal interpretations persist in subsamples with failed placebo checks, producing the binding ceiling despite the otherwise detailed execution.
Sections 4–5.3 and 6.2–6.3; Appendix E
Reporting
3.6
The estimands, worked examples, and progression from results to decomposition make the argument followable. Central mechanism claims in the abstract and conclusion are more categorical than the identification checks and decomposition justify.
Abstract; Sections 4, 5.1, 6.2–6.3, and 7
Positioning
2.8
The paper connects the national results with subnational RD evidence, accountability theory, and electoral-manipulation research. Its conclusion does not carry through the acknowledged identification limitation for autocratic and unfair-election settings.
Introduction; Sections 2, 5.1, 6, and 7
Reliability
Caveats3 of 4 checks
Results checkClean
The reported numbers and statistics are consistent with the stated results.
Reporting and transparencyNoted
Tables E.9 and E.10 repeat the same descriptive title despite reporting different sample restrictions. Their notes distinguish the samples, so this reporting error does not change the conclusions.
Conflict of interest: declared
Reference IntegrityNot checked
Reference integrity check could not run: references could not be reliably parsed from the manuscript; verification skipped rather than risk false verdicts.
0 references in manuscript
Post-Publication RecordClean
No retraction or expression of concern found in Crossref (includes Retraction Watch).
Last checked: 20 Oct 2026
Sources: Crossref (incl. Retraction Watch) ✓
Readiness
Where this paper’s evidence sits on the path from initial observation to real-world use.
Mechanistic
Use of the national incumbency measure as a democratic-quality indicator is proposed rather than operationally tested. Practical translation therefore requires validation and an application protocol beyond the reported analysis.
Section 7, final paragraph
AI-generated, human-governed. Something look off? Contact us to request a review.