A systematic literature review evaluating a decade of research into the United Nations Sustainable Development Goals has revealed that quantitative methodologies used to assess interlinkages between targets remain fragmented, conceptually inconsistent, and prone to significant analytical distortions. Published in the journal Energy, Sustainability and Society, the investigation conducted by researchers Rega Sota and Sandra Venghaus evaluated peer-reviewed quantitative studies exploring interactions across the 2030 Agenda. Their findings indicate that despite a rapid expansion of analytical papers, the scientific community has not converged on unified, technically sound methods. Instead, widespread reliance on inconsistent terminology, unvetted correlation metrics, and unverified causal claims continues to produce incomparable operational findings, undermining the potential of the global goals to provide robust decision support for policy interventions.
The 2030 Agenda, unanimously approved by the United Nations in 2015 to guide humanity toward sustainable development, defines sustainability through 17 interconnected goals, 169 specific targets, and 231 operational indicators. While the official framework presents these goals as indivisible and explicitly criticizes policy formulation in administrative silos, the United Nations avoids prescribing any standardized methodology for interlinkage assessment. Consequently, researchers examining whether progress in one area accelerates or undermines advancement in another have introduced a broad array of disparate frameworks. As the review notes, the goals themselves vary significantly in composition, interpretation, and degree of quantification, making systematic tracking challenging. Without officially recognized evaluation guidelines, analytical efforts have proliferated into fundamentally divergent streams, creating what previous critiques characterized as a confusing sea of tools that leaves the framework’s policy leverage substantially weakened.
To rigorously map the scientific landscape, the authors carried out a structured literature search across Web of Science, Scopus, and Science Direct covering publications from 2015 to 2023. Using paired search strings such as SDG and interlinkages or SDG and interactions, the team compiled an initial corpus of 8,008 entries, which was reduced to 6,894 records after eliminating duplicates. Subsequent text screening to confirm the specific meaning of the SDG acronym further refined the collection to 6,264 papers. Because the study focused strictly on quantitative and methodological evaluations rather than descriptive policy overviews, qualitative papers were eliminated based on title and abstract screening, leaving 616 candidates. These were evaluated against three explicit hierarchical criteria assessing whether they introduced a quantitative method, offered a conceptualization with clear quantification potential, or addressed technical gaps affecting operationalization. This filtration left 140 core papers, and applying literature snowballing uncovered 13 additional records, yielding a final analytical corpus of 153 studies that were evaluated in full.
The earliest efforts to quantify interactions grew directly out of qualitative frameworks, most notably textual analysis reports and an integer scoring scale extending from minus three to plus three introduced by Nilsson and colleagues in 2016. In that scale, interactions range from cancelling effects at negative three to indivisible synergies at positive three. Although this seven-point scoring rubric appeared in 27 of the reviewed studies, the authors point out that its reliance on context-specific examples frequently blurs distinctions between numerical increments and hampers consistency. Because the scale primarily functions as an adjunct to subjective expert assessments, different expert panels regularly assign conflicting scores to identical target interactions. While alternative scoring methods have attempted to introduce resource-specific scoring criteria—such as assessing mutual infrastructure benefits versus competition over limited land or water—subjective matrix evaluations continue to suffer from fundamental reproducibility limits.
Over time, qualitative scoring was widely combined with or supplanted by network analysis, which has been deployed across the literature for three primary purposes: identifying priority targets, visualizing interactions extracted from textual discourse, and asserting causal relationships among targets. In prioritization studies, such as investigations mapping interactions across Sweden’s national targets, interaction matrices are converted into network graphs to evaluate metrics like node centrality. However, the review found that concepts like centrality are assigned conflicting definitions across studies, while connections are frequently extended beyond direct linkages to vaguely formulated indirect pathways without mathematical standardization. Similarly, studies using text-mining metrics such as cosine similarity to construct discourse networks from UN reports or academic literature define and deploy the metric in conflicting manners, severely impeding cross-study comparability.
A particularly acute methodological pitfall emerged in studies attempting to derive causal inferences from network structures. For example, some published investigations applied the transitive reduction algorithm to directed graphs derived from Granger-causality testing between SDG indicators, describing the process as the removal of redundant connections to compress the graph. The authors explain that because Granger causality is not mathematically a transitive relation, pruning reachability paths using information compression algorithms designed for graph simplification introduces severe logical fallacies. Using graph compression routines for causal discovery distorts the underlying empirical relationships, generating unintended analytical conclusions and falsely implying established causal pathways where none have been verified through sound econometric identification.
Correlation analysis represents the most common yet most criticized quantitative approach across the reviewed literature, appearing in 36 separate studies. Researchers have applied Pearson correlation, Spearman rank correlation, multiple factor coefficients, and autoregressive models to indicator time series, target pairs, or aggregated indices. A central vulnerability of this literature is the mechanical mapping of correlation signs directly onto the conceptual labels of synergy and trade-off. In the majority of studies, any positive correlation coefficient is automatically classified as a synergy, while negative coefficients are treated as trade-offs. The authors emphasize that this practice introduces major errors because it fails to distinguish between stimulant indicators, where an increase represents progress, and destimulant indicators, where progress requires a numerical decline, as well as moderate indicators that are desirable only within bounded intervals.
Without careful indicator normalization, a positive correlation can simply describe a scenario where an improving stimulant indicator moves in tandem with an increasing destimulant indicator, which represents a clear win-lose dynamic rather than a mutual benefit. Even when datasets undergo normalization or sign reversals, a positive correlation routinely captures two simultaneously deteriorating indicators. Labeling a shared decline as a synergy represents an analytical fallacy that severely distorts policy conclusions. Furthermore, influential global correlation frameworks that establish arbitrary significance thresholds, such as a correlation coefficient cutoff of 0.6 to designate non-interlinkages, fail to eliminate underlying conceptual confusions. These studies frequently run into spurious correlations when a limited pool of indicators is mapped across all 169 targets, and they regularly disagree with expert elicitations because short-term statistical snapshots fail to reflect long-term systemic dynamics.
In response to data scarcity during the early years of the 2030 Agenda, econometric techniques and complex simulation models emerged more slowly but have gained traction as longer time series became available. Econometric investigations, featured in 20 surveyed papers, encompass linear regressions, multiple regressions, difference-in-differences designs, and generalized method of moments estimators examining sectoral intersections like the water-energy-food nexus. Although econometric methods impose more rigorous statistical checks, address endogeneity, and test causal directionality, their technical complexity has ironically limited their uptake by policymakers, who frequently express a preference for easily interpretable scoring rubrics over technically demanding econometric models. Meanwhile, 14 studies deployed simulation frameworks, including computable general equilibrium and integrated assessment models. While capable of multi-sectoral evaluations, these models rely heavily on restrictive assumptions, such as universal market clearing or predetermined cost competition, that restrict their empirical validity.
The review also documented emerging conceptual approaches designed to overcome aggregate data limitations by linking goals through concrete technological and physical mechanisms. Bottom-up estimation techniques have mapped interlinkages by calculating the direct resource and energy requirements of basic human needs, while other recent frameworks use energy technologies as explicit entry points, mapping technological supply chains directly to SDG targets through industrial classification systems. In parallel, systems studies have applied causal loop diagrams to conceptualize systemic feedbacks and archetypal behaviors. However, the authors observe that findings derived at global aggregation levels routinely fail to hold when scaled down to regional or local jurisdictions, reflecting substantial contextual heterogeneity across income levels and geographic settings.
To overcome widespread inconsistencies and establish scientific comparability, the authors issue several concrete recommendations for future research and institutional governance. Methodologically, researchers must abandon the simplistic and misleading synergy and trade-off dichotomy, which conceals critical empirical dynamics and creates an unjustified bias toward positive associations. Instead, studies should adopt nuanced multi-state configurations that explicitly identify causal and affected targets alongside their positive or negative trajectories. Furthermore, authors strongly caution against extensive data imputation, noting that imputing missing observations often relies on unrealistic randomness assumptions and creates artificial data smoothing that degrades diagnostic precision; where unavoidable, missing observations should remain well below established empirical thresholds such as five percent.
Finally, Sota and Venghaus urge the United Nations Statistical Commission to establish a coordinated benchmarking initiative that assesses identical, standardized databases across multiple competing quantitative methodologies at the indicator, target, and goal levels. Such inter-methodological testing would expose specious robustness—instances where flawed methods arrive at coincidentally similar conclusions—and clarify which analytical frameworks generate reliable empirical insights. The authors also recommend that regional and sectoral studies, particularly those focused on the critical water-energy-food nexus, adopt systematic methodological triangulation by running multiple distinct analytical models over the same datasets. By grounding analytical frameworks in rigorous causality, theoretical consistency, and complete reporting of data treatments, the scientific community can transform SDG interlinkage research from an abstract discursive exercise into an actionable foundation for global policy.
Subject of Research: Systematic literature review evaluating quantitative methodologies used to assess interlinkages among the United Nations Sustainable Development Goals
Article Title: Talking past each other: a review of the quantitative study of interlinkages between the sustainable development goals
Article References: Sota, R., & Venghaus, S. (2026). Talking past each other: a review of the quantitative study of interlinkages between the sustainable development goals. Energy, Sustainability and Society, 16(1), Article 14. https://doi.org/10.1186/s13705-026-00572-0
Image Credits: AI Generated
DOI: 10.1186/s13705-026-00572-0
Keywords: Sustainable Development Goals, United Nations Agenda 2030, interlinkage methodology, systematic literature review, correlation analysis, network analysis, econometric modeling, integrated assessment models, computable general equilibrium, synergies and trade-offs, data imputation, causality, water-energy-food nexus, indicator normalization, policy decision support, scientific comparability
News Source: Denise Maddox. (November 13, 2026). Systematic Review Finds Conflicting Quantitative Frameworks Hamper Global Sustainable Development Goals. Scienmag.



