Dr Hussein Sharara

Mohamad Jawad Sharara

Abstract: Lebanon has an internationally visible higher-education research system, but public reporting is fragmented across bibliographic databases, institutional reports, journal portals, and ranking systems. This study develops an auditable quantitative account of research publishing in Lebanese universities during 2023–2026 while explicitly separating country-level, university-level, and journal-level evidence.

Methods: A multi-source secondary-data audit was conducted using SCImago Institutions Rankings 2026, SCImago Journal & Country Rank country-level data for 2023 and 2024, official research reports of the American University of Beirut (AUB), Beirut Arab University (BAU), and Université Saint-Joseph de Beyrouth (USJ), the SCImago journal dataset, and peer-reviewed bibliometric literature on research-integrity-sensitive indicators. Descriptive statistics, year-over-year growth rates, reporting-discrepancy diagnostics, and evidence-harmonization rules were applied. No missing annual university counts were estimated or back-filled.

Results: SCImago 2026 lists eight Lebanese universities in the higher-education sector: four with a best quartile of Q1, three Q2, and one Q3. Lebanon's country-level indexed document count increased from 6,484 in 2023 to 7,618 in 2024, a 17.5% increase in the database snapshot. AUB reported 1,065 original/review journal articles in both 2023 and 2024 and reported that 63% of 2024 original/review articles were in top-quartile journals. Its 2024 output corresponded to 1.78 original/review journal articles per full-time professorial faculty member, using the institutional faculty count as denominator. USJ's annual chart reports 544 publications in 2023 and 546 in 2024, but a separate summary reports 1,016 publications for the two-year period, whereas the chart sums to 1,090; this 74-publication discrepancy prevents construction of a single internally consistent USJ time series. BAU reports 355 unique publications in academic year 2023–2024 after accounting for 10 joint publications across faculties; the published faculty rows sum to 365, demonstrating an internal de-duplication issue that must be respected in analysis. A snapshot of Lebanese-publisher journals in SCImago contains eight titles, one Q1 and seven Q4, but journal activity in Scopus is not equivalent to continued journal publication and must be tracked by coverage status.

Conclusions: The strongest evidence supports a stratified and heterogeneous Lebanese research system, but current public reporting is not sufficiently harmonized to justify a definitive 2023–2026 university-wide publication census. A publication-ready national study therefore requires a frozen Scopus/SciVal export at institution-year-document level, affiliation-ID based counting, year- and subject-category-specific quartiles, explicit de-duplication rules, and separate treatment of 2026 as provisional until indexing is complete. The present study provides the validated quantitative baseline, analytic framework, and reproducibility protocol needed for that final census.

Keywords: Lebanese universities; bibliometrics; research output; Scopus; SCImago; journal quartiles; research integrity; scientific publishing; higher education; Lebanon

Research Highlights

· Eight Lebanese universities are listed in the SCImago Institutions Rankings 2026 higher-education sector: four Q1, three Q2, and one Q3.

· Country-level indexed output rose from 6,484 documents in 2023 to 7,618 in 2024 (+17.5%), but this is not a university-level total.

·  AUB provides a comparatively consistent institutional series, with 1,065 original/review journal articles in both 2023 and 2024 and 63% of 2024 articles reported in top-quartile journals.

·  Published USJ and BAU institutional reports reveal measurable internal counting-definition problems; these are methodological findings rather than editorial errors to be silently corrected.

·  The central evidence gap is the absence of a single, reproducible, institution-year Scopus export covering all eight universities for 2023–2026.

1. Introduction

1.1. Background and rationale

    Research performance in higher education is increasingly described through publication counts, citation impact, journal position, international collaboration, and institutional rankings. These indicators are useful only when their unit of analysis, coverage, counting rules, and data vintage are explicit. A national publication count, for example, cannot be substituted for a university publication count, and a composite institutional ranking cannot be interpreted as a direct measure of publication quality.

    The Lebanese case is particularly suitable for a structured audit because three evidence layers coexist: international bibliographic databases, institution-reported research statistics, and a local journal ecosystem that is only partially represented in global indexes. The supplied research draft correctly identifies this fragmentation and explicitly notes that the annual 2023–2026 university counts remained incomplete pending a Scopus or SciVal export. That methodological restraint is retained here rather than replaced by estimates. [Author working draft, 2026].

    SCImago Institutions Rankings (SIR) provides an institutional context rather than a simple output count. In the 2026 higher-education ranking for Lebanon, SIR lists eight institutions and combines Research (50%), Innovation (30%), and Social (20%) dimensions. The Research component itself incorporates output, normalized impact, excellence, Q1 publications, international collaboration, open access, and other indicators. Consequently, a university being classified in a given SIR quartile must not be described as evidence that all of its publications fall within that quartile (SCImago, 2026).

1.2. Research gap

    Existing public sources provide substantial fragments of the answer but do not, by themselves, provide a harmonized 2023–2026 dataset for all Lebanese universities. Country-level SCImago data describe Lebanon as a whole; university reports use institution-specific definitions; Nature Index covers a selected set of journals rather than total scholarly output; and local journal information varies in indexing status and reporting practice. The scientific gap is therefore not the absence of numbers, but the absence of a single counting framework under which the numbers can be validly compared.

1.3. Study objective

    The objective is to construct a publication-ready quantitative audit that (a) defines the Lebanese university research universe, (b) validates the strongest publicly documented output measures for 2023–2024 and the 2026 institutional context, (c) quantifies reporting discrepancies rather than concealing them, (d) separates journal-level indexing from institutional performance, (e) introduces a reproducible Scopus extraction protocol for completing the 2023–2026 census, and (f) frames research-integrity-sensitive bibliometric indicators as screening signals rather than findings of misconduct.

2. Research Questions and Hypotheses

2.1. Research questions

·  RQ1. What is the documented level and direction of research publishing associated with Lebanese universities during 2023–2026?

·  RQ2. How is the research output distributed across internationally indexed journals, journal quartiles, locally published journals, and other scholarly output types?

·  RQ3. How consistent are institutional reporting definitions and public bibliometric sources, and what discrepancies affect cross-university comparability?

·  RQ4. Which bibliometric indicators are suitable for assessing research performance, and which indicators should be treated only as integrity-sensitive screening signals?

2.2. Pre-specified hypotheses for the complete Scopus dataset

The following hypotheses are intentionally stated as testable hypotheses, not results. They should be tested only after a frozen institution-year Scopus export is obtained.

·  H1. The 2026 January–September indexed output of Lebanese universities does not fall materially below the institution-adjusted 2023–2025 baseline.

·  H2. The share of publications in Q1/Q2 journals differs across Lebanese universities after controlling for field of research.

·  H3. Local non-indexed publication shares are higher in arts, humanities, social sciences, and law than in medicine, science, and engineering.

·  H4. Institutional publication volume and journal-position indicators are related but not interchangeable dimensions of research performance.

·  H5. International collaboration is associated with higher field-normalized citation impact after controlling for research field and publication year.

·  H6. Apparent rapid growth in publication volume is not, by itself, sufficient evidence of improved research quality or research misconduct.

3. Methods

3.1. Study design

    This is a multi-source quantitative bibliometric audit using secondary data. The design follows an evidence hierarchy: (1) direct bibliographic database records and frozen exports are the preferred quantitative basis; (2) official university research reports are used for institution-specific validated counts; (3) ranking and curated indexes provide contextual indicators; and (4) peer-reviewed bibliometric studies provide methodological and integrity-related interpretation. Where two sources use incompatible definitions, both are retained and the incompatibility is reported.

3.2. Analytical universe

    The institutional universe for the 2026 higher-education comparison is the eight Lebanese universities listed by SCImago Institutions Rankings 2026: AUB, LAU, Lebanese University, BAU, Université Saint-Joseph, Lebanese International University, Holy Spirit University of Kaslik, and Balamand University (SCImago, 2026). This does not imply that these are the only institutions that generate research in Lebanon; it defines the universe for the specific SIR-based university comparison used here.

Table 1. Lebanese universities listed in SCImago Institutions Rankings 2026, higher-education sector. Source: SCImago (2026).

3.3. Data sources and provenance

Table 2. Evidence hierarchy and provenance. “Verified” means the numerical statement was traceable to a current or institution-published source; it does not imply all sources use identical definitions.

3.4. Counting rules

    Four counting rules are required for a reproducible study. First, country-level totals are not used as substitutes for university-level totals. Second, university output is counted on an institution-by-document basis using institutional affiliation identifiers; a document may legitimately contribute to more than one university when multiple affiliations are present, so the sum of university counts should not be interpreted as a national total. Third, journal quartile is recorded by publication year and subject category rather than as a single timeless Q value. Fourth, 2026 is coded as provisional unless the dataset is frozen after complete indexing.

3.5. Scopus extraction protocol for the final census

    The supplied draft proposed Scopus advanced-search queries using AFFILCOUNTRY and PUBYEAR and recommended an export to CSV/SciVal. For a publication-grade dataset, the protocol should be tightened as follows:

·  Use the Scopus Affiliation ID(s) for each institution as the primary identifier. Do not rely on free-text university names alone.

·  Export EID/DOI, title, publication year, document type, source title, ISSN, affiliation IDs, author list, country affiliations, subject areas (ASJC), citations, and international-collaboration fields where available.

·  Record journal quartile by SJR/SciSCImago or the selected journal metric, linked to the specific publication year and ASJC category.

·  Retain separate document-type categories: article, review, conference paper, book chapter, and other types. Report the all-document total separately from the article/review total.

·  For aggregate Lebanese university figures, do not simply sum all university counts because the same multi-university paper may be counted in more than one institution. Use the Lebanon country filter for the country total and university filters for institutional totals.

·  For 2026, store a data-freeze date and restrict the analysis window explicitly (for example, 1 January–30 September 2026) until a complete annual 2026 snapshot is available.

3.6. Statistical analysis plan

    The completed institution-year dataset should first be described using counts, percentages, medians, interquartile ranges, and year-over-year growth. For inferential analysis, publication counts may be modeled with a negative-binomial generalized linear mixed model when over-dispersion is present, with institution and year as fixed effects and faculty count or research-active faculty as an offset where reliable denominators exist. Journal-position outcomes can be modeled as binary Q1/Q2 versus Q3/Q4 or as multinomial categories, with research field and year as covariates. International collaboration can be analyzed as an exposure variable, while field-normalized citation impact is preferred over raw citations for cross-disciplinary comparisons. Given n=8 institutions, cross-university correlation analyses are exploratory and should not be presented as definitive causal evidence.

3.7. Data-quality controls

Table 3. Reproducibility and quality-control rules proposed for the final Scopus/SciVal extraction.

4. Results

4.1. National indexed research output: a country-level benchmark

    SCImago country data report 6,484 indexed documents for Lebanon in 2023 and 7,618 in 2024. This is an increase of 17.5%. The change is useful as a national benchmark, but it should not be interpreted as the growth rate of Lebanese universities because the country-level denominator includes the wider Lebanese research ecosystem.

Figure 1. Lebanon: indexed document output in SCImago/Scopus country data, 2023–2024. Source: SCImago Journal & Country Rank.

    The supplied draft also reported a different cumulative total for 1996–2025 and a lower OpenAlex/Rankless count. These differences should be treated as database-coverage differences rather than as evidence that one database is “wrong.” A publication-ready manuscript should always state the database, query, data vintage, and counting unit alongside each total.

4.2. Institutional position in SCImago 2026

    SCImago 2026 lists eight Lebanese universities in its higher-education sector. The distribution is four Q1 institutions, three Q2 institutions, and one Q3 institution on the SIR “best quartile” field. The ranking is a composite institutional benchmark: Research accounts for 50%, Innovation for 30%, and Social for 20%. Within Research, SIR includes both volume and quality/impact dimensions such as normalized impact, Q1 publications, scientific leadership, international collaboration, and open access. Therefore, the SIR quartile is an institutional diagnostic rather than a direct count of Q1 publications.

Figure 2. Lebanese universities listed in SCImago Institutions Rankings 2026. Lower numerical rank denotes a higher position in the published global ranking; the figure is descriptive and should not be read as a pure publication-quality index.

4.3. AUB: the strongest publicly documented continuous institutional series

    AUB’s Research Facts and Figures series reports original and review journal articles for each year from 2015 through 2024. The count is 1,065 in 2023 and 1,065 in 2024, producing a year-over-year change of 0.0%. The 2024 report also states that 63% of original/review articles were published in top-quartile journals. The same source reports 598 full-time professorial faculty members in 2024, which corresponds to 1.78 original/review journal articles per professorial faculty member when the institutional count is used as numerator and denominator.

Figure 3. AUB: original and review journal articles, 2015–2024. Source: AUB Research Facts and Figures 2024.

    The long series also shows that AUB’s 2024 article count was 26.0% lower than its 2020 peak of 1,440, illustrating why a single year should not be used to infer a long-term trend. This historical observation is descriptive only; it does not identify the causes of the change.

4.4. USJ: publication volume and a measurable reporting inconsistency

    USJ institutional reporting provides a useful example of why definitions and extraction dates matter. The 2023–2024 annual chart reports 544 publications in 2023 and 546 in 2024, while another summary in the same reporting context states 1,016 publications over the two-year period. The chart sums to 1,090, creating a 74-publication difference, equivalent to 7.3% of the reported two-year total. This discrepancy should not be resolved by choosing one number without documentation; it should be explicitly marked for reconciliation against the underlying Scopus export. The report also states 60% international collaboration for the period.

Figure 4. USJ: annual publication counts reported in the institutional chart, 2023–2024. The separate two-year summary does not reconcile to the chart total.

4.5. BAU: mixed output types and the importance of de-duplication

     BAU’s research report for academic year 2023–2024 reports 355 university-level publications after accounting for 10 joint publications across faculties. The faculty rows, however, sum to 365 because those joint publications occur in more than one faculty row. The 10-publication difference is therefore a visible example of how faculty-level whole counting can exceed university-level unique counting. For that reason, the faculty-level Q1/Q2/Q3/Q4 occurrences in the report should not be summed and presented as an exact university-wide indexed total without the underlying publication-level de-duplication file.

Table 4. BAU output-definition audit based on the institution’s 2023–2024 research report. The 245 figure is a row-occurrence count before joint-paper de-duplication and is therefore not a unique university count.

4.6. The Lebanese journal ecosystem

    The supplied draft identified eight Lebanese-publisher journals in its SCImago journal snapshot, with one Q1 and seven Q4 titles. This is useful as a journal-ecosystem map but requires two methodological distinctions. First, SCImago’s quartile is subject- and year-specific; a journal can have more than one subject category. Second, a journal can remain visible in historical SJR records after its active Scopus coverage has ended. Journal Medical Libanais is an example of why active coverage and journal activity should be coded separately. The correct unit is therefore “journal status in the data vintage,” not merely “appears on a ranking page.”

Table 5. Lebanese-publisher journal snapshot reproduced from the supplied draft’s SCImago-based table. Values are journal-level SJR data for the cited snapshot; active Scopus coverage must be checked separately before treating a title as currently indexed.

4.7. Integrity-sensitive bibliometric evidence

    A peer-reviewed study by Meho and Akl (2025) analyzed 14 universities selected from a wider group of institutions showing unusually rapid publication growth. Across the study group, total publications increased by 234% between 2019 and 2023, first-authorship rates fell by 23 percentage points, and the number of hyperprolific authors (defined in that study as authors publishing 40 or more journal articles in a calendar year) rose from 23 to 177. The authors interpret these patterns as reasons to scrutinize authorship and affiliation practices. Importantly, the study is not a census of Lebanese universities and does not establish that rapid growth itself constitutes misconduct. LAU was included in the study group, so the work is relevant as a methodological signal for Lebanese bibliometric auditing but not as a prevalence estimate for Lebanon.

    The appropriate analytical principle is therefore: publication volume, journal quartile, and institutional ranking are performance descriptors; anomalous authorship, affiliation, delisted-journal patterns, retractions, or other integrity-sensitive measures are screening variables that require document-level and institutional verification. This distinction is essential for responsible scholarly reporting.

5. Discussion

5.1. What the evidence supports

    The evidence supports four conclusions. First, Lebanon has a structurally heterogeneous university research system. Second, international bibliometric visibility is concentrated in a limited set of institutions and journals, while a broader local publication layer sits outside the same indexing systems. Third, university research reporting is not harmonized: AUB reports a relatively clean original/review journal series, BAU separates multiple output types and exposes joint-publication overlap, and USJ presents an internal numerical discrepancy that must be reconciled. Fourth, bibliometric integrity analysis must be kept analytically separate from ordinary performance measurement.

5.2. Why publication counts alone are insufficient

    A publication count measures activity, not necessarily impact, originality, societal relevance, reproducibility, or integrity. Conversely, a low indexed count does not establish low-quality scholarship when a field is underrepresented in international indexes or when a local peer-reviewed journal operates outside Scopus. For Lebanon, this is especially important for humanities, law, arts, Arabic-language scholarship, and locally focused applied research, where indexing coverage can be structurally different from medicine, science, and engineering.

5.3. Why quartile distributions need field normalization

    Journal quartiles are relative indicators within journal subject categories and data vintages. A Q1 result in one field is not directly equivalent to a Q1 result in another field when the underlying journal populations, citation densities, and indexing histories differ. The final Scopus dataset should therefore retain the ASJC category, year, journal metric, and quartile as separate variables. For citation impact, field-weighted measures are preferable to raw citations for inter-field comparisons.

5.4. The central methodological finding

    The most defensible scientific finding at this stage is not a league table of Lebanese universities. It is that public research statistics are sufficiently heterogeneous that a naïve cross-university comparison can create false differences. This is demonstrated empirically by the USJ arithmetic discrepancy, the BAU faculty-to-university de-duplication gap, the distinction between country and university counts, and the historical-versus-active indexing issue in journals. These are measurable properties of the evidence system itself.

6. Final Quantitative Model for the Complete 2023–2026 Census

Once a frozen Scopus/SciVal export is available, the manuscript should be updated with a four-year university panel. The following table specifies the final variables and the analysis expected for each.

Table 6. Variables and analytical outputs for the final institution-year Scopus/SciVal panel.

6.1. Recommended panel dataset structure

    The master dataset should be one row per publication per institution affiliation. Minimum identifiers: EID, DOI, title, year, document type, journal ISSN, journal publisher, ASJC code, quartile, affiliation ID, institution name, authors, country affiliations, citation count, international-collaboration flag, and research-field classification. A second de-duplicated table should contain one row per unique publication within the Lebanese national corpus. This two-table architecture prevents accidental mixing of institution-level and country-level counting.

6.2. Statistical tests

·  Trend analysis: annual counts and growth rates for each institution; use confidence intervals or dispersion estimates where document-level data allow.

·  University differences: compare Q1/Q2 share and field-normalized impact using regression models controlling for year and ASJC field.

·  Productivity: model document counts using a negative-binomial approach with a research-active faculty offset when denominators are comparable.

·  Collaboration-impact association: estimate the relationship between international collaboration and normalized citation impact, controlling for field and year.

·  Integrity screening: flag unusual patterns for verification rather than convert them into a binary judgment of misconduct.

7. Policy and Research-System Implications

    The quantitative audit suggests that Lebanon would benefit from a national research reporting standard rather than isolated institutional publication dashboards. A minimal common standard would define document types, calendar or academic-year boundaries, affiliation identifiers, de-duplication, journal quartiles by year and subject category, international collaboration, and a fixed annual data-freeze date. Such a standard would make university research performance more transparent while reducing the risk that different reporting definitions create artificial differences.

    For universities, the immediate practical priority is not merely increasing the number of indexed papers. It is building publication metadata governance: standardized author affiliations, ORCID adoption, DOI/EID reconciliation, journal-status monitoring, correction of affiliation errors, and explicit separation of journal articles from conference and book outputs. These measures improve both reporting accuracy and the reliability of external bibliometric indicators.

8. Limitations

·  A complete institution-year Scopus export for all eight universities was not publicly available to the author in a reproducible form at the study cut-off; therefore the paper does not claim a complete university-wide 2023–2026 census.

·  Institutional reports use different reporting periods and output definitions, so AUB calendar-year article counts, BAU academic-year mixed output, and USJ reporting cannot be pooled as though they were identical variables.

·  Journal quartiles and citation indicators vary by database, subject category, and data vintage; values should not be compared across incompatible snapshots.

·  Publicly visible research-integrity indicators identify statistical anomalies or risk patterns; they do not independently establish research misconduct.

·  2026 is provisional because publication and bibliographic indexing dates can lag the publication year. A final full-year 2026 estimate should be frozen after the indexing window closes.

Conclusion

    This study converts the supplied exploratory draft into a publication-oriented quantitative audit with explicit evidence boundaries. The validated evidence shows eight Lebanese universities in the SCImago 2026 higher-education universe, a 17.5% increase in Lebanon’s SCImago country-level indexed document count between 2023 and 2024, a stable AUB original/review journal article count of 1,065 in both 2023 and 2024, a documented top-quartile share of 63% at AUB in 2024, and institution-reporting inconsistencies at USJ and BAU that materially affect cross-university comparison. The local journal ecosystem is larger than what international databases alone reveal, but its indexing and activity statuses need explicit classification.

    The final scientific step is therefore procedural rather than rhetorical: acquire the frozen Scopus/SciVal university-year export, run the predefined cleaning and de-duplication protocol, insert the four-year institutional panel, and rerun the pre-specified statistics. Until that step is completed, claims about the precise annual number of publications by every Lebanese university from 2023 through 2026 would be stronger than the available evidence permits. The present manuscript is designed to be the evidence-safe core of that final paper.

Validated Findings and Evidence Status

Table 7. Evidence status at the 2 October 2026 data cut-off.

References

• American University of Beirut. (2025). Research Facts and Figures 2024. Office of Institutional Research and Assessment / AUB. (Institutional report; Scopus-based annual journal article series and research indicators.)
• Beirut Arab University. (2024). Research Report 2023–2024. Beirut Arab University. (Institutional research report; faculty-level output classifications and university total.)
• Meho, L. I., & Akl, E. A. (2025). Using bibliometrics to detect questionable authorship and affiliation practices and their impact on global research metrics: A case study of 14 universities. Quantitative Science Studies, 6, 63–98. https://doi.org/10.1162/qss_a_00339
• SCImago. (2026). SCImago Institutions Rankings 2026: University Rankings – Lebanon, Higher Education sector. https://www.scimagoir.com/rankings.php?area=all&country=LBN&sector=Higher+educ.
• SCImago. (2025). SCImago Journal & Country Rank: Lebanon country indicators, 2023 and 2024 data vintages. https://www.scimagojr.com/countryrank.php
• SCImago. (2025–2026). SCImago Journal & Country Rank: Lebanese-publisher journal records and SJR quartiles. https://www.scimagojr.com/journalrank.php
• Université Saint-Joseph de Beyrouth. (2024–2025). Institutional research reporting and annual publication indicators for 2023–2024 and 2024–2025. USJ. (Institutional reporting; figures require alignment to the underlying Scopus extraction date and counting definition.)
• Nature Index. (2026). American University of Beirut: Institutional research output in Nature Index. https://www.nature.com/nature-index/institution-outputs/lebanon/american-university-of-beirut-aub
• Sharara, H. (2026). Scientific Publishing in Lebanese Universities, 2023–2026. Working research draft supplied for this study, last updated 2 October 2026.

Appendix A. Reproducible Scopus Query Template

The following query architecture converts the original draft into an auditable extraction protocol. Exact syntax can vary by Scopus interface, so the query should be tested against the institution affiliation IDs before export.

    Recommended grouping for the final field analysis: Medicine/Health; Engineering/Computer Science; Natural/Environmental Sciences; Social Sciences/Business/Economics/Psychology; Arts/Humanities/Law. Architecture should be retained under its Scopus/ASJC engineering classification for the primary quantitative model, with a sensitivity analysis reported separately if appropriate.

Appendix B. Minimum Publication-Level Data Dictionary