Introduction

Scientific production in Latin America represents only 3% of international production ^{(RÍOS GÓMEZ; HERRERO SOLANA, 2005)}. Despite this, from 1996 to 2007, scientific production in Latin America increased by 137% ( ^{SANTA; HERRERO SOLANA, 2010b}, p. 384). Likewise, according to other studies, from 2010 to 2015, Latin American production increased by 37%, according to articles published in Scopus (ALBORNOZ; BARRERE; SOKIL, 2017). This indicates that there is an upward trend.

However, during the first time period, Peru ranked last in terms of scientific production among ten countries, representing only 0.03% of the total international production. (SANTA; HERRERO SOLANA; [-@]75, p. 387). Furthermore, in Peru, there are only 29 indexed journals (in Scopus, Emerging Sources Citation Index (ESCI) or SciELO), which belong to 14 Peruvian universities ( ^{ESTRADA-CUZCANO; BARRIONUEVO-FLORES; ALHUAY-QUISPE, 2018}), out of a total of 143 higher education institutions in the country ( ^{SUNEDU, 2019}). This data reveals the challenging situation prevailing in scientific research production in Peru. Therefore, it is important to investigate the aspects that are related to the results of scientific research in the country.

This research addresses the problem of scientific research production in professors teaching in a particular program at Universidad Científica del Sur ² (Southern Scientific University) (UCSUR) in Peru. The following aspects are analyzed: teaching performance, hours dedicated to teaching, hours dedicated to research, researcher status recognized by the institution, gender, and academic degree of the professor. The main objective is to determine the aspects related to scientific research production in Google Scholar and Scopus. These variables have been considered because, first, it is necessary to ask whether a higher teaching performance is related to a higher scientific research production, according to different authors ( ^{ARTÉS; PEDRAJA-CHAPARRO; SALINAS-JIMENÉZ, 2017}; ^{COATE; BARNETT; WILLIAMS, 2001}; ^{HATTIE; MARSH, 1996}; ^{PALALI et al., 2018} ; ^{UZ ZAMAN, 2004}). Second, it may be possible that a professor, who spends more time on research and is recognized as a researcher, may have a higher number of research studies. Third, a similar point applies to the relation between academic degree and scientific production-as demonstrated, for example, by the research of Mejía, Valladares-Garrido and Valladares-Garrido ( ²⁰¹⁸), regarding physicians registered with the Peruvian Physician Association.

These aspects, mentioned above, shall be analyzed in the context of a human historical period caused by the outbreak of the Covid-19 pandemic and the consequent changes in social habits ( ^{AYALA-COLQUI, 2020}, ²⁰²²).

Health restrictions forced researchers and university professors to work from home, which had significant personal, social, and academic repercussions that affected scientific production ( ^{LAUDA-RODRIGUEZ et al., 2020} ), especially for female researchers due to gender inequalities in the distribution of household and childcare responsibilities ( ^{INNO; ROTUNDI; PICCIALLI, 2020}). In most cases, this has implied implementing a distance and virtual teaching practice model. In this sense, the main research question is: What are the aspects related to scientific research production in Google Scholar and Scopus of professors from a private university in Lima, Peru, during the pandemic caused by Covid-19?

The variable “teaching performance” is defined as the set of competencies of the professor that allows the development of the teaching-learning process ( ^{MONTES DE OCA; MACHADO, 2011}; ^{RUEDA, 2009}). This is generally approached in relation to the evaluation of teaching performance, which is a complex process that, by involving cognitive, instrumental, and axiological aspects, ( ^{TEJEDOR; GARCÍA-VARCÁRCEL, 2010}), allows to evaluate teaching performance (ELIZALDE; REYES, 2008; ^{RUEDA, 2009}; ^{VAILLANT, 2008}). This evaluation may be conducted using different models or strategies such as the evaluation based on the opinion of professors, academic supervisors, students, standardized instruments or rubrics, student achievements, and a skills-based model ( ^{TEJEDOR, 2012}). In this study, teaching performance is determined through a “teacher satisfaction survey” completed by students, who grade their professors according to their teaching skills performance.

The variables: hours dedicated to teaching, hours dedicated to research, researcher status recognized by the institution, gender, and academic degree of the professor, do not require further explanation. These variables have been considered due to previous research focused on aspects related to scientific research publication ( ^{ANGULO- GIRALDO, 2020}; ^{MEJÍA; VALLADARES-GARRIDO; VALLADARES-GARRIDO, 2018}; ^{PEREYRA-ELÍAS et al., 2014} ). Researcher status is an internal category of the university where professors of our institution teach: It is possible to become an associate researcher (and enjoy a few restricted hours for research), a part-time senior researcher (whose hours dedicated to research amount to an average of 20), and a full-time senior researcher (whose hours dedicated to research tend to be 40).

The variable “scientific publication” is understood in this study, exclusively, as the sum of publications in Scopus and Google Scholar (discounting, of course, research in the first database that is repeated in the second one). However, as discussed further below, the analyses will unbundle both databases in order to closely analyze the articles only published in Scopus, in Google Scholar versus the articles published in both databases. Theoretically, it is important to distinguish between scientific production and scientific publication since they are empirically different phenomena. Scientific production involves both the outcome material of the acquired scientific output, as well as the activities related to research ( ^{CHÚA; OROZCO, 2016}). In contrast, the scientific publication refers to the process by which scientific production is exposed, made visible and shared ( ^{MONTCUSÍ; TORMO, 2017}). Both are studied by scientometrics and bibliometrics, the former being the quantitative analysis of the development of scientific knowledge and, the latter, involves statistical studies of scientific activity ( ^{ARAUJO; ARENCIBIA, 2002}; ^{SPINAK, 2001}). Moreover, the publications may be made on different media. However, journals are the main and substantive means of scientific production dissemination ( ^{AGUADO-LÓPEZ et al., 2014} ; MENDOZA; PARAVIC, 2006; ^{MIGUEL, 2011}; ^{PATALANO, 2005}).

Journal indexing, understood as the publication of journals in a renowned database, is “the most significant indicator of the production capacity of a researcher and, therefore, it is an essential factor in the overall scientific evaluation” ( ^{AGUADO-LÓPEZ et al., 2014} , p. 571). The two most widely used multidisciplinary databases, both of which are indexed, are Web of Science (under the domain of Thomson Reuters) and Scopus (under the domain of Elsevier). This is due to their relatively wide coverage, accessibility, and utility ^{AGUADO-LÓPEZ et al., 2014} ; ^{CORTÉS, 2008}; ^{SANTA; HERRERO SOLANA, 2010b}; ^{SANTA; HERRERO SOLANA, 2010a}; ^{PATALANO, 2005}).

However, these databases do not cover the entire scientific production of a researcher. In contrast, Google Scholar offers an alternative that brings together the production of several databases, as it has the advantage of being free of charge, greater coverage, and access to a wide range of documents such as books, articles, conference papers, theses or any other publication with an abstract (TORRES-SALINAS; RUIZ-PÉREZ; DELGADO-LÓPEZ-CÓZAR, 2009). That is why our scientific production variable includes Google Scholar, in addition to Scopus, because it is wider and more inclusive.

In terms of background, in Peru, there is a number of research studies that address the factors related to the scientific research production of professors from human medicine program. First, Parra ( ²⁰¹⁰) focuses specifically on gastroenterologists. In this study, a qualitative methodology is employed, in which interviews with physicians are included, who provide their perceptions of the aspects related to publication. The multivariate analysis performed by the author allows concluding that the relevant aspects are divided into categories such as “personal” (belonging to a scientific society), “work” (support for research in the workplace and number of jobs) and “academic” (type of bibliographic research and comprehension level of authentic articles) ( ^{PARRA, 2010}, p. 4). Second, an article by Pereyra-Elías et al. ( ²⁰¹⁴) was found. This paper focuses on Peruvian medical program professors (32 schools) in 2011, especially, those who teach academic courses related to research, in order to determine the aspects related to publication. Among these aspects are age (under 40 years old), to be a university professor with a compulsory thesis, and to work in a university with a high level of research production. Third, the paper by Atamari-Anahui, Sucasaca-Rodríguez and Marroquín- Santa Cruz ( ²⁰¹⁶), addresses scientific production of thesis advisors from a university in Cuzco. However, the text simply provides publication statistics (indicating a low level of publications) without explaining the reasons for this situation based on a statistical test. Fourth, the article by Merino-Soto and Salas-Blas ( ²⁰¹⁶) analyzes, qualitatively, the motivations and difficulties for conducting research in Peru; from the sample results, it was found that the main motivations are social and individual, and the most important hindrances are difficulties of a personal, instrumental, and environmental nature. Fifth, the publication by Atamari-Anahui et al. ( ²⁰¹⁶) identified that physicians from Cusco, who teach in medical schools, are more likely to publish compared to those who do not teach in universities. Sixth, the article by Mejía, Valladares-Garrido and Valladares- Garrido ( ²⁰¹⁸) focuses on the publication of physicians by determining the academic degree as a variable. As a result, it was found that those who have a higher chance to publish are those who have both a master’s degree and a doctorate. Seventh, Chachaima- Mar, Fernández-Guzmán and Atamari-Anahui ( ²⁰¹⁹) in their paper, highlight that “being a coordinator of an academic course increased the possibilities to publish” (p. 9).

Finally, regarding research that addresses the pandemic scenario and the fact that many journals allowed free access to their papers with the aim to foster medical research on Covid-19, Torres-Salinas ( ²⁰²⁰) mentions that there was an exponential increase in publications on the pandemic (R2 =0.92), especially, in medical journals. Gregorio- Chaviano, Limaymanta and López-Mesa ( ²⁰²⁰) agree on the exponential increase in articles and add that, in Latin America, the countries that published the most were Mexico, Brazil and Colombia. Ortiz-Núñez ( ²⁰²⁰) adds that, in the case of Scopus, publications on Covid-19 were mostly in English, with multiple authorship, a high number of collaborations, a high visibility in social networks and publications in first quartiles. The author refers to medical publications, which mainly dealt with three fields: the description of the new disease, clinical studies, and possible treatments. Peru has not been the exception and the enormous effort in medical research is reflected, for example, in its participation in the Solidaridad (Solidarity) project, a massive global clinical trial to seek treatment for Covid-19 ( ^{SOTO et al., 2020} ). Likewise, the impact of the pandemic on the Peruvian academic community is reflected in two Peruvian medical journals that greatly increased their research on the subject of the pandemic: Acta Médica Peruana (Peruvian Medical Journal) (ISSN 1728-5917) and Revista Peruana de Medicina Experimental y Salud Pública (Peruvian Journal on Experimental Medicine and Public Health) (ISSN 1726-4634) ( ^{RODRÍGUEZ-MORALES et al., 2020} ; ^{ROMANI, 2020}).

These documents clearly show a great number of publications regarding Covid- 19; however, they do not focus on the specific and tangible aspects that, within the pandemic context, are related to publications. Therefore, these papers are highly descriptive and merely report on the increase in research in the context of the disease without investigating the observable and measurable reasons behind it, beyond the need to understand the causes and treatments of this new viral outbreak. On the other hand, the aim of this research is precisely to address the latter: to identify the internal institutional aspects related to scientific research production in the context of the Covid-19 pandemic.

Methodology

Results

Table 1 presents a summary of the variables used in Models 1, 2, and 3, while Table 2 does the same for Models 4, 5, and 6, respectively. It was noted that imputing values had significant consequences for Model 3 ( Scopus). Moreover, it appears to severely modify the coefficients of all predictor variables in Models 4, 5, and 6, except for gender. In other words, the expansion of the sample generated considerable changes in the estimations of the final adjusted models (see Table 3 and Table 4).

Some patterns in the model fit shown in Tables 3 and 4 should be mentioned. Using imputed values, the models with the best fit, both in logistic and gamma regressions, were models 2 and 5 ( Google Scholar). Those with the worst fit were models 3 and 6 ( Scopus). This relationship was not found among the models with no imputed values, due to the changes introduced by the added case values. However, in all cases, only results with models showing optimal fit will be discussed.

Discussion

The models which relate the number of publications and those appearing in Google Scholar exhibit higher levels of significance compared to the models regressing publications in Scopus. Additionally, the first models demonstrate significant associations with a broader range of variables. These findings substantiate the low efficacy of predictors commonly identified in the bibliography as determinants of scientific productivity ( ^{RØRSTAD; AKSNES, 2015}).

Considering that university professors in Peru can hold either a master’s degree or doctorate, the academic degree has shown to be highly favorable and statistically significant for scientific production in the conducted logistic regressions. Previous research conducted by Mejía, Valladares-Garrido, and Valladares-Garrido ( ²⁰¹⁸) found that a higher academic degree is related to higher scientific production. However, these findings contrast with those reported by ^{Castro-Rodríguez et al. (2020)}, who did not find a significant relationship between these variables among dental professors at the Universidad Nacional Mayor de San Marcos. Needless to say, in the country, only 18% of the professors hold a doctorate, while 49% possess a master’s degree (SUNEDU, 2020).

The second variable whose importance permeated all the models was research hours. This variable has received limited attention in the academic bibliography on the subject. A study by Spanish authors measured “research intensity” through path analysis models ( ^{BERBEGAL-MIRABENT et al., 2018} ) and found that it acts as a moderator between teaching quality and student satisfaction with the professor. The authors examined all possible relationships among the three variables, including research hours as a result. In the context of the pandemic, this variable may have been favored by the fact that, as Torres-Salinas ( ²⁰²⁰) demonstrates, in several places in the world, special attention and research care was devoted to the issue of Covid-19: considering that these professors already had hours reserved for research, the pandemic may have concentrated their research efforts on such topics, driven by the urgency of the new disease.

However, it should be noted research hours were measured “intramurally”, i.e., only within a university, while research hours in other institutions were left aside. In terms of academic degrees, only postgraduates have on average more than 0 hours of research (master’s degree=1, doctorate=5, n=162). The reason for this is due to the internal policies of the institution. In fact, institutional characteristics such as university funding ( ^{BARLETTA et al., 2017} ; ^{RHAIEM, 2017}), optimal human capital utilization by the university, and optimal working environment for research ( ^{KANNEBLEY et al., 2018} ; WAY et al., 2019) are crucial in scientific productivity. Furthermore, the increase of research hours in the university would theoretically facilitate local and intramural scientific collaboration, as well as its presence (and a less individualistic and more egalitarian or hierarchical culture) seems to facilitate the increase of scientific articles and their quality ( ^{ABRAMO et al., 2017} ; ^{CHEN; MA, 2015}).

A relevant issue is related to the gender variable. Research reveals that women produce fewer articles in general and in Google Scholar. Moreover, five of the six regressions show that gender is negatively related to the scientific productivity variables. Despite greater scientific production in the context of Covid-19, the gender gap has widened, so that men have been rewarded while women, who have taken on more responsibilities in their own households ( ^{MYERS et al., 2020} ; ^{VINCENT-LAMARRE et al., 2020} ), who have had to bear more of this domestic “unpaid work” ( ^{FEDERICI, 2018}), have been penalized. Regarding this, an analysis of articles submitted for review in Elsevier journals between February and May 2020 showed that women submitted fewer Covid-19-related manuscripts in 2020, especially in health and medical journals, with significant values for the areas of life sciences, physical sciences, and engineering (^{SQUAZZONI et al., 2020}).

This research, however, also finds a significant increase commensurate with the degree. Although gender inequalities are common in academia and also in health-related disciplines ( ^{BEAUDRY; LARIVIÈRE, 2016}), women with doctorates tend to match their male peers over time ( ^{RØRSTAD; AKSNER, 2015}). Our larger base confirms that assumption (the ratio of total publications per person progresses from 7/2 for bachelors, 5/2 for masters, and 6/4 for doctors).

One finding of this study was the low importance of teaching quality factors, such as hours and performance scores. Many international studies confirm that the relationship between teaching quality and scientific publication is ambiguous and rather depends on each local context ( ^{TIGHT, 2016}), and some authors find different relationships depending on the quality and productivity measures used ( ^{PALALI et al., 2018} ). However, even the latest aforementioned authors have not found any relationships between the presence or quantity of publications and teaching quality. The same applies to technology transfer, which highlights the requirement of full dedication to research ( ^{BEAUDRY; LARIVIÈRE, 2016}). Nevertheless, consider that the teaching quality score was an average derived from several variables. Among these variables, the professor proficiency in the use of the virtual platform was the one that showed a slight increase in significance in the models. Regarding other matters, it is important to mention that the “scientific publication” variable is a factor that inevitably has an impact and influence in the classroom. For instance, previous studies have shown that a research-oriented professor is perceived to be better in terms of their teaching practice ^{(ROSE; MCKINLEY, [2020])}. And, in some cases, a positive correlation has been found between being a researcher and being a good professor ( ^{ARTÉS; PEDRAJA-CHAPARRO; SALINAS-JIMENÉZ, 2017}; ^{PALALI et al., 2018} ). For this reason, Boyer ( ¹⁹⁹⁰) stated that “great professors” have as a necessary condition the possession of knowledge acquired both through experience and teaching practice, as well as through research. Hence, although our research shows results on scientific productivity, the knowledge of the factors that determine it can indirectly help to improve teaching practice in the classroom since researching professors will have a better self-perception of their teaching practice, as well as updated and first-hand knowledge.

Returning to the question of factors related to productivity, other factors must also be considered. Latin America is known to produce fewer scientific papers than other regions in the world, although this trend is expected to change in the coming years ( ^{JAVED; LIU, 2018}). However, there is a scarcity of publications on methods to improve academic efficiency in the health academic areas of developing countries ( ^{OBUKU et al., 2018} ). Despite this, some reviews of interventions focused on literacy for university professors have been found with excellent results ( ^{FRANTZ; AMOSUN, 2011}; ^{HAVNAER et al., 2017} ).

Regarding general recommendations, the model of scientific effectiveness of Rhaiem ( ²⁰¹⁷) indicates that institutional decisions are fundamental for the increase of scientific publications. Recommendations include an emphasis on gender and academic degree differences, as well as the creation of ideal work environments to encourage and retain highly research-performing professors. This recommendation is important because teaching activities interfere with research activities. Finally, the United Way model seems appropriate to the needs of health career professors in order to level knowledge inequalities, albeit it is not the subject of this study ( ^{HAVNAER et al., 2017} ).

Limitations

This study stands out from others because it divides scientific productivity into the presence or absence of research and its quantity ( ^{PALALI et al., 2018} ) The combined use of gamma and logistic regressions is an adequate innovation to our variables; other studies use OLS regressions or a mixture of linear and logistic regressions. This is because many studies measure both quality and quantity of publications as dependent variables. Furthermore, it was crucial to differentiate between publications in Google Scholar and Scopus, thereby identifying the differences between these repositories. Finally, this research takes into account academic degree and research hours in the regressions, which is uncommon in the bibliography.

However, there are several limitations to be acknowledged. The first limitation is related to the small size of our sample, whereas other studies have samples that consist of thousands of researchers. The second limitation is the relative scarcity of independent variables. Future studies should incorporate variables related to scientific collaboration, funding, age, informal strategies (e.g., participation) or formal strategies (e.g., being a lead author) for positioning of professors ( ^{RODRÍGUEZ; RUBIO, 2016}; VUONG et al., 2018), as well as the literacy skills of professors. Thirdly, future studies should consider both the quantity and quality of publications and differentiate the individual contribution of professors ( ^{VUONG et al., 2018} ).

The academic degree variable exhibited a significantly positive association with scientific productivity in our logistic regressions. However, this variable was not included in the gamma regressions due to the limited number of positive cases, which resulted in biased estimates and multicollinearity issues.

Another limitation was the presence of extreme values and data dispersion, which decreased the quality of the regressions. Future research is necessary to account for unexplained residuals. It is recommended to strike a balance between the available data and theoretically relevant variables. Finally, it is recommended to have access to the data repository and to documents detailing the procedures used to collect the data (a final limitation of this study).

Conclusions

This study had the following findings: the main factors associated with scientific production in a private university in Lima, Peru, are the academic degree. This determines the presence of publications, and research hours, which are associated with both publications and their increase. It is important to note that in the Peruvian context, according to SUNEDU (2020), only 18% of professors hold a doctorate, while 49% have a master’s degree. This fact has a direct impact on the studied scientific productivity. In addition to the findings, it should be noted that in Peru, postgraduate studies are not free of charge in any case. Therefore, individuals who aspire to obtain a higher academic degree must invest both additional time (beyond their working hours) and money (as the Peruvian government does not provide scholarships for national studies). This situation poses challenges to the training of doctoral professors and their future relevance in scientific production. Despite this, it is important to mention that in this research, the quality and dedication of teaching are null or negatively related to the presence of publications.

Furthermore, it is concerning to note another finding that reinforces the gender gap in academia: being male is associated with a greater presence of scientific publications. This finding is consistent with other Latin American studies. For instance, Alarco et al. ( ²⁰²¹) emphazise that only 17.3% out of 1750 Latin American researchers listed in Webometrics are women, which indicates their underrepresentation. The research conducted by ^{Colther, Rojas-Mora, and Wong} identifies that the gender of female students and professors in Chilean universities has an impact on the scientific productivity of their institutions.

Additionally, it is important to highlight the gender gap in academia on a global scale. For instance, in France, Mairesse and Pezzoni ( ²⁰¹⁵) found that male researchers have a scientific productivity that is 2/3 higher than that of their female counterparts in French universities. Similarly, in Italy, Filandri and Pasqua ( ²⁰²¹) discovered that female researchers have fewer publications compared to male colleagues, which directly impacts their career advancement in academia.

The latter finding is significant because, although there is a demonstrated correlation between research hours and scientific productivity, the gender gap may be hindering women researchers’ access to dedicated research time. Therefore, it remains to influence policies specifically aimed at addressing the gender gap in a transversal manner in the promotion of research in all its dimensions.

It is recognized that the increased family burdens faced by women compared to men make it challenging to meet criteria of excellence. In the field of science, these domestic work hours are not taken into consideration ( ^{CABALLERO-VILLALOBOS et al.} ). In particular, during the pandemic, the personal, social, and academic disruptions that affected scientific productivity were accentuated in the case of women researchers due to gender inequity in the distribution of household and childcare responsibilities ( ^{INNO; ROTUNDI; PICCIALLI, 2020}).

These findings confirm trends explained in the existing bibliography. In the context of the pandemic, we observed that the scientific production of professors accumulated disparate numbers whose main factors were only grade, hours of research, and male gender.

In general, findings showed that the determinants of publication in Google Scholar are not the same as those of publication in Scopus. In fact, the models related to Scholar are more robust in relation to the other variables, indicating that publication in Scopus is very scarce and limited to professionals who only work at the university studied. The adoption of modern scientific effectiveness models is recommended to increase the collaboration and quality of researchers, as well as a literacy intervention model to shorten the cognitive distances between them.

Regarding the limitations, the study focused on a very small sample size of professors, which contrasts with the cited background studies that examined thousands of professors. Furthermore, the study only considered a specific school and did not consider all schools within the studied university. Moreover, other potentially relevant independent variables were not considered in the study, such as scientific collaboration, funding, age, informal strategies (e.g., participation), formal strategies (e.g., being the lead author), and others. Indeed, future research can be expected to work with larger samples, comparing results among different schools within the same university, and even across different universities. In addition, future research should incorporate the aforementioned unaddressed independent variables and include variables related to the quantity and the quality of publications. This includes considering traditional scientometric standards such as author H-index and citation count, as well as exploring alternative metrics (altimetric).