Few self-citations among Chilean ecologists

We determine the occurrence of self-citations among 36 Chilean ecologists with the highest h index values recorded in Web of Science. Because the practice of self-citation is perceived as negative by inflating a given researcher’s impact factor, we evaluate if those ecologists (five of them having been awarded the National Prize in Natural Sciences) tend unduly to self-citation, or alternately, receive citations from others ostensibly because their peers recognize their theoretical and empirical output. We use a recently proposed self-citation estimate easily calculated from h index values recorded in the restricted-access Web of Science (Wos) database and the open-access Google Scholar’s (GS) Researcher Profiles and compare these metrics. The Chilean ecologists showed low self-citation values, independently of their status as National Prize awardees. Their publications were highly cited by unrelated peers, likely on account of their novelty or quality. Among middle-aged (50–60 year) and young (< 50 year) Chilean ecologists open-access GS h index values are significantly correlated with those from WoS, thus rendering expeditious this method of citation assessment.


Background
Self-citations (or auto-citations) are those that an author makes of his/her own previous work, whereas allocitations are those made to work not conducted by a given citing author, neither as leader nor as collaborator [1]. Without penalizing self-citations when they are used to put an author's line of research in context, when they describe a specific technique or methodology or study site [2], or have lines of research with few practitioners and/or long-term monitoring along a theme [3], it is assumed that the allocites represent a greater impact than the autocites and that the autocites/allocites ratio is lower as the scientometric impact increases [1].
Self-citations are often perceived negatively, as they may convey a misleading impression of a researcher's impact ( [4], and references therein]). In fact, among n = 107 Ecology journals scrutinized, auto-citations accounted for 16.2 ± 1.3% (mean ± SE) of their Impact Factor in 2004 [5].
Along the same line, an analysis of the publication output of 120 Chilean ecologists found evidence that self-citations significantly increased h-index values [6]. Here, using a quick-and-efficient ("back of the envelope") recently proposed metric [4], we evaluate whether a sample of n = 36 Chilean ecologists grouped by age in 3 non-overlapping classes incur in this inflationary practice.
Because allocites are proportionally more abundant among young researchers (because they do not have many self-authored papers to cite; see Fig. 1 in [3]), low self-citation and higher number of allocites are expected in more recognized senior researchers (because they prefer to cite others to give broader appeal to their research), and higher self-citation is hypothesized for those of intermediate age. Given this rationale, here we compare the 3 age groups previously considered by [7], adjusted for the 10 year elapsed since its publication (from Table 2

Methods
A simple test to determine the proportion of selfcitation in an author's total production from the h-index as reported by Google Scholar (GS) was recently published [4]. The self-citation test (T) assumes that citation patterns follow an exponential curve (see Table 1 in [4]). That proportion, which varies between a minimum of 0 (null autocitation) and a maximum of 1 (full autocitation), is obtained by squaring the h-value and dividing it by the total number of cites (self + allo-cites). According to [4], a test value of 0.35 or more indicates high ratios of self-citation, with values closer to 0.2 indicating low ratios. GS is a broadly used platform that contributes to the Open Science movement [8].
To assess the self-citation pattern of the most-cited Chilean ecologists (i.e., the 10% most cited ones) we obtained from Table 2 of [7] the names of 36 out of 120 ecologists with institutional address in Chile grouped in three age classes: young (< 50 year), middle-aged (50-60 year), and senior (> 60 year). We did not use the entire databases provided by [7] and [6] because ca. 20% of a sample of n = 36 Chilean ecologists does not possess a Researcher Profile in Google Scholar. On the other hand, to use the entire database (n = 120) involves committing errors when assigning names of researchers to a given metric (e.g., [9]).
Five of those ecologists have been awarded Chile's National Prize for Natural Sciences. We reviewed the GS Profiles of each of n = 36 as of June 20, 2020 and October 19, 2020, except for seven who did not have a Researcher's Profile made, thus reducing our sample to 12, 11 and 13 ecologists distributed among the respective age classes: young, middle-aged, and senior. All statistical analyses were performed with the computer package developed by Richard Lowry © http://www.vassarstats.net.

Results
Results are in Table 1, together with the h-index obtained from the Web of Science (WoS), not corrected by time of first publication, as proposed by [7]. On average, the GS h-index is 2.8 times larger than the WoS h-index for senior age-class, 3.1 for middle-aged class, and 4.4 for junior age-class; this is because the latter index considers a broader variety of publications including books, book chapters, and annals, not only papers in journals. The two indices are significantly correlated in the cases of the middle-aged (r = 0.816, P = 0.014, df = 6) and the youngage class (r = 0.666, P = 0.018, df = 10), but not so in the case of the senior-age class (r = 0.598, P = 0.089, df = 10).
In the case of the WoS h-index, the arithmetic means of the 3 age groups differed significantly (One-way ANOVA, F = 29.1, P < 0.0001, df = 2; multiple contrasts for all group comparisons different at P < 0.01, Tukey HSD test). Something similar occurred in the case of the GS h-index (F = 7.3, P = 0.003, df = 2; multiple contrasts different between 1st. and 3rd. groups at P < 0.1, Tukey HSD test). In all 3 age classes, the GS self-citations T test value was not statistically significantly (H = 1.73, P = 0.421, df = 2, Kruskal-Wallis test) and did not exceed the 0.35 empirical threshold [4]. Thus, it is possible to predict the proportion of self-citations from the h-index reported by WoS, especially when it is influenced by selfcites [6].
It is noteworthy that the awardees of the National Prize in Natural Sciences have not incurred in inflationary practices (their test values ranged from 0.229 to 0.311, below the 0.35 critical threshold). Notice that their WoS h-index values ranged from 16 to 29, whereas GS h-index values ranged from 49 to 80. In comparison, the mean WoS h-index for a sample of n = 18 of the most cited ecologists of the world was 45, and those for the editors of high impact journals of Ecology (n = 187) ranged from 9 to 33 [10]. Using the SciELO-Chile database, [11] found a low frequency of self-citation in Natural Sciences, including senior researchers, compared to other disciplinary areas such as Social Sciences and Humanities.

Conclusions
The 10% most-cited Chilean ecologists do not unduly rely on self-citations to increase their Researcher Profiles. To the contrary, their publications have attracted citations from unrelated peers (institutionally speaking), likely because they are theoretically or empirically relevant. Contrary to our predictions, we did not find differences in the pattern of GS self-citations among the 3 age groups of Chilean ecologists compared. On a more speculative vein, they may approach the upper test boundary when they refer more often to their long-term research, either thematic or site-based.
Manuscript preparation: JRR, FMJ. All authors read and approved the final version of the manuscript.