Continue to Work on Building a Publication Record

Perhaps the most comprehensive indication of a young investigator's research accomplishments is his or her record of publication.

It is through publications in the scientific literature that the achievements of scientists are primarily known to the scientific community, and it is through citations to these publications that the impact of a given scientist's contributions can most conveniently be measured. This is not to suggest that these two measures, publication counts and counts of citations, constitute an all-inclusive or sufficient criterion of scientific accomplishment, but only to indicate the general significance of [these] criteria … (NRC, 1977, p. 46).

Unlike most of the indices examined in the previous two chapters, publication records are not limited to those employed in the academic sector and are available for the most recent as well as the older graduates. In fact, many young investigators, ineligible or not ready to apply for their own research grants, have established their "research credentials" by authoring or coauthoring articles that have appeared in refereed journals. Some individuals begin while still in graduate school--frequently in collaboration with their faculty mentors. Others contribute to the scientific literature as part of their involvement in postdoctoral research activities. As a result, most biomedical scientists interested in pursuing careers in research have already compiled records of publications by the time they have completed their graduate and postgraduate training. Moreover, they continue to build on these records throughout their careers in research--irrespective of whether they work in universities, government or industrial laboratories, or some other research setting. Thus a strong record of published articles provides empirical evidence of a young investigator's participation in and contribution to biomedical research.

The chief limitation of publication and citation counts as measures of research productivity lies with the difficulty of compiling data that accurately reflect individual contributions to the scientific literature. The process of identifying individual authors (and coauthors) and linking them with former NIH predoctorals and other biomedical scientists in the two comparison groups involved extensive data processing and verification. For the purposes of this analysis publication records were compiled for a sample of 1,773 biomedical scientists who had earned their doctorates in one of three fiscal years: 1967, 1972, or 1977. All of these individuals had been randomly selected to participate in the NRC biennial Survey of Scientists and Engineers and had responded to at least one of the six surveys conducted during the 1973–81 period--thereby having provided at least some information on their postgraduation employment histories. The publication data examined in this analysis include all articles that appeared during the 1970–80 period in a set of 275 carefully selected journals¹ covering a broad range of biomedical research areas. A total of approximatley 21,100 articles² were intially linked with the 1,773 individuals in the sample, solely on the basis of author's last name and first initial. Since this list obviously included many papers by individuals who were not in the sample but who had names similar to those of sample members, the set of articles identified with each individual was carefully scrutinized to determine whether or not the individual had in fact authored each of the articles attributed to him or her. This determination was made on the basis of fields of research, institutional affiliation, full names of authors (when known), and other available clues. After thorough scrutiny the list was reduced to approximately 8,700 articles that were determined to have been authored or coauthored by sample members during the 11-year span between 1970 and 1980.

These articles were collated (by computer) with source files from the Science Citation Index (Institute for Scientific Information, 1970–80), which provided specific citations to each article. For the 8,700 articles identified, a total of 67,800 citations were found in papers that appeared in the 275 selected biomedical journals during the 1970–80 period. Because of the complexity and uncertainties involved in this collation process, there was some reason to suspect that a significant number of citations might have been missed. To investigate this possibility a random sample of 100 articles published in 1978 was selected, and citations to each article were looked up using a library copy of the 1979 and 1980 volumes of the Science Citation Index. This "look-up" uncovered as many as 857 citations (in this two-year span) to the 100 articles, compared with 449 citations that had been derived by the computer collation. Most of the new citations discovered in the library volumes, however, were from articles in books, foreign journals, and other publications³, not included in the set of 275 biomedical journals, and the correlation found between the annual number of computer-derived citations to each article and the number discovered through the library search--excluding those not in the biomedical journal set--is as high as .97. Furthermore, the correlation between annual number of computer-derived citations from the restricted journal set and the citation count based on all scientific journals covered in the Science Citation Index, is .92. On the basis of these findings we conclude that the computer derivation of citation data presented in this chapter produced highly accurate indices of research productivity, even though the restriction to the 275-journal set has resulted in a substantial underestimation of the total citation count.

Record of Publication In terms of the frequency with which individual scientists have produced published articles, the early career records suggest an easily recognizable pattern. Figure 7.1 describes the publication activity of FY1972 Ph.D. recipients--a group for whom we have publication data prior to their graduation, as well as for several subsequent years. Relatively few of these individuals contributed papers that were published before they had completed their doctoral training, but many produced articles in the two years immediately after graduation--presumably reflecting the publication of their doctoral dissertations. The publication frequency gradually declines over the next several years, as some young scientists finished their postdoctoral apprenticeships and moved on to career positions not involving extensive research activity or into junior faculty positions in which research time is limited. Many individuals, on the other hand, have continued to be involved in research as evidenced by the fact that in each year since 1978 more than one-fourth of all FY1972 Ph.D.s had contributed one or more published articles.

FIGURE 7.1

Percent of former NIH predoctorals and other FY1972 biomedical science Ph.D.s who had one or more articles published in a particular year, 1970–80. See Table 7.1.

Of particular interest to this analysis is the finding that throughout their early careers former NIH trainees and fellows have been more likely to produce articles than have members of either comparison group. For the FY1972 cohort, the differences are greatest in the years immediately after graduation when approximately 40 percent of the NIH predoctorals contributed papers annually. Furthermore, 84 percent of this group had at least one article published sometime during the 1970–80 period, compared with 64 percent of the FY1972 biomedical science Ph.D.s in Group I and 56 percent of those in Group II (Table 7.1). For the FY1967 and FY1977 cohorts, similar differences were found between the NIH-supported graduates and the two comparison groups.

TABLE 7.1

Percent of the FY1967, FY1972, and FY1977 Ph.D. Recipients Who Had One or More Articles Published During the 1970-80 Period.

Total Articles and Citations The tendency for former NIH predoctorals to publish more frequently than their colleagues is reflected, as might be expected, in the total numbers of articles each individual produced during the 1970–80 period (Figure 7.2). With regard to the FY1967 Ph.D.s, for example, those who had received NIH training support produced, on the average, one-third more articles than those in either Group I or Group II. The differences in the total citations to papers by members of each group are even greater. A FY1967 graduate who had been supported by the NIH accrued an average of more than 66 article citations during the 11-year span (Figure 7.2). The comparable figures for Groups I and II are 39 and 31 citations, respectively. For the more recent cohorts, the average citation counts per individual are, of course, considerably smaller, but heavily favor the NIH predoctorals, nevertheless.

FIGURE 7.2

Average number of articles by former NIH predoctorals and other biomedical science Ph.D.s that were published during the 1970–80 period and average number of article citations received per individual. See Tables 7.2 and 7.3.

Citation Rates The higher citation counts (per individual) attributed to former NIH trainees and fellows reflect, in part, the fact that these individuals have typically produced more papers than have their biomedical science colleagues. Of further interest is whether or not the NIH-supported group has also accrued more citations per published article. Several studies (Jones, 1980) have demonstrated that this citation index is a meaningful measure of the research productivity of scientists--and one that is highly correlated with peer judgments and other measures of the quality of the scientific contribution. Figure 7.3 presents findings from an analysis of the citations to papers authored by FY1967, FY1972, and FY1977 Ph.D.s. Articles written by former NIH predoctorals in each cohort have received more citations per paper than have their Group I colleagues' articles, which in turn have been cited more frequently than articles by those in Group II. With regard to the FY1967 cohort, for example, papers authored by NIH-supported graduates were referenced in the biomedical research literature an average of 8.3 times during the FY1970–80 period, compared with an average of 6.4 and 5.3 citations to papers by graduates in Groups I and II, respectively. Somewhat surprising perhaps is the finding that papers by FY1972 Ph.D.s have been cited more frequently than have those by FY1967 Ph.D.s, even though the latter group has been publishing considerably longer (and thus their papers have had greater opportunity to accrue citations). As illustrated in Figure 7.3, this finding is applicable to the two comparison groups as well as to the NIH predoctorals. A possible explanation for the lower citation rates of FY1967 graduates is that the data reported do not include articles published prior to 1970 when many of these individuals were involved in postdoctoral training.

FIGURE 7.3

Average number of citations per article published during the 1970–80 period. See Table 7.4.

TABLE 7.2

Average Number of Articles Published by FY1967, FY1972, and FY1977 Ph.D. Recipients During the 1970-80 Period.

TABLE 7.3

Average Number of Citations to Articles Published by FY1967, FY1972, and FY1977 Ph.D. Recipients During the 1970-80 Period.

TABLE 7.4

Average Number of Citations per Article Published by FY1967, FY1972, and FY1977 Ph.D. Recipients During the 1970-80 Period.

Publication Records of Former Postdoctorals Results presented in Chapter 4 reveal that, in comparison with other biomedical science Ph.D.s, a larger fraction of the NIH-supported trainees and fellows planned to take postdoctoral appointments after graduation. As discussed, these appointments have afforded young scientists opportunities to devote their full energies to research pursuits and at the same time compile requisite records of publications. Consequently, one might hypothesize that the differences observed in the publication records of former NIH predoctorals and their colleagues may be largely attributed to the fact that more of the former group have had postdoctoral research opportunities. To test this hypothesis separate analyses were made of the publication and citation records of graduates who indicated that they would take postdoctoral appointments after completion of their doctoral requirements and those with other postgraduation plans. Findings for the FY1972 cohort are described in Figure 7.4. The publication counts include only those articles published within three years following receipt of the doctorate (i.e, 1973–75)--a time when many of these graduates were involved in postdoctoral research training. Not at all surprising is the finding that graduates with definite plans for postdoctoral study produced approximately twice as many papers as did their colleagues. Furthermore, their papers have accrued, on the average, more than twice as many citations per article. Of greater interest is the finding that, among those who planned postdoctoral study, the NIH-supported graduates surpassed individuals in either comparison group--in terms of both the number of articles authored and the average number of citations per article. With respect to the publication records of those who did not have definite plans to take postdoctoral appointments, the findings are similar, although the differences between the NIH group and each of the two comparison groups are not as large. On the basis of these results we reject the hypothesis that the tendency for former NIH predoctorals to take postdoctoral apprenticeships primarily accounts for their superior records of publication.

FIGURE 7.4

1973–75 publication and citation rates for FY1972 Ph.D.s with definite plans for postdoctoral study and corresponding rates for other FY1972 graduates. See Table 7.5.

TABLE 7.5

Publication and Citation Rates for FY1972 and FY1977 Ph.D Recipients with Definite Plans for Postdoctoral Study, Compared with Rates for Other Graduates.

Summary The principal findings presented in this chapter--that the NIH-supported graduates have produced more papers than their colleagues and that their papers have been cited more frequently--reinforces the conclusions reached in earlier chapters. There can be no doubt that, in comparison with other biomedical scientists, members of the NIH group have been more successful in their pursuits of careers as independent investigators. This success is measured by their demonstrated involvement in research activities (e.g., applying for federal research grants or contributing to the biomedical research literature) and by their research accomplishments (e.g., the receipt of grant awards or high article citation rates). Whether their success may be attributed to the superior ability and research potential of those selected to receive NIH predoctoral support, the quality of the graduate training they received, or perhaps other factors cannot be ascertained from this analysis. This important issue is discussed at some length in the following chapter, which provides an overall summary and interpretation of the study findings.

1

For a profile of these journals, see Narin, 1983.

2

The publication records--including names of all authors and coauthors, article title, and journal and date of publication--were derived from the MEDLARS system, a computerized information system maintained by the National Library of Medicine.

3

For a list of these other journals, see Appendix C.

thornhillsureat.blogspot.com

Source: https://www.ncbi.nlm.nih.gov/books/NBK547302/

Share this post