Influences on Authorship Issues: An Evaluation of Giving Credit

Abstract

A survey on authorship issues was conducted with academic chemists in Ph.D.-granting institutions in the United States. Six hundred faculty members responded. The respondents reported a wide range in their attitudes and behavior regarding giving credit in a publication. The various guidelines for authorship are independent of academic background factors such as the relationship between the senior author and the contributor-potential author. However, the survey data reveal significant context-dependency by the respondents. Many respondents would give more credit to their own student than to another professor's student for the exact same contribution to a research project. The survey data further shows that the faculty who received their Ph.D. in the 1940s, 1950s, and 1960s are the most likely to provide authorship, while those who received their Ph.D. in the 1990s and 2000s would most likely give either no credit or acknowledgments.

Keywords:

• acknowledgment
• authorship
• ghost author
• guest author
• publication
• responsible conduct of research
• RCR

INTRODUCTION

The field of responsible conduct of research (RCR) encompasses every aspect of the planning, funding, conducting, reviewing, and reporting of research (Macrina, 2000; Shamoo and Resnik, 2003; Steneck, 2007; White House Office of Science and Technology Policy, 2000). The most intensely studied aspects of RCR include research misconduct, protection of the rights of human subjects and the welfare of laboratory animals, conflicts of interest, and authorship practices. Of all the areas of RCR, authorship involves everyone who participates in a research project—and, in fact, sometimes it involves individuals who have not participated in the research. Complicating the practice of RCR is the fact that different organizations provide guidelines that are typically consistent with each other but not identical (Claxton, 2005b; Tarnow et al., 2004).

For example, the Committee of Journal Medical Editors' (ICJME) statement on Authorship and Contributorship provides specific types of contributions that authorship criteria should be based upon, e.g., acquisition, analysis and interpretation of data, and drafting, revising, and giving final approval of the manuscript to be published (International Committee of Medical Journal Editors, 2010). Similarly, the American Astronomical Society states that “All persons who have made significant contributions to a work intended for publication should be offered the opportunity to be listed as authors … contributed significantly to the inception, design, execution, or interpretation of the research to be reported” (American Astronomical Society, 2010). In contrast, the American Chemical Society's Ethical Guidelines for Publication states, in part, that co-authors should “have made significant scientific contributions to the work reported” without delineating any specific types of contributions (American Chemical Society, 2006).

Authorship decisions are serious, given today's highly competitive environment in science. Omitting individuals who meet the criteria for authorship (“ghost authorship,” see Flanagin et al., 1998; Johnson, 2005; Rennie et al., 1997) can have serious consequences for many tangible necessities as well as rewards in one's profession (for another definition of “ghost authorship”, see: Shamoo and Resnik, 2003). Indeed, omitting an individual from authorship who meets the criteria of authorship can be considered a form of plagiarism (Bouville, 2008) in some circumstances. Plagiarism along with fabrication and falsification (Claxton, 2005a) constitute the major forms of misconduct of science (Macrina, 2000; Steneck, 2007; White House Office of Science and Technology Policy, 2000). Including an individual as an author of a publication who clearly fails to meet the criteria for authorship (“guest” or “honorary” authorship, see Bates et al., 2004; Flanagin et al., 1998) dilutes the value of being a deserving author, demoralizes the participants of the research program, and corrupts the practice of research. Guest authorship may well be characterized as falsifying the research record in that the authorship information is not accurately represented.

Authorship practices have two interpenetrating yet obverse facets: receiving credit and giving credit. In the first paper in our studies on authorship issues (Seeman and House, In press), we examined the experiences of 600 respondents to our survey, all academic chemists in Ph.D.-granting institutions in the United States, regarding their experiences in receiving, or not receiving, credit for their research contributions (Seeman and House, In press). In this article, we examine the respondents' perceptions and behaviors in giving credit to others. In particular, we wished to examine the age dependency of senior authors in their decisions about giving credit as well as the context-dependency, if any, of their decisions. Future articles will focus on the “rules” used by the participants in making authorship decisions, how the respondents developed or adopted such rules, and conflict resolution when the decisions of the senior authors are not (at least initially) accepted by their research colleagues.

METHODS

Full details about the survey, the treatment and analysis of the data, and a summary of the questions can be found in the first article in this series (Seeman and House, In press). A short summary of the key points is as follows. The e-mail addresses of 3,990 faculty, including emeritus faculty, from departments of chemistry from 152 colleges and universities granting Ph.D. degrees representing all 50 states in the United States were obtained from the institution's Web sites. The survey was programmed in the Survey Crafter® software and administered via the internet. Approvals were obtained from The University of Richmond Institutional Review Board (IRB). Six hundred individuals referred to herein as the “respondents” (a 14% response rate taking into consideration undeliverable e-mails) made it through the entire survey. No identifying information such as respondents' e-mail addresses, university affiliations, or gender were collected which could identify any respondent or location. Statistical Package for the Social Sciences® (SPSS) was used for the statistical analyses, and UCINET® was used for some specific manipulations of the data when we created a matrix with the data from Question 23.

RESULTS

Question 23 of the survey was:

A suggestion was made by Person A that permitted the successful completion of your research project which otherwise would not have been achieved. Person A did not work on the project beyond that suggestion. What credit would you give to that person in your publication?

As shown in Table 1, seven different scenarios were examined for Question 23 (Q23a–Q23g). In these seven scenarios, the substantive contribution of Person A to the respondent's research project is exactly the same, that being “a suggestion … that permitted the successful completion of your research project which otherwise would not have been achieved.” The seven scenarios differed by (a) the relationship between the respondent and Person A; and (b) the setting in which the essential intellectual contribution was made. The relationship of Person A to the academic was one of four “types:” the respondents' own graduate student or postdoctoral student; another student in the academic's department; a colleague in the same department; and a colleague from another institution. The settings included the respondent's research group meeting, a departmental seminar, an informal setting such as in an office, and an e-mail or letter. Question 23 was designed to quantify, evaluate, and understand the context-dependency in giving credit, if such a context-dependency was revealed. Providing an acknowledgment was the most prominent response for the seven scenarios in Q23 (69–81% of the responses; see Table 1 for more details).

Table 1: Data describing the respondents' policies regarding giving credit to others

Q23. A Suggestion was made by Person A that permitted the successful completion of your research project which otherwise would not have been achieved. Person A did not work on the project beyond that suggestion. What credit would you give to that person in your publication?
	Scenario		Number of responses ^a (%)
Person A	Setting	Question number	Do nothing	Provide an acknowledgment	Provide co-authorship
Your graduate student or postdoctoral student who was not working on the project	At your research group meeting	Q23a	48 (8.5%)	393 (69%)	126 (22%)
A student in your department	At a department seminar	Q23b	101 (18%)	410 (72%)	56 (10%)
A colleague in your department	In an informal discussion (e.g., the hall or an office)	Q23c	46 (8.1%)	446 (79%)	75 (13%)
	At a department seminar	Q23d	86 (15%)	422 (74%)	59 (10%)
A colleague from outside your department	In his/her office or your office	Q23e	36 (6.3%)	460 (81%)	71 (13%)
	At a seminar	Q23f	89 (16%)	423 (75%)	55 (10%)
	Letter or e-mail sent to you	Q23g	15 (2.6%)	462 (81%)	90 (16%)
^aThirty-three respondents provided “yes” responses to two or three alternatives in at least one row. The numbers and percentages in this table reflect only the 567 respondents who provided a single response in each row. See text for more details.

To be certain that our initial analyses were not based on “averaged” data and to look for underlying trends, we examined the responses to Q23 in several ways. Table 2 and Table 3 illustrate the context-dependency found in giving credit. These two tables summarize the data for the following types of questions. For Table 2, what credit would respondents give to students in their department if they would give co-authorship to their own student for the identical substantive contribution? If they would give an acknowledgment to their student? If they were to give “nothing” to their student?

Table 2: ^{a, b} Survey responses to Q23: A suggestion was made by Person A that permitted the successful completion of your research project which otherwise would not have been achieved. Person A did not work on the project beyond that suggestion. What credit would you give to that person in your publication? Two scenarios were presented in the sequence shown

First (or control) scenario		Second scenario
If Person A were your own student at your research group meeting, you would grant		If Person A were a student in your department at a department seminar, you would grant
		Response [Within each row, n (%)]
Response	Respondents (n)	Co-authorship	Acknowledgment	Nothing
Co-authorship	126	55 (44%)	64 (51%)	7 (5.6%)
Acknowledgment	393	1 (.3%)	335 (85%)	57 (15%)
Nothing	48	0 (0%)	11 (23%)	37 (77%)
Total	567	56	410	101
^aOf the 600 survey respondents, 567 selected only one answer (co-authorship, acknowledgment, or nothing for each scenario. See text for additional discussion.
^bThis table is read in the following fashion: Consider the first row of this table only. Of the 600 respondents to the survey, 126 said they would, under the first stipulated situation, grant their own graduate student co-authorship. Those 126 respondents were then asked what they would do under a second scenario, when Person A was a student in their department other than their own graduate student. Of those 126 respondents, 55 would grant someone else's student co-authorship, 64 would grant someone else's student an acknowledgment, and seven would provide no recognition to someone else's student.

Table 3: ^{a, b} Survey responses to the following two scenarios: A suggestion was made by Person A that permitted the successful completion of your research project which otherwise would not have been achieved. Person A did not work on the project beyond that suggestion. What credit would you give to that person in your publication?

First (or control) scenario		Second scenario
If Person A were a colleague in your department at a departmental seminar, you would grant		If Person A were a colleague from outside your department at a seminar, you would grant
		Response [Within each row, n (%)]
Response	Respondents(n)	Co-authorship	Acknowledgment	Nothing
Co-authorship	59	52 (88%)	6 (10%)	1 (1.7 %)
Acknowledgment	422	3 (.7%)	407 (96%)	12 (2.8%)
Nothing	86	0 (0%)	10 (12%)	76 (88%)
Total	567	55	423	89
^aSee footnotes a and b in Table 2.
^bThis table is read in the following fashion: Consider the first row of this table only. Of the 600 respondents to the survey, 59 said they would, under the first stipulated situation, grant co-authorship to a colleague in their own department. Those 59 respondents were then asked what they would do under a second scenario, when Person A was a colleague in another department. Of those 59 respondents, 52 would grant the other faculty member co-authorship, six would grant an acknowledgment, and one would provide no recognition.

Table 3 compares credit given (or not given) for a suggestion made during a departmental seminar. What credit would respondents give to a colleague from outside their department for a substantive, critical research suggestion if they would give co-authorship to their departmental colleague for the identical substantive contribution? If they would give an acknowledgment to their departmental colleague? If they were to give “nothing” to their departmental colleague? Consider the “co-authorship” row of Table 3. Of the 59 respondents who would give co-authorship to their own departmental colleague, one would give nothing, and six would give an acknowledgment to an outside departmental colleague. Consider the “nothing” row of Table 3. Of the 86 respondents would give nothing to their own departmental colleague, 10 would acknowledge the contribution of their outside departmental colleague.

Table 4 lists the response trends observed for Q23. Each row in Table 4 represents a unique response profile. For example, row 1—the response by 307 academics—is an acknowledgment for each of the Q23 scenarios (Q23a–Q23g). Twelve unique response profiles (rows 1–12) represent the responses of 475 of the academics. Row 17 indicates that each of 66 respondents provided a unique response profile. The shaded rows in Table 4 represent all “acknowledgements”, all “nothing”, and all “co-authorship” in rows 1, 2, and 7 respectively. Table 5 provides an indication as to the uniformity of the response profiles. The number of respondents who were within one (or within two) responses of a uniform response to the seven scenarios in Q23 are listed in Table 5.

Table 4: Unique responses to Q23a–Q23g, where the lower case letter “a” refers to the first scenario in Table 1, “b” is the second scenario, etc. The letter “N” refers to “Do Nothing”, “Ack” refers to “Acknowledgment,” and “Co-A” refers to “Provide Co-authorship.” An entry of “0” is a “no” response, and an entry of “1” is a “yes” response

	Responses to the seven scenarios (A–G) in Q23																					Total respondents
	Q23a Your student			Q23b Another's student			Q23c Your colleague, informal discussion			Q23d Your colleague, seminar			Q23e Outside colleague, informal discussion			Q23f Outside colleague, seminar			Q23g Outside colleague, letter or e-mail			For each unique response	Cumulative	Cumulative total (%)
Rows	N	Ack	Co-A	N	Ack	Co-A	N	Ack	Co-A	N	Ack	Co-A	N	Ack	Co-A	N	Ack	Co-A	N	Ack	Co-A
1	0	1	0	0	1	0	0	1	0	0	1	0	0	1	0	0	1	0	0	1	0	307	307	51%
2	1	0	0	1	0	0	1	0	0	1	0	0	1	0	0	1	0	0	1	0	0	47	354	59%
3	1	0	0	0	1	0	0	1	0	0	1	0	0	1	0	0	1	0	0	1	0	36	390	65%
4	0	1	0	0	0	1	0	1	0	0	0	1	0	1	0	0	0	1	0	1	0	23	413	69%
5	0	0	1	0	0	1	0	0	1	0	0	1	0	0	1	0	0	1	0	1	0	10	423	71%
6	0	1	0	0	0	1	0	1	0	0	1	0	0	1	0	0	1	0	0	1	0	9	432	72%
7	0	0	1	0	0	1	0	0	1	0	0	1	0	0	1	0	0	1	0	0	1	9	441	74%
8	0	0	1	0	1	0	0	1	0	0	1	0	0	1	0	0	1	0	0	1	0	8	449	75%
9	0	1	0	0	1	0	0	1	0	0	1	0	0	1	0	0	1	0	1	0	0	7	456	76%
10	1	0	0	0	1	0	1	0	0	0	1	0	1	0	0	0	1	0	1	0	0	7	463	77%
11	1	0	0	0	1	0	0	1	0	0	1	0	0	1	0	0	1	0	1	0	0	6	469	78%
12	0	1	0	0	0	1	0	1	0	0	0	1	0	1	0	0	1	0	0	1	0	6	475	79%
13	Three unique distributions were reported by five respondents each.																					5	490	82%
14	Two unique distributions were reported by four respondents each.																					4	498	83%
15	Six unique distributions were reported by three respondents each.																					3	516	86%
16	Nine unique distributions were reported by two respondents each.																					2	534	89%
17	Sixty six unique distributions were reported by one respondent each.																					1	600	100%

Table 5: Frequency of respondents who provide the exact same response for seven (all) or six or five of the seven scenarios in Q23

	All seven scenarios	Six of the seven	Five of the seven
No credit	9	12	11
Acknowledgment	307	67	31
Co-authorship	47	4	5
Totals	353	83	47
Cumulative totals	353	436	483
All counts are only of those participants who did not have any multiple answers in any of the Q23 questions.

To understand what is driving the responses to Q23, the respondents were placed into artificially though rationally constructed groups based on their answers to all of the Q23 questions. The procedure was as follows. The answers to all Q23 questions were summed so that each respondent had three digits that represented their responses, for instance (0, 3, 4) or (0, 7, 0). The first digit is the sum of all of their answers for “no credit given,” the second digit is the sum of all their answers for “give an acknowledgement,” and the third digit is the sum of all their answers for “co-authorship.” Examination of the distribution of the responses to Q23 led to nine groupings as shown in Table 6. These nine groups (Group 1–Group 9) were then characterized according to what they represent and then used as independent variables in correlations with the academic background factors and other data within the survey dataset. Table 7 illustrates the relationship between Groups 1–9 and the year the recipients received their Ph.D. Table 8 illustrates the relationship between the Groups and the total career publications. Other relationships were also examined (see the Discussion section).

Table 6: Grouping of respondents to their responses to the seven Q23 scenarios, namely Q23a–Q23g (see Table 1)

Group	Response characteristics	Sum of responses for each category			Responses
		No credit	Acknowledment	Co-authorship	Frequency	Percent
1	Co-authorship primarily	0	0, 1, or 2	7, 6, or 5, respectively	57	9.5%
2	Co-authorship and acknowledgement split	0	3 or 4	4 or 3, respectively	19	3.2%
3	Co-authorship but biased to acknowledgement	0	5 or 6	2 or 1, respectively	56	9.3%
4	Always acknowledgement	0	7	0	307	51.2%
5	Nothing but biased to acknowledgement	1 or 2	6 or 5,respectively	0	40	6.7%
6	Nothing and acknowledgement split	3 or 4	4 or 3, respectively	0	44	7.3%
7	Nothing primarily	5, 6, or 7	2, 1, or 0, respectively	0	32	5.3%
8	Situational–had at least one choice in all columns	≥ 1	≥ 1	≥ 1	12	2.0%
9	Multiple responses for one or more scenarios	The sum of responses ≥ 8			33	5.5%

Table 7: Number of respondents in groups 1–9 (see Table 6) as a function of decade of receipt of their Ph.D. degree (Q48)

Decade receiving Ph.D.	Group 1	Group2	Group3	Group4	Group5	Group6	Group7	Group8	Group9	Total
	Number of faculty respondents (n)
1940s	1	0	0	0	0	0	0	0	0	1
1950s	13	0	0	0	0	0	0	0	0	13
1960s	20	17	37	0	0	0	0	0	2	76
1970s	0	0	3	102	0	0	0	0	1	106
1980s	0	0	0	109	0	0	0	0	0	109
1990s	0	0	0	94	39	0	0	8	14	155
2000s	0	0	0	0	1	44	32	4	16	97
Total	34	17	40	305	40	44	32	12	33	557
Number of emeritus faculty respondents (n)
1940s	3	0	0	0	0	0	0	0	0	3
1950s	17	0	0	0	0	0	0	0	0	17
1960s	3	2	16	0	0	0	0	0	0	21
1970s	0	0	0	2	0	0	0	0	0	2
Total	23	2	16	2	0	0	0	0	0	43

Table 8: Number of career publications as a function of Q23 groups (see Table 6)

	Q23 Groups									Total
Total number of publications ^a	1	2	3	4	5	6	7	8	9
0 to 10	0	0	1	2	3	6	8	2	6	28
11 to 20	0	0	1	21	3	19	11	2	6	63
21 to 30	0	1	2	22	7	7	8	1	4	52
31 to 40	1	0	1	26	5	7	2	4	7	53
41 to 50	3	0	4	50	13	2	0	1	3	76
61 to 80	3	0	3	48	5	3	2	1	2	67
81 to 100	3	1	8	23	3	0	0	0	0	38
101 to 150	10	7	11	45	0	0	0	1	1	75
151 to 200	5	3	7	38	0	0	0	0	0	53
200 plus	31	7	18	30	0	0	0	0	3	89
Total	56	19	56	305	39	44	31	12	32	594
^aSix of the 600 respondents did not provide an answer the ‘total number of publications’ question.

DISCUSSION

For the 600 respondents as a whole, providing an acknowledgment was the most prominent response for the seven scenarios in Question 23, i.e., when an individual makes a suggestion that permits the successful completion of the respondents' research project which otherwise would not have been achieved. That person does no work on the project beyond providing that essential suggestion. In today's competitive environment of science, providing an acknowledgment simultaneously acknowledges another's contributions without sharing much, if any, of the credit. An acknowledgment is “present, tangible, and real” but simultaneously is essentially transparent. An acknowledgment is only seen if one goes to the acknowledgment section of a paper. Acknowledgments are neither indexed nor quantified in any way in the literature or abstracting services. Curricula vitae do not include a section listing acknowledgments. Thus, an acknowledgment is far distant from being a co-author. That is, on some “credit axis,” doing nothing and giving an acknowledgment are closer to each other than is authorship.

Thus, a senior author giving an acknowledgment to a colleague or student depletes one's own status little but has several potentially important consequences: the senior author is able to say that “appropriate credit was given” and thereby may avoid an allegation of plagiarism. An assertion of plagiarism by an insufficiently-credited individual will typically fall on deaf ears. Except in highly visible (and extremely rare) instances, the scientific community is hardly responsive to the apparent “whining” of others. There are few effective processes that address such problems before an allegation of wrongdoing transitions into a serious charge of plagiarism, one of the forms of misconduct of research.

Some interesting trends are seen by examining columns four and six in Table 1, namely, the “do nothing” and “co-authorship” columns. Academics are most likely to give co-authorship to one's own student. Academics are least likely to credit a colleague's student or a colleague inside or outside their department. Giving a suggestion at a seminar is the least likely mode of intellectual contribution to be rewarded. This may be because the very purpose of a seminar-–as well as a research group meeting—is the transmission of results by open communication and idea-sharing. Nonetheless, the rules of plagiarism still obtain. An idea is still the property of the conceiver. Of course, multiple simultaneous discoveries in science are well known (Merton, 1973), and the recipient of an idea may well have had that idea already. It is sometimes awkward to say, “I have already thought of that.” There is clear tension between free discussion of ideas, ownership of ideas, and the perception of intellectual thievery.

As shown in Table 1, academics appear to be more likely to give co-authorship when a specific proactive, substantive action is undertaken by the idea generator, e.g., when a letter or e-mail containing the suggestion is sent. In addition, though this may not play a conscious role in a senior author's behavior toward scientific publication, a written suggestion may carry with it legal ramifications, i.e., it may be a contemporaneous, tangible, documented and dated record of inventorship. Suggestions received during an informal interaction, especially as part of a seminar, usually do not lead to co-authorships. Seminars may be viewed as “group think” experiences, a group activity of sharing ideas freely, openly, and collegially without credit responsibilities. Lack of providing credit for suggestions given during a seminar could also be a memory effect. That is, speakers rarely take notes of questions posed and comments made during their presentations.

There is a clear inhomogeneity in this data set. The dissimilarity of responses (“nothing” versus “acknowledgment” versus “co-author”) indicates that the nature of the relationship between respondent and Person A influences the behavior of many respondents. That is, many respondents provide credit to Person A based on Person A's relationship with the respondent rather than entirely based on the actual substance of the interaction.

The constancy in each individual respondent's behavior to the exact same assistance but by different individuals and/or in different contexts was next evaluated (Table 2 and Table 3). For example, would individual respondents give more or less credit to their own students compared with their colleague's students given identical scientific contributions (Q23a versus Q23b)? Consider the data in the first row of Table 2. Each individual surveyed was asked what form of credit each would give to their own student-–co-authorship, acknowledgment, or nothing—if that student made a suggestion that permitted the successful completion of the professor's research project which otherwise would not have been achieved. One hundred twenty six of the 600 respondents said they would grant co-authorship to their own student. Those 126 respondents were then asked what credit they would grant a student under the exact same scenario except that that individual was a student of another professor. Only 55 of those 126 (44%) would grant someone else's student co-authorship. Sixty four of those professors would grant someone else's student an acknowledgment, and seven would provide no recognition at all to someone else's student—even though, under the exact conditions, those professors would grant their own student co-authorship. The same type of downgrading in credit (moving from co-authorship to acknowledgment or no credit whatsoever; or moving from giving an acknowledgement to giving no credit) is also observed for those academics who would give their own student an acknowledgment (n = 393). Of those 393 academics, 335 would provide another professor's student an acknowledgment, while 57 would not credit the other professor's student at all in the manuscript.

The most obvious reason to give more credit to one's own student than another faculty member's student is academic favoritism perhaps stemming from a greater awareness of one's own students accomplishments, capabilities, needs or, simply, closer association and favoritism. Favoritism constitutes testimony to the family-like feelings in the group. Alternatively, departmental politics may play a role: providing a colleague's student with authorship may entail—according to the ACS rules of authorship (American Chemical Society, 2006) and the International Committee of Medical Journal Editors (International Committee of Medical Journal Editors, 2010)—shared responsibility and accountability for the results. This could require a research interaction with the student outside that student's formal relationship with the other colleague. A “research triangle” could then result leading to conflict in interest issues. A faculty member may simple shy away from a potentially messy situation by not providing co-authorship to another faculty member's student.

At the other extreme, of the 48 professors who would give their own student no credit, neither co-authorship nor an acknowledgment, 11 of them (23%) would give another professor's student an acknowledgment. This is an example of upgrading in credit (a move from giving no credit to either an acknowledgement or co-authorship; or a move from giving an acknowledgement to giving co-authorship). Unfortunately, those who upgraded credit were so rare that the numbers were not sufficient to test for academic background factor relationships. Further, in the first scenario (Table 3), Person A is a colleague in the respondent's own department. In the second scenario, Person A is a colleague from another university. While there is some degree of downgrading (data row 1; 10%) and upgrading (data row 3; 12%), 88%–96% of the respondents would give the same degree of credit regardless of whether Person A is a faculty member from their own institution or another institution.

We tested for a difference in basic academic background factors for those who downgraded credit. There are two statistically significant related variables: field of expertise (chi-square test; p = .005) and receiving a Ph.D. in the United States (chi-square test; p = .029). There are some striking differences for the field of expertise variable. Academics who stated their primary field of expertise was organic chemistry were much less likely to downgrade credit, while those in polymer chemistry were much more likely to downgrade credit. Respondents who received their Ph.D. degree at a United States institution or had less than 10 single-authored publications were less likely to downgrade credit, but those receiving the Ph.D. degree outside the United States or those having 11 to 30 single-authored publications were more likely to downgrade.

We next evaluated if there were any “favored” response trends in the respondents' responses to these seven scenarios. Since each of the seven scenarios for Q23 shown in Table 1 has three possible responses, there are 2,187 (= 3⁷) unique combinations of answers for Q23a–Q23g (for those individuals providing a single response per Q23 row). We asked to what extent are there several unique response patterns that occur much more than others? Are there groups of individuals who generally give co-authorship or acknowledgements no matter what the situation or person involved? Are there groups that lean more towards giving no recognition?

As shown in Table 4, row 1 represents the most frequent response, selected by essentially 50% of the respondents. These respondents would provide an acknowledgement for all Q23a–Q23g scenarios. Row 2 (all “nothing,” 8% of the respondents) and row 7 (all co-authorship, 2% of the respondents) also represent uniformity of responses within a single credit modality. Three responses (rows 2, 3, and 4) each garnered from 4– 8% of the respondents. One hundred and seventy seven respondents (rows 5–17) fall into 94 unique response patterns.

Table 5 shows the number of respondents who were within one or within two responses of a uniform response to the seven scenarios in Q23. To illustrate the reading of Table 5, 307 respondents said that they would provide an acknowledgment for all seven scenarios. Sixty-seven other respondents said they would provide an acknowledgment in six of the seven scenarios in Q23, and 31 respondents said they would provide an acknowledgment in five of the seven Q23 scenarios. Table 4 and Table 5 together demonstrate an important observation from the data set: there is the wide diversity of responses for the respondents.

To more fully characterize the actual and predicted behavior of the respondents to giving credit, each respondent was placed into one of nine groups based on their answers to all of the Q23 questions. The grouping procedure is fully described in the Results section. As can be seen in Table 6, there is some symmetry in the definition of these groups: Group 1 and Group 7 (either co-authorship or nothing, one extreme or the other); Group 2 and Group 6; Group 3 and Group 5. Group 4 is always an acknowledgment and can be considered a “mean” response. Group 8 (situational) involves respondents who marked at least once in each column. We interpret “situational” as either (a) a real sensitivity to differences in each of the seven situations posed; these individuals say that they can respond in various directions; or (b) the respondents could be individuals who respond to the immediate situation without firm preset principles or ideas or past experience which they then formulate into rules. Group 9 is comprised of the 33 individuals who responded, at least once, to more than one option to a specific scenario.

Which, if any, of the variables in the survey, i.e., the responses to the other questions, are related to the nine Groups and distinguish one Group from the others? Group membership was used as a binary dependent variable, and the significant relationships between each Group and other variables in the survey data were determined. In total, 62 variables were examined for their contribution towards membership in each Group. Appropriate tests were used in accordance with the different types of variables. Each test provides a probability that the relationship is not spurious. Two of the variables were significant for all nine Groups, namely, “year of receipt of the Ph.D.” (Q48) and “total number of lifetime publications” (Q50e). The relationships between these two variables and Groups are detailed in Table 7 and Table 8, respectively. Indeed, these two tables illustrate some rather dramatic year of Ph.D.–credit relationships.

As shown in Table 7, professors receiving their Ph.D. degrees in the 1940s, 1950s, and 1960s are the only respondents to grant co-authorship (Group 1) or co-authorship and acknowledgement split (Group 2). Analogous results were obtained from the emeritus faculty data who received their Ph.D. degree mostly in the 1950s. The only other group that contains co-authorship choices (Group 3) consists of 37 respondents who received their Ph.D. degrees in the 1960s and three from the 1970s. The other 1970s graduates (102 of the 103 remaining) gave not a single co-authorship but only acknowledgments. Every single respondent who received the Ph.D. in the 1980s, all 109 of them, gave only acknowledgments (Group 4 only). Ph.D. recipients from the 1990s were mostly either all acknowledgment or biased toward acknowledgment with some “no credit.” Seventy six of the 97 faculty who received their Ph.D. in the 2000s gave primarily no credit or some acknowledgment (Group 7 and Group 8). Very clearly, there is a dramatic trend toward giving or not giving credit: the eldest faculty is more likely to give credit; the youngest faculty hoards credit; and those in between straddle both behaviors. The shaded cells in Table 7 illustrate this age-credit trend.

We emphasize one further consideration in this analysis. Respondents who fall into Group 8 selected at least one of each of the possible responses, “nothing”, “acknowledgment”, and “co-authorship” in at least one of the seven Q23a–Q23g scenarios. In addition, Group 9 respondents selected multiple responses for one or more of the seven Q23 scenarios. Of the two of the 1960's-Ph.D. respondents and one of the 1970's-Ph.D. respondents who are in Group 9, one respondent in each bracket chose “nothing”. Of the 14 1990's-Ph.D. respondents and the 16 2000's-Ph.D. respondents who are in Group 9, three and four chose “co-authorship” at least once, respectively.

We speculate that some senior faculty members will anticipate that the credit is still going to be assigned to themselves, regardless of the number of other unknown junior authors and regardless of the order of the names on the paper. This conclusion brings to mind the so-called “Matthew effect” and may well be a corollary to it. According to Merton, “The Matthew effect consists of the accruing of greater increments of recognition for particular scientific contributions to scientists of considerable repute and the withholding of such recognition from scientists who have not yet made their mark” (page 446, in particular, and for more details, Chapter 20 in Merton, 1973). We cite a corollary to the Matthew effect, namely, that senior authors may not consider it much of a credit sacrifice to add a name or several names—or even adding “starred authors”—to their articles when it is the senior authors who are so well known that credit will flow automatically and primarily if not solely to them. To the extent that this corollary to the Matthew effect is operating within the academic chemical community in the United States, it must only be doing so for those academics who are actually giving credit more frequently. The younger respondents, as shown in Table 7, are most certainly not doing so.

All of Group 8 respondents (“situational,” having at least once choice from each credit possibility) received their Ph.D. in either the 1990s or 2000s. And 30 of 33 respondents who were in Group 9 (provided multiple responses for at least one of the scenarios in Q23) also received their Ph.D. in either the 1990s or 2000s. Apparently, a segment of the younger faculty members have less firm models of credit, and some have a more flexible style than their elder colleagues. These younger faculty members may also have trouble anticipating an exclusive response, perhaps because of their lack of experience in being senior authors. We emphasize, however, that the vast percent of the youngest respondents provided little additional credit to others in the Q23 scenarios. That the eldest cohorts are the most likely to give credit disagrees with Robinson et al. who concluded that “mid-to-late career scientists may still be evaluated on the quantity of publications for promotion or maintaining professional standing … these individuals may feel pressured to pursue authorship without having substantially contributed to the work” (Robinson et al., 1999).

Table 8 shows the number of total career publications by the Groups. In the section above, we showed that the responses to Q48 (decade of receipt of the Ph.D.) and Q50e (number of publications) are highly and positively correlated, reflecting the same fundamental characteristics of the respondents. Hence, it is expected that Table 7 and Table 8 will reveal similar characteristics, and they do. We shall not walk through Table 8 as we did for Table 7. The data indicates quite clearly that with time, age, and productivity, the model shifts from one likely to give credit to others (with high number of publications) to one unlikely to give credit to others (with low number of publications). As in Table 7, the shaded cells in Table 8 illustrate this age-credit trend. This observation can be explained by taking into account the pressure that academics feel to publish to obtain tenure, grants, and professional recognition at the earliest stages of an academic's career.

Several other notable relationships were observed between Group membership and behavior. Question 1, Have you had problems with credit practices? is only related to membership in Group 5 (“acknowledgment but biased to nothing”). This result indicates that the perception of not getting appropriate credit in the past does not dictate how the respondents act regarding giving credit to others. However, for the subgroup of respondents who indicated they had a problem with their professor or teacher (Q2), five statistically significant relationships were observed. Respondents who reported having had a problem with their professor or teacher were more likely to give nothing and less likely to give co-authorship (Group 1) (chi-square test; p = .015). But such a strong relationship between Group and Q9 (problem with a “departmental colleague” was not found nor was a problem with a “non-departmental colleague.” However, respondents who reported having had a problem with their colleagues in another institution are more likely to globally give co-authorship (chi-square test; p = .09). Group 7 (“giving no credit”) was not statistically related to problems with an inside colleague or with an outside colleague but was found with a problem with one's professor or teacher. It is reasonable that experiences during one's education are more influential than professional experiences than experiences later in one's professional career.

The chemistry discipline (Q47) showed significant differences in having authors in the first two Groups (heavily biased towards co-authorship). Physical chemists, theoretical chemists, and “other” (subdisciplines of chemistry not specified in the survey) were the most likely to give credit in terms of providing co-authorship and acknowledgments. The least likely to give credit were analytical chemists, bio/medicinal chemists, organic chemists, and inorganic chemists. The maximum percentage of respondents by discipline in Group 9 (multiple responses) belonged to polymer chemists.

Group 1 (“co-authorship”) and Group 4 (“acknowledgment”) respondents were more likely to have asked to be removed from authorship (Q33; see the following section for a fuller discussion of this topic) (chi-square test; p = .005 and .073, respectively). Group 7 individuals (“give no credit”) were inversely correlated to being asked to have their name removed from authorship (chi-square test; p = .006). These relationships indicate that if asked to be removed from authorship, an individual would have some sense of what is correct and apply that sense to giving/not giving credit. In addition, respondents who do not give credit are likely not to ask to have their name removed from authorship and visa versa, a consistent finding for a person making decisions primarily on a self-serving basis.

These results speak to both context-dependence and publication practices which have some relationship with age of the respondent. None of the codes of conduct dealing with authorship that we are aware of, and there are many, provide standards or guidance regarding authorship that are context-dependent, i.e., the relationship between the person providing a service and the researcher using that service or the context in which the service is provided. Thus, the decision to provide co-authorship—or not—to an individual should be, according to the codes of conduct, independent of an individual's relationship with the researcher or the research project. That the respondents in this survey would provide a range of credit behaviors based on the seven scenarios in Table 1 is inconsistent with the various standards of responsible conduct of research.

One unambiguous relationship difference in terms of professional status, experience, and power is the professor–student pair. Should students be considered any different than professional colleagues with regard to the rules of authorship (Fine and Kurdek, 1993)? A recent survey showed that 50% of NIH-funded postdoctoral F32 fellows were either unaware of or did not refer to RCR guidelines dealing with authorship (Barrett et al., 2005). Fairness also speaks to conduct that is “giver-context neutral.” Of course, even within a faculty member's own research group, not all students are considered or treated equally, as discussed by Sullivan and Ogloff (1998). Indeed, Sullivan and Ogloff suggest that “it seems more fair to err on the side of the student (who is in the position of less power) when the contributions of supervisor and student are approximately equal” (Sullivan and Ogloff, 1998, p. 234). Fine and Kurdek suggest that, given “that students have less power and competence than non-student collaborators, then justice would be served by giving students differential treatment (Fine and Kurdek, 1993, p. 1143),” i.e., be more likely to give credit to students than the standards of authorship would demand. This suggestion further places authorship issues into the realm of subjectivity.

Except for the student–faculty relationship, context-dependency of RCR rules is not much discussed in the literature. Garrett and Bird point out that the quality of communication among scientists must be uniformly high regardless of the context: “informally in the elevator or over lunch, or in a formal paper or conference presentation” (Garrett and Bird, 2000, p.436). Several studies have shown that large percentage of co-authors fail to meet the authorship standards of their community (Bates et al., 2004; Flanagin et al., 1998; Goodman, 1994), though the underlying reasons for this failure were not determined. A survey by Vernon et al. suggest that “some of [their] respondents saw any criteria as guidelines rather than rules and wanted to retain the power to interpret them” (Vernon et al., 1997).

In this survey, Q23 was intentionally designed to provide a hypothetical situation in which there was simultaneously (a) a suggestion by an idea-provider that was essential to the successful completion of another researcher's project and (b) there was no further association by the idea-provider with the project. The Ethical Guidelines to Publication of Chemical Research of the American Chemical Society states clearly that “the co-authors of a paper should be all those persons who have made significant scientific contributions to the work reported … ” (American Chemical Society, 2006), met by hypothetical (a) above. However, the guidelines continue, “and who share responsibility and accountability for the results,” a criterion similar to the statement on Authorship and Contributorship by the International Committee of Medical Journal Editors (ICMJE) (International Committee of Medical Journal Editors, 2010) and clearly not met by hypothetical (b) above. Based on the American Chemical Society (ACS) (American Chemical Society, 2006) and ICMJE guidelines, an individual as characterized by Q23—in all of the scenarios, Q23a—Q23g—would not deserve authorship. That Q23-individual, as described, could not share responsibility and accountability for appropriate portions of the research results. In contrast, the American Astronomical Society's Ethics Statement (American Astronomical Society, 2010) indicates that this idea-provider—an individual characterized by all of the scenarios Q23a–Q23g—does deserve authorship since there are no “responsibility and accountability” criteria in AAS's standards.

Given that researchers apparently do not feel bound to-–and may not even be knowledgeable of—the standards of professional societies regarding authorship (Barrett et al., 2005; Funk et al., 2007; Seeman and House, 2010), it is logical and consistent that many of the respondents did not choose “authorship” for the Q23 scenarios. An interesting and even critical teaching stems directly from these considerations: should an individual provide critical intellectual contributions to a research project, and should the eventual senior author(s) of the publication—likely, a faculty member—envision that the idea-provider ought to be an author of the eventual publication, then the idea-provider need be brought into the research project during the course of the research, and as soon as possible, thereby establishing that the idea-provider had and took the opportunity to become accountable and responsible for the results.

The Role of Asymmetry in Credit Issues

There are many reasons that explain the large percentage of reports of credit and related RCR problems (Martinson et al., 2005; Seeman and House, In press) in the academic community in the United States. Consider the professor–student relationship. The hierarchical nature of academics causes an inherent asymmetry in terms of responsibility, authority, position, and ultimately in power (Brown-Wright et al., 1997; Lawrence, 2002; MacDonald and Williams-Jones, 2009). While many research groups today work in teams—as judged by the substantial number of multi-authored publications—the professor is typically the senior author and simultaneously a benevolent, or otherwise, dictator. It is the professor in whose laboratory the research is performed, who provides the financial underpinnings, research direction, and ultimately letters of recommendation. This asymmetry in power leads away from a democratic environment where the methods of conflict avoidance can prosper (Robinson et al., 1999; Welsh et al., 2008) and fosters a system in which arbitrary decisions such as those illustrated in Table 2 (context-dependent assignment of credit) are widespread. Lawrence summarized that “the scientific community supports the natural tendency of the experienced to take advantage of the inexperienced” (Lawrence, 2002, p. 835). In such a setting, it is not surprising that inconsistent behavior toward giving credit would be the norm and that disappointment and anger would be both a short term and long term consequence. For example, “ghost authorship” can have opposite representations. A student who actually contributed and deserves authorship may not be so credited, one form of ghost authorship. Alternatively, as discussed above, a student who did not meet the criteria for authorship may be given co-authorship for various reasons, e.g., to conscious or unconscious favoritism or a desire by the faculty member to help the student obtain a job.

Fine and Kurdek “recommended that both faculty and students participate in the authorship decision-making process early in the collaborative endeavor” (Fine and Kurdek, 1993, p. 1141). “Participation” can have various meanings, all of which can include open dialogue and mutual sharing of values. In one extreme, “participation” can mean a sharing of the research experience by student with the professor combined with the faculty member's dominance regarding policy. In this form of participation referred to as “parentalism” by Fine (Fine and Kurdek, 1993, p. 1143), the faculty member makes the decision when the work is ready for publication while the student performs the research without knowing, for certain, who will be authors and in what order the authors' names will be listed. “Participation” alternatively can entail a more democratic process among non-equals in which published authorship standards are agreed upon and openly discussed, best done in advance of the research being performed (Robinson et al., 1999). Many professors prefer the role of benign dictator and may be unwilling to encourage a discussion on authorship in a pseudo-equal hierarchy. In all cases, faculty behavior becomes a role model for the students' ultimate behavior when they become independent scientists. To the extent that education is intended to prepare one for future work experiences, the asymmetry in the work environment surely finds a parallel with that in academia.

Both “guest authorship” and “ghost authorship” occur. One of the survey questions probed whether the respondents believe that they have not gotten appropriate credit for work performed (“ghost authorship”); 50% responded in the affirmative (Seeman and House, In press). Alternatively, it may be the faculty member whose name appears on a publication without deserving authorship (“guest authorship”) (Fine and Kurdek, 1993), or the faculty member actually does deserve authorship based on one of the published standards but not in the perception of the student. In other instances, students completing their Ph.D. or postdoctoral students are competing for positions and research funding and request authorship on publications in which they may not meet the standards—but will receive authorship due to the enthusiastic—or pseudo-obligatory—support of their mentor. Indeed, the drive by some faculty members to promote their students from their (i.e., the faculty member's) own ego and reputation may well lead to a conflict in interest situation which may promote instances of ghost authorship for the student who may benefit from an expanded, if not deserved, curriculum vitae.

Calls for faculty to be vigilant against their own possible misuse of this asymmetry by, for example, performing a regular “explicit, written self-audit” (MacDonald and Williams-Jones, 2009, p. 122) may well be overly optimistic though certainly well meaning. In fact, Woodward and Goodstein have argued quite forcefully and cogently that some “plausible-sounding rules for defining ethical conduct might be destructive to the aims of scientific inquiry” (Woodward and Goodstein, 1996). For example, these authors suggested that certain principles “might be damaging or unworkable according to our analysis of how science works [including, the principle …] The choice and order of authors on a multiple-author publication must strictly reflect the contributions of the authors to the work in question” (Woodward and Goodstein, 1996, pp. 479–480). This particular conclusion by Woodward and Goodstein runs counter to all of the guidelines or codes of conduct dealing with authorship that we have seen. It may be that Woodward and Goodstein are considering the balance or tension between scientific productivity versus a healthy environment in which to do science. Indeed, Woodward and Goodstein conclude that “the practice may be functionally useful and may involve a little deception, since conventions regarding authorship may be well understood by those who participate in a given area of science” (Woodward and Goodstein, 1996, p. 487). Surely, the asymmetry in power between faculty and students is well understood by all involved—indeed, all faculty were, at one time, students themselves and thus susceptible to the same power—asymmetry with their own professors. Such knowledge, often gained by personal experience, does not make it fair. Fairness and related issues dealing with faculty-student relationships was discussed qualitatively by Sullivan and Ogloff (Sullivan and Ogloff, 1998).

Brown-Wright et al. conducted a survey of faculty and graduate students at a single institution and compared the roles and expectations of both with regard to student ability and training needs as well as criteria for co-authorship. About 50% of each group agreed that a graduate assistant (GA) ought to be a co-author of a publication “if the GA assists in typing, proofreading, the literature search, and/or coding of data” (Brown-Wright et al., 1997, p. 413). These specified support activities do not meet any of the currently available authorship guidelines (see, for example, American Chemical Society, 2006; International Committee of Medical Journal Editors, 2010), consistent with an environment ripe for contention and disharmony, as our survey results indicate (see also Seeman and House, In press).

Brown-Wright et al.'s study also led to the recommendation of formal training in various forms of the students. Indeed, this may be a case of “the blind leading the blind.” Formal training by all involved must also be preceded by agreement on the standards and norms of concern within the community. Indeed, the trainers themselves need to be (better) trained (Wilcox, 1998). On the other hand, Funk et al. conducted a survey of NIH postdoctoral trainees and found that education in and awareness of RCR did not increase the level of appropriate behavioral responses examined in their survey (Funk et al., 2007). Awareness of publication guidelines is another serious issue (Barrett et al., 2005). We also question whether faculty members, especially younger faculty members, will be willing to have their research groups operate in a more democratic fashion regarding policy matters.

Asymmetry can be found in professor–professor relationships as well, though this is an area not as well studied as the professor–student relationship. Asymmetry can be created by professional standing either formal, e.g., tenured versus untenured at the same institution, or less formal, e.g., reputation or standing within the community. Competition for resources or competition for recognition as well as personality characteristics such as ego and self-confidence can become behavioral driving factors. Open communication, so very critical to conflict resolution as well as conflict minimization (Robinson et al., 1999; Welsh et al., 2008), may be difficult in a milieu of self-interests and territorialism, and nearly impossible when physical distance is present. With the above factors plus cultural differences, hectic travel schedules, poor communication skills, and non-existent conflict resolution processes, even faculty within the same department may not build sufficiently strong relationships to prevent professional clashes. Indeed, the lack of conflict resolution skills and few models for productive negotiations within an academic environment based on tenure and independence, faculty members “playing well” with one another is not a well-honed capability.

It is tempting to speculate about the role of culture in authorship issues just as the role of culture in research misconduct has been explored by Davis (Davis, 2003). Hierarchical structures in academic chemistry make for a much larger asymmetry in the relationships between faculty and students in some countries relative to others. Our model would predict a range of credit problems as a function of these hierarchical structures. This would, indeed, form a testable hypothesis for a future study. Confounding factors would include the degree to which students accept asymmetry as a natural fact of their culture without attendant resentment. Indeed, the transformation into a more democratic academic culture may not proceed or be experienced in an orderly and parallel fashion by the different populations within the scientific community. It would be such inhomogeneity that could foster the kind of resentment that is evidenced in this study. Furthermore, the experiences in one country or in one culture may not be superimposable onto others.

With (1) increased attention, as discussed above, to the ethics of authorship, (2) the suggestion (Garfield, 1982; Rennie et al., 1997; Resnik, 1997) and now reality (International Committee of Medical Journal Editors, 2010) of replacing authorship credits with contributions made by “contributors” rather than by “authors,” and (3) incorporation in some journals of listings of actual contributions of each author, the number of “ghost” and “guest” authors may well decline. The dramatic relationship between respondents' age (and number of publications) and mode of giving credit suggests varying authorship standards by individuals—with increasing likelihood of giving more credit—over time.

CONCLUSIONS

A survey was conducted to examine the behavior or anticipated behavior of academic chemists in Ph.D.-granting institutions in the United States. Six hundred respondents provided information dealing with their giving, or not giving, credit to others, i.e., graduate students and colleagues, in scientific publications. A significant context-dependency was observed in the responses for the exact same substantive contributions. For example, academics are more likely to give co-authorship to one's own student rather than to another professor's student or to colleagues, whether those colleagues are within their own department or from an outside department. Downgrading credit was much less likely done by organic chemists and more likely done by polymer chemists. Giving a suggestion at a seminar is the least likely mode of intellectual contribution to be rewarded. While there were several highly populated response trends across the examined authorship scenarios, a great deal of diversity in responses was observed across the 600 respondents.

Faculty who received their Ph.D. degree in the 1940s–1960s period (and three of 106 respondents who received their Ph.D. in the 1970s) were the primary respondents who gave or reported that they would give co-authorship in the examined scenarios. Only two of the 305 respondents who received their Ph.D. in the 1940s—1980s would give no credit. Every one of the 109 faculty who received their Ph.D. in the 1980s reported giving only an acknowledgment. In contrast, faculty giving mostly no credit and some acknowledgments are populated primarily by respondents receiving their Ph.D. degree in the 1990s and 2000s. Only 19 of the 252 faculty respondents who received their Ph.D. in the 1990s and 2000s would give co-authorship. Analogous relationships with credit were found with “total number of publications,” a reasonable observation given that the two variables “year of receipt of Ph.D.” and “total number of publications” are correlated with each other.

That younger faculty will provide little credit to others, the eldest will provide the most credit, and the middle-aged group provide a mixture of credit and no-credit is a reflection of community culture, degree of already-attained (or not attained) professional positions, and internal and external pressures to produce and to be recognized, i.e., a more highly honed and required competitive spirit and behavior. That such dramatic age-related and number-of-publications related relationships are observed in survey questions dealing with hypothetical scenarios suggests a real difference in attitude and behavior among these individuals. That is, the respondents were apparently not providing some “high standard behavior pattern” as opposed to what they have done or would actually do. That today's youngest chemistry faculty members are the least likely to give credit is consistent with a highly competitive environment.

The criteria for authorship from several major professional organizations, one being the American Chemical Society, include the requirement that the individual bear responsibility and accountability for the results. This criterion indicates that an idea-provider needs to become part of the research project as early as possible, thereby providing that individual the opportunity to become responsible and accountable for at least some of the results. A senior author, knowing and accepting this “responsibility and accountability criterion,” may effectively and essentially if not consciously make an authorship decision within days if not hours after receiving an “essential” idea from a student or fellow colleague.

Finally, a very detailed discussion is provided on the asymmetry of relationships in the scientific milieu. This asymmetry can easily lead to the use and abuse of power within the realm of publication. Further, one's experiences as a graduate student may well form the model of one's behavior as a senior researcher.

ACKNOWLEDGMENTS

We thank R. Kirk Jonas, Chair of the University of Richmond Institutional Review Board, for his continuing prompt and thorough assistance. We thank Larry D. Claxton, Roald Hoffmann, Carol E. Jones, Frank Macrina, Chris McCarty, William H. Myers, Gary H. Posner, David B. Resnik, Loren Shea, Judy Stamberg, and two anonymous reviewers for very helpful comments.

Notes

Garfield, E. (1982). More on the ethics of scientific publication: Abuses of authorship attribution and citation amnesia undermine the reward system of science. Current Contents, Number 30 (July 26): 5–10; also published in Garfield, E. (1981–82). Essays of an Information Scientist, 5: 621–626.