Dearth by a Thousand Cuts? Accounting for Gender Differences in Top-Ranked Publication Rates in Social Psychology

Abstract

Publication in the Journal of Personality and Social Psychology, a flagship indicator of scientific prestige, shows dramatic gender disparities. A bibliometric analysis included yoked-control authors matched for Ph.D. prestige and cohort. Though women publish less, at slower annual rates, they are more cited in handbooks and textbooks per JPSP-article-published. No gender differences emerged on variables reflecting differential qualifications. Many factors explain gender discrepancy in productivity. Among top publishers, per-year rate and first authorship especially differ by gender; rate uniquely predicts top-male productivity, whereas career-length uniquely predicts top-female productivity. Among men, across top-publishers and controls, productivity correlates uniquely with editorial negotiating and being married. For women, no personal variables predict productivity. A separate inquiry shows tiny gender differences in acceptance rates per JPSP article submitted; discrimination would be a small-but-plausible contributor, absent independent indicators of manuscript quality. Recent productivity rates mirror earlier gender disparities, suggesting gender gaps will continue.

This volume discusses the pervasive nature of subtle bias across a variety of contexts. Given that bias against women is often subtle, perhaps it is evident even in the practices of those who are arguably at the front lines of the efforts to identify and eradicate it: social psychologists. Here, we examine potential sources of gender disparity in publication rates within our flagship empirical journal and whether gender bias may be exhibited in this group, one that would safely be considered to be against prejudice.

Women have made substantial progress in social psychology, composing the majority of undergraduate majors and even graduate students in many departments. However, at each higher level of the professional hierarchy, women are less and less represented, and cohort effects do not account for these patterns; this pattern is common to many sciences (National Academy of Sciences, 2006). For example, despite the fact that women now earn 44% of doctoral degrees in Science, Engineering, and Math fields, the percentage of women who are full professors in science and engineering has remained at 10% for the past fifty years (Rosser & Taylor, 2009). Moreover, women are less likely to be chosen as leaders than men in a variety of academic settings, despite the fact that female leaders are often as effective, or even more effective, than male leaders (Eagly, Johannesen-Schmidt, & van Engan, 2003; Rosser, 2003). Finally, women more often than men leave the field at every stage of their professional careers (National Academy of Sciences, 2006). Many factors could account for this, but one factor might be perceived scientific impact. We explore here disparities in a top-ranked publication process, as a case study of one indicator in one academic field. Scientific contributions are formally and informally measured mainly by publication rate in a field’s best empirical journals. Because publishing in the field’s flagship journal increases one’s scientific impact, it is important to consider factors that predict women’s visibility in the journal.

Though subtle bias is a global phenomenon stretching across professional domains and societies, this paper focuses on a particular context—authors published in the Journal of Personality and Social Psychology (JPSP). In a bibliometric analysis, we counted how many articles the top 30 men and top 23 women identified by Quiñones-Vidal, Lopez-García, Peñaranda-Ortega, and Tortosa-Gil (2004) as frequent JPSP authors had published in JPSP from 1965 to 2004. We noticed a stark gender difference in JPSP authorship identified but not examined by Quiñones-Vidal et al. (2004): With the exception of Shelley Taylor, there is no overlap between the men’s and women’s distributions. Note that a comparable analysis of the top 20 male and female Journal of Experimental Social Psychology authors showed the same gender divide, with men’s publications ranging from 17-11, and women’s from 8-4. Thus, we have evidence that JPSP is not unique vis-à-vis gender representation.

The current paper explores potential predictors of this gender disparity in JPSP publications. Specifically, we describe publication patterns of male and female authors since JPSP’s inception and examine the role of professional and personal variables in predicting JPSP visibility as well as other indicators of scientific impact. We stress again that this analysis is specific to this context: social psychologists in the United States. Our hope, however, is that this case study suggests advice for current and future female investigators that could help to narrow the gender gap and to increase women’s scientific impact more broadly.

Method

Bibliometric analyses

Using PsycINFO, we counted how many articles the 30 men and 23 women identified by Quiñones-Vidal et al. (2004) as frequent JPSP authors had published in JPSP from 1965 to 2004 (to extend Quiñones-Vidal et al.’s analysis, which had stopped at the year 2000). Tables 1 and 2 show the stark sex difference in JPSP authorship identified by Quiñones-Vidal et al. (2004). Using data from Tesser and Bau (2002), we noted the standardized ranking of these same authors for citations in the Gilbert, Fiske, and Lindzey (1998) Handbook of Social Psychology and in the Higgins and Kruglanski (1996) Social Psychology: Handbook of Basic Principles. The original Tesser and Bau list had 160 authors. Any of the 53 JPSP top authors not on this list were assigned a value of −1.00.

Table 1.

Scientific Impact of the Most Productive Male Authors in the Journal of Personality and Social Psychology

Author	JPSP 1965–2004	Handbook Citations	Textbook Citations
Wyer, Robert S.	52.00	−.10	.00
Petty, Richard E.	39.00	1.77	49.00
Diener, Ed	38.00	−1.00	.00
Mischel, Walter	34.00	−.22	.00
Cialdini, Robert B.	32.00	−.22	38.00
Insko, Chester A.	31.00	−1.00	.00
Mikulincer, Mario	31.00	−1.00	.00
Swann, William B.	30.00	−.27	.00
Greenberg, Jeff	30.00	−.53	7.00
Baumeister, Roy F.	30.00	.36	40.00
Cacioppo, John	29.00	1.56	42.00
Batson, C. Daniel	28.00	−1.00	27.00
Zuckerman, Miron	28.00	−1.00	.00
Bandura, Albert	27.00	−.53	12.00
Carver, Charles, S.	27.00	−.34	.00
Wegner, Daniel M.	26.00	.15	.00
Holmes, David S.	25.00	−1.00	.00
Komorita, Samuel S.	25.00	−1.00	.00
Cooper, Joel	25.00	−.60	.00
Zanna, Mark P.	25.00	.65	39.00
Higgins, E. Tory	25.00	3.33	10.00
Lepper, Mark R.	24.00	−1.00	.00
Snyder, Mark	24.00	.51	41.00
McCrae, Robert R.	24.00	−1.00	.00
Ross, Michael	24.00	−.43	.00
Pyszczynski, Tom	24.00	−.76	.00
Sarason, Irwin	23.00	−1.00	.00
Tesser, Abraham	23.00	.13	17.00
Feather, N. T.	22.00	−1.00	.00
Spanos, Nicholas P.	21.00	−1.00	.00

Note. Total number of articles = 846.

Table 2.

Scientific Impact of the Most Productive Female Authors in the Journal of Personality and Social Psychology

Author	JPSP 1965–2004	Handbook Citations	Textbook Citations
Taylor, Shelley E.	24.00	1.89	43.00
Chaiken, Shelly	20.00	2.34	44.00
Park, Bernadette	19.00	−1.00	.00
Eagly, Alice H.	18.00	1.94	50.00
Berscheid, Ellen	17.00	.51	29.00
Langer, Ellen	17.00	−1.00	.00
Major, Brenda	17.00	−1.00	.00
Fiske, Susan T.	16.00	3.30	48.00
Wortman, Camille	16.00	−1.00	.00
DePaulo, Bella M.	15.00	−1.00	.00
Matthews, Karen A.	15.00	−1.00	.00
Rusbult, Caryl E.	15.00	−1.00	.00
Dweck, Carol S.	15.00	−1.00	.00
Harackiewicz, Judith M.	15.00	−1.00	.00
Rodin, Judith	14.00	−1.00	.00
Andersen, Susan M.	14.00	−1.00	.00
McFarland, Cathy	14.00	−1.00	.00
Cantor, Nancy	13.00	−.64	.00
Spence, Janet T.	13.00	−1.00	.00
Walster, Elaine	13.00	−1.00	27.00
Nolen-Hoeksema, Susan	12.00	−1.00	.00
Grusec, Joan E.	12.00	−1.00	.00
Mackie, Diane	12.00	−.15	.00

Note. Total number of articles = 356.

Using data from Proctor and Bujak (2001), we also noted these authors’ rank order for textbook citation frequency (recoded from high to low). Proctor and Bujak used 12 full-length Social Psychology textbooks published between 1996 and 2001 to determine the 50 most cited social psychologists. Specifically, they collected data from the name index, recording the number of pages on which each author appeared. The number of pages for each author was summed and ranked. To control for bias, textbook authors’ scores were based on the other texts. Any of the 53 JPSP top authors not on this Proctor-Bujak list were assigned a rank order of zero. The handbook and textbook citations represent the second and third scientific impact variables in Tables 1 and 2, respectively.

We also coded whether authors were APS or APA Fellows as an indicator of scientific prestige. There were no sex differences on these variables, both ts(51) < 1.00, ns. For male authors, APS and APA Fellow status were related, r(28) = .49, p < .01. They did not covary reliably for women, r(21) = .27, ns. Because Fellow status did not predict our scientific impact variables, including JPSP authorship, for either gender, all rs < .29, ns, it will not be discussed. Finally, we counted the total number of published papers for these 53 authors, using PsycINFO. Although quantity of output does not necessarily represent quality, we presumed it would positively correlate with (at least) some of the scientific impact variables.

Institutional status

We used 1997 National Research Council (NRC) scores (available at www.socialpsychologynetwork.org) to code for the status of the authors’ graduate Psychology Ph.D. program. These ratings pertained to the top 185 U.S. programs and ranged from a rating of 49 (Boston University) to 72 (Stanford). Data for two authors who obtained foreign PhDs (one male, one female) and one male author who obtained an M.D. were not available.

Generating the control group

We generated a control group of authors, who had published at least once in JPSP, for comparison against the 53 JPSP top authors. First we generated a list of random numbers between 1 and 7,993 (the number of articles in JPSP between 1965 and 2004; #1 being the first editorial by Daniel Katz regarding the launching of JPSP and #7,993 being the last article in volume 87, issue 6). From the random number list, we looked up the first author of the corresponding article. The author was included in the control group and yoked to one of the original 53 authors if 1) the two authors received their Ph.D.s within three years of one another and 2) their Ph.D. granting institutions were within three points of each other, according to the 1997 NRC ranking score. If those conditions were not satisfied, then the random number was excluded and the next article (according to the next number in the random sequence) was chosen and the process repeated until all 53 authors had a yoked author, controlled for career age and graduate degree quality.

Personal factors websurvey

We examined personal factors using a brief websurvey emailed to the original 53 and corresponding control group of authors. We asked about authors’ family status (e.g., “Current marital status,” “Years married,” “Previously married?”, “Number of children”), department service (e.g., “Have you ever been Department Chair or served in another major administrative position (Dean, Provost)?”; “If yes, how many years?”), job satisfaction (e.g., “Overall, how would you describe your psychology career satisfaction?” on a 1 (disappointing) to 10 (highly satisfying) scale, and success negotiating with editors (“Consider only your submissions to JPSP: To the best of your recollection how often did you negotiate with an Editor about a paper (e.g., trying to change a rejection to a revise and resubmit)?”, “Of the times you negotiated with a JPSP editor, to the best of your recollection how often did it result in acceptance by JPSP?”. The number of respondents for the original and control authors were 44 and 36, respectively. All responses were anonymous.

Results

Gender Differences in Scientific Impact for Original 53 Authors

Consistent with Quiñones-Vidal et al. (2004), the gender difference in JPSP authorship from 1965 to 2004 was robust (Ms = 28.20 vs. 15.47, respectively for men and women, t(51) = 9.0, p < .001, d = 1.57). However, among these frequently published JPSP authors, there were no reliable sex differences for the number of handbook citations (standardized Ms = −.25 vs. −.29 for men and women, t(51) = .14, ns, d = .04), or for the number of textbook citations (rank order Ms = 10.73 vs. 10.48 for men and women, t(51) = .05, ns, d = .01). This suggests that JPSP visibility serves to ameliorate the large gender discrepancies found by Tesser and Bau’s handbook (2002) and Proctor and Bujak’s textbook (2001) textbook analyses of scientific citation impact. That is, per JPSP article published, top female authors were more cited in handbooks and textbooks than top male authors. Thus, our first counsel to female authors is to publish in the field’s flagship journal to increase their scientific impact.

Is there evidence of gender bias in JPSP article acceptance? We obtained access to APA’s general records of JPSP manuscripts submitted to articles published by gender. The gender difference is small, about 4%. Specifically, male authors had 135 articles accepted of 751 submitted (18%) and female authors had 65 articles accepted of 455 submitted (14%); however, it is not clear whether the cause is inferior quality, lack of persistence, or bias. To the extent that bias is a possibility, we recommend blind review for everyone, not the practice during the time period covered by these APA data.

Although not shown in Tables 1 and 2, men published more total articles than did women, resulting in a large effect size; Ms = 134.57 vs. 74.83, t(51) = 5.02, p < .001, d = 1.15. However, computing the percent of published articles that appeared in JPSP (1965–2004) found no sex differences (both Ms = 23%). Thus, men produce more articles but research quality (as measured by JPSP acceptance) appears to be equal for both genders. This suggests that women would need to make up the difference in sheer volume, not so much per-article quality, to catch up to male JPSP authors. Arguably, women might choose a different model, publishing fewer but higher impact articles, a point we explore next.

Predictors of Scientific Impact Variables

Table 3 shows the relationships among the scientific impact and productivity variables, separately for women and men. The first column presents number of JPSP publications (1965–2004). Our first surprise was that JPSP authorship is strongly related to textbook and handbook citations, but only for women. This observation buttresses our first counsel to women (that publishing in JPSP increases their scientific impact). For men, these correlations were negligible, suggesting that they have more independent sources of scientific impact than do women (that is, JPSP and textbook/handbook citations are separate indicators). A second surprise is that women who produce more total articles tend to appear in JPSP (r = .29, ns), whereas men’s total output does not at all predict their JPSP visibility (r = .04, ns). Thus, perhaps women may have to work harder than men do to achieve success in the journal. That is, their sheer productivity may have to be higher for their work to appear in JPSP. This relationship also supports our second counsel to women (to produce more overall output to close the JPSP gender gap).

Table 3.

Correlates of Scientific Impact for Women and Men

	JPSP (1965–2004)	Textbook Citations	Handbook Citations	N Articles
Women
Textbook	.56**
Handbook	.55**	.92**
N Articles	.29	.31	.37†
Percent	.09	−.12	−.21	−.79**
Men
Textbook	.12
Handbook	.15	.57**
N Articles	.04	.19	.22
Percent	.34†	−.17	−.20	−.86**

Note. Percent = the number of JPSP articles (1965–2004) divided by N Articles (the total of published articles, 1965–2004).

^†p < .10.

^*p < .05.

^**p < .01.

Tesser and Bau (2002) found the relationship between handbook and textbook citations to be robust, and stronger than the link between sheer number of articles (output) and scientific impact. Table 3 echoes their findings for both genders. However, the magnitude of the handbook and textbook citation correlation for women is nearly twice that for men (rs = .92 vs. .57, respectively, p < .05), again suggesting that men draw scientific impact from more independent sources than do women.

Finally, men’s JPSP visibility covaried with how much of their total output was published in the journal (the percent index), p < .05. There was no comparable linkage for women. That is, men who appear often in JPSP tend to use the journal as a home for their published research, whereas women do not. Thus, our third counsel to women is to submit more of their research to the journal.

Gender Differences in Variables Predicting JPSP Visibility

Because JPSP visibility strongly relates to women’s scientific impact, it is important to uncover potential predictors of their success in the journal—predictors that might help to explain the gender gap in JPSP authorship. For example, possibly men began publishing in JPSP earlier, compared with women. However, we found only an unreliable 3-year difference between their initial publications (Ms = 1972 vs. 1975, ns, d = .46). Thus, men did not appear to have a significant “running start.” However, and by contrast, men’s most recent publication marked a later year (M = 2001), compared with women (M = 1996), resulting in a large effect (d = .78). Taking the difference between initial year and most recent year as an index of JPSP publishing period, we found that men have been publishing longer in the journal compared with women (Ms = 28.30 vs. 20.78 years, d = 1.08)—a difference of 7.5 years. This suggests that, although they started around the same time, women tend to end their JPSP publishing careers before men do. In addition to length, we computed the JPSP publication rate (ratio of JPSP publications to length of time publishing in JPSP). Men’s publication rate was considerably faster, compared to women’s (Ms = .94 vs. .76, p < .05, d = .72).

Do men and women show differences in JPSP authorship order? This would follow just from men publishing a higher quantity in general; however, it is possible the gender gap may vary by authorship position. Analyses revealed that men had more first and second authorships than did women, t(51) = 5.14, 2.66, ps < .05, ds = 1.16 and .68 for first and second authorships, respectively; Ms = 14.76 vs. 7.70 for men versus women for first authorships and 8.43 vs. 5.53 for second authorships. Third authorships favored men, but did not reliably differ by sex, t(51) = 1.38, ns, d = .38; Ms = 1.80 vs. 1.21. Gender disparity is by far the greatest for first authorship (nearly twice that for 2^nd authorship and twice or more for later authorships). From this we can conclude that men are especially often first authors (14.76) compared with later authors (8.43, 1.80, .80), whereas women are also more likely to be first (7.70), but nearly as likely to be second (5.53). Quiñones-Vidal et al. (2004) suggested that men’s greater tendency to network could reap an advantage. If so, men may publish more often as “late” (fourth or higher) authors, compared with women. Indeed, combining fourth through highest authorships (seventh was the highest), we found exactly this tendency (Ms = .80 vs. .17 for men and women, t(51) = 2.15, p < .05, d = .58). Moreover, women never appeared later than fourth author.

We also analyzed demographic variables. On average, marital status (1 = single, 2 = married) did not reliably differ by gender; Ms = 1.80 vs. 1.74 for men and women, t(51) = 0.75, ns, d = .13. Although men tended to have more children, this difference was nonsignificant; Ms = 1.28 vs. .86 for men and women, t(51) = 1.57, ns, d = .45. Plausibly, male authors were older than female authors, and this was borne out. On average, men received their PhDs five years earlier; Ms = 1969 vs. 1974, t(51) = 2.06, p < .05, d = .58; however, as stated above, we found only an unreliable 3-year difference between men’s and women’s initial JPSP publications (Ms = 1972 vs. 1975).

Accounting for the Sex Difference in JPSP Success

Initial analyses suggested that publishing longer and faster are two variables that, not surprisingly, predict JPSP success, and both showed a gender difference that favors men. Taken together, they might account for the gender difference found by Quiñones-Vidal et al. (2004). Figure 1 shows the results of a mediational analysis (Baron & Kenny, 1986) predicting JPSP publications (1965–2004) from publishing rate and length. As can be seen, the substantial link between gender and JPSP publications was reduced to nonsignificance after accounting for publication career length and speed. Thus, to eliminate the gender gap in JPSP publications, women need to (a) continue to publish in the journal and (b) publish at a faster rate.

Figure 1.

Regression analyses testing length and rate of publication in JPSP as mediators of the relationship between author gender and JPSP publications (1965–2004). Coefficients in parentheses reflect a bivariate analysis. A dashed arrow indicates successful mediation. Author gender was coded 2 (male) 1 (female).

But of course length and rate add up to sheer numbers, and women have less control over when they start to publish than the rate at which they publish in JPSP. Separate mediational analyses for length and rate yielded a decrease in the effect of author gender on JPSP articles, but the gender gap remained reliable in each case, both betas > −.68, ps < .001. Thus, adjusting for the fact that men have been publishing longer in the journal than women was not a sufficient explanation for the overrepresentation of men. Instead, the faster rate at which men publish in JPSP is necessary to include to fully account for the gender discrepancy in JPSP publications.

Correlates of JPSP Variables

The above analysis is not particularly satisfying with respect to illuminating the gender gap in JPSP success. Suggesting to women that they need to publish longer and faster does not explain why men have longer publishing careers at JPSP or produce JPSP papers at a faster rate. In addition, other variables (beyond length and rate) might covary with JPSP success. Table 4 shows potential correlates of JPSP productivity, as well as length and rate of publishing in the journal, separately by gender. Table 4 includes variables that were derived from JPSP publications and demographics taken from the websurvey.

Table 4.

Correlates of JPSP Articles and JPSP Publishing Length and Rate, By Gender

	Women			Men
	JPSP Articles	Publishing Length	Publication Rate	JPSP Articles	Publishing Length	Publication Rate
JPSP Variables
Length	.55**			.20
Rate	−.15	−.84**		.59**	−.59**
First Author	.34†	−.05	.27	.43*	.17	.37*
Second Author	.21	.38†	−.27	.68**	.18	.29
Third Author	.15	.09	−.06	.02	.19	−.14
Late Author	.42*	.09	.04	−.17	−.31†	.07
Initial Year	−.12	−.24	−.04	−.01	−.76**	.42*
Recent Year	.36†	.57**	−.69**	.27	.22	−.17
Demographics
PhD Year	.04	−.24	.05	−.02	−.57**	.30†
Marital Status	.10	.37†	−.38†	.34†	.16	.12
Children	.21	.21	−.22	.14	.26	−.11

Note. Publishing Length = the number of years publishing in JPSP. Publication Rate = number of JPSP articles divided by Publication Years. Late author = sum of fourth through seventh authorships. Initial Year marks the date of authors’ first JPSP publication. Recent Year marks the most recent JPSP publication year. Marital status was coded 1 = single, 2 = married. Children = number of children.

^†p < .10.

^*p < .05.

^**p < .01.

JPSP variables

As already noted, and as is obvious, length and rate of publishing in JPSP predicts sheer number of JPSP articles; Table 4 reveals that for women, length was a reliable correlate of total number of JPSP articles, whereas rate was not. Thus, women may have to work several years longer to achieve high JPSP visibility. By contrast, for men, rate was a reliable correlate, whereas length was not. Thus, men who work quickly achieve success in the journal, and they do not necessarily have to earn their status over a long period time. Not surprisingly, publishing length and rate were negatively related for both genders, p < .05, suggesting that a longer publishing career covaried with slowing down.

Table 4 also reveals correlations between authorship order and JPSP visibility. First and second authorships correlated positively with men’s JPSP success. Surprisingly, first authorship covaried marginally positively with women’s success, whereas second authorship showed only a weak positive link. Presuming that first authorship demands more time and energy than second authorship, this is an indication that women may have to work harder than men to achieve JPSP success. Interestingly, “late” authorship was negatively linked to JPSP success for men, albeit weakly. However, it was positively related to JPSP success for women. As a result, cultivating networks appears to be good advice for women, because it may promote women’s JPSP visibility. Finally, authorship order was not reliably related to publishing length or rate for either gender, with one exception. Men showed a positive relationship between first authorship and publication rate, whereas for women, this link was unreliably positive. In addition, men showed a marginally negative relationship between late authorship and publishing length, suggesting greater networking among younger male authors. Women showed a marginally positive link between second authorship and publishing length, suggesting that teamwork may extend women’s JPSP publication career.

Demographic variables

As noted, men tended to obtain their doctoral degrees earlier than women did. Table 4 shows that date of degree was related to neither men’s nor women’s JPSP success. However, younger men are more likely to produce faster, whereas Ph.D. year and rate are not at all related for women. This suggests that younger women are not likely to close the gender gap. Rather, for women, it is a career-long process.

Might family demands hinder women’s JPSP success? As can be seen, marital status and family size (number of children) were unreliably related to JPSP success for either gender, although there was a tendency for married men to publish more often in the journal, compared with unmarried men. For women, marital status positively covaried with publishing length, but negatively covaried with publishing rate, albeit marginally so in each case. To the extent that marriage slows down women’s rate of JPSP publishing, it may augment the gender gap. The positive link between marriage and publishing length may also reflect the fact that older women (as measured by PhD year) tend to be married (r =.42, p < .05), whereas age and marriage are unrelated for men (r = .17, ns).

Institutional status

Another possible explanation for the gender gap in JPSP authorship concerns the institutional status of our authors. If men graduated from more prestigious programs, they may be better equipped for an academic career. Unexpectedly, results showed that women graduated from more prestigious programs than men did, t(48) = 2.33, p < .05 (Ms = 67.00 vs. 63.71, d = .66). However, PhD program status was unrelated to JPSP authorship variables, including number of publications, publishing years, and rate, for both genders (all ps > .30). Thus, although women tended to graduate from more prestigious programs, this difference cannot inform the gender gap in JPSP publications. Readers may find it noteworthy that 30% of the authors graduated from prestigious private institutions (5 from Stanford, 4 from Yale, 4 from Harvard, and 3 from Princeton). Three authors each graduated from some of the most prestigious public universities--University of Michigan, Ohio State University, and University of North Carolina - Chapel Hill--resulting in an additional 17%.

A second possibility was that men might tend to be hired at more prestigious institutions, compared with women. Assuming that institutional prestige reflects greater publishing resources (e.g., collaborations with other faculty, internal research funds, and graduate students), it might have explanatory power. However, we found no gender differences in the status of authors’ first or most recent departments, both ts < 1.00, ns. (Status was not coded for 5 male and 2 female authors housed in foreign institutions or government agencies.) In addition, institutional prestige, whether first or most recent, was unreliably related to the authors’ number of JPSP publications, as well as their rate of production, all ps > .31. However, men’s most recent status predicted their publishing length at JPSP, r(22) = .43, p < .05. By contrast, women did not show this relationship, r(19) = .20, ns. Thus, being housed in a prestigious institution may help men (but not women) to sustain their JPSP publishing career.

Finally, to examine whether men tended to gravitate, over time, to departments with higher status more often than women, we computed the difference in status between authors’ first and most recent institution. This analysis yielded no sex difference, t(51) < 1.00, ns. That is, men did not show greater upward mobility than women. However, upward mobility tended to covary with number of JPSP publications, but only for men, r(22) = .39, p =.06. Women did not show this link, r(19) = .13, ns. One implication of this finding, as well as the relationship between men’s most recent status and their JPSP career length, is that men may make greater use of the institutional resources available to them. Alternatively, these results could reflect that men’s upward mobility is more dependent on their JPSP publication record, compared with women’s. The latter seems unlikely, given the importance of women’s JPSP publications for predicting other career status indicators (e.g., handbook and textbook citations). In either case, the fact that institutional status variables did not differ by sex preclude them as mediators of the gender gap in JPSP authorship. Although they may shed some light on why men outperform women at the journal, they cannot account for the gap.

JPSP editorial status

Although to date, there have only been two female editors at JPSP (Patricia Devine and 2009 incoming editor, Laura A. King), the higher influx of women into the field from 1965 to the present ought to be reflected in greater female editorial positions, including associate and consulting editors.

We computed the percent of men and women from the original 53 who had served as editor, associated editor, or consulting editor for JPSP, across four decades (Table 5). While the percent of highly productive men serving as editor has increased a miniscule amount over time (from 0 to 7%), none of the most productive women have served as JPSP editor. The percent of productive men and women serving as associate editors has remained fairly stable over time (3–7% for men, and 4–9% for women). However, the percent serving as consulting editors increased for both men and women between 1965 and 2004. Collapsing across all four decades, the years served by an individual as an associate or consulting editor correlates with scientific impact for both genders. For men, years spent as an associate editor marginally relates to total number of JPSP articles, r(28) = .33, p = .07; years spent as a consulting editor positively correlates with handbook citations, r(28) = .59, p < .05, and textbook citations, r(28) =.52, p < .05, both of which are indicators of scientific impact. Similarly, for women, years spent as a consulting editor positively relates to total number of JPSP articles, r(21) = .46, as well as handbook citations, r(21) = .57, and textbook citations, r(21) = .59, all ps < .05. This suggests that accepting invitations to become a consulting editor at JPSP may be beneficial for women’s productivity and visibility in the field. Alternatively, perhaps women are only asked to serve as consulting editors once they have become successful. Correlations that account for a lag demonstrated that for women, total number of JPSP articles in decade 1 positively related to years spent as consulting editor in decade 2, r(21) = .59, p < .05, and number of JPSP articles in decade 3 positively related to years spent as consulting editor in decade 4, r(21) = .44, p < .05; however, this relationship disappeared for female consulting editors who served from 1985 to1994. That said, we see a similar pattern for men: total number of JPSP articles in decade 1 positively related to years spent as consulting editor in decade 2, r(21) = .36, p < .05, number of JPSP articles in decade 2 positively related to years spent as consulting editor in decade 3, r(21) = .45, p < .05, but, this relationship disappeared for male consulting editors who served from 1995 to 2004. Overall, the lagged correlations suggest that being a consulting editor comes as a result of productivity.

Table 5.

Percent of Top-Published JPSP Authors Serving as JPSP Editors by Decade and Gender

Time Span	Men			Women
	Editor	Associate Editor	Consulting Editor	Editor	Associate Editor	Consulting Editor
1965–1974	0%	3%	17%	0%	0	4%
1975–1984	0%	7%	53%	0%	4%	26%
1985–1994	3%	3%	47%	0%	9%	57%
1995–2004	7%	3%	70%	0%	4%	52%

Note. Men N = 30; Women N = 23. Authors were counted as new editors each decade even if they were already counted in the previous decade.

JPSP Authorship by Decade

To examine whether the gender gap in JPSP publications might be modified by decade, we measured authors’ publication frequency over four decades (1965–1974–1975–1984–1985–1994, and 1995–2004). The first column of Table 6 shows the results in effect sizes, collapsed across authorship order. In each time period, men outscored women in total number of JPSP publications. However, the first decade showed only a marginally significant sex difference, p = .05. The next two decades revealed an increased gender gap, which has not been diminished in the most recent decade (all ps < .01). In fact, the difference between the first and last decade’s gender gap is modest (d = .16).

Table 6.

Effect Sizes for the Gender Gap in JPSP Publications by Decade and Author Order

Time Span	Authorship Combined	First Author	Second Author	Third Author	Fourth or More Author
1965–1974	.53†	.60*	.19	.19	.32
1975–1984	.88**	.73**	.59*	.18	.09
1985–1994	.74**	.65*	.38	.18	.18
1995–2004	.69**	.48†	.42	.38	.51†
1965–2004	1.57**	1.16**	.68*	.38	.58†

Note. Effect sizes are Cohen’s d. Positive effects indicate that men published more than women. By convention, small, medium, and large effect sizes correspond to .20, .50, and .80, respectively (Cohen, 1988).

^†p < .10.

^*p < .05.

^**p < .01.

Table 6 also shows effect sizes for gender differences as a function of JPSP authorship order, by decade. Gender differences in first authorship occur in every time period, but have decreased steadily since the second decade, resulting in a marginal difference for the most recent decade, p = .08. By contrast, the gender gap in late authorship—a possible indicator of networking—has increased in the most recent decade, resulting in a marginal advantage for men, p = .07. Third authorship is the only variable that has not yet shown a reliable sex difference, but the gender gap has doubled in the most recent decade, compared with earlier time periods. Thus, while women appear to be closing the gap when it comes to first authorship, the networking gap has grown.

Finally, Table 7 shows tests of within-decade differences in JPSP publication for women and men. The top row for each gender reflects the total number of papers, collapsed across authorship order. As can be seen, compared with the first time period, women produced more papers in the second and third decades. However, the most recent decade shows a return to the publication level of the first. That is, the first and fourth decades do not reliably differ for women. Although women’s fourth decade average is not different from the second’s, it is reliably smaller than their output in the third decade. By contrast, men produced more papers in all time periods, compared with the first. Moreover, their output in the most recent time period is comparable to their second and third decade’s averages. This is another sign that the gender gap is not on the wane.

Table 7.

Mean JPSP Publications as a Function of Gender, Time Period and Authorship Order

	1965–1974	1975–1984	1975–1984	1995–2004
Women
Total	2.13 a	4.76 b, c	5.04 b, c	3.87 a, b
First Author	1.22 a	3.09 b	2.48 a, b	1.09 a
Second Author	.74 a	1.39 a, b	1.96 b	1.78 b
Third Author	.17 a	.26 a	.48 a	.56 a
Late Author	.00 a	.04 a	.13 a	.08 a
Men
Total	4.86 a	8.53 b	7.76 b	7.03 b
First Author	3.63 a	5.43 a, b	4.17 a, b	2.37 a
Second Author	.87 a	2.63 b	2.77 b	2.90 b
Third Author	.30 a	.40 a, b	.60 a, b	1.10 b
Late Author	.06 a	.06 a	.23 a, b	.67 b, c

Note. Means not sharing a subscript differ within rows at the p < .05 level or higher. Late author = sum of fourth through seventh authorships.

The remaining rows in Table 7 indicate JPSP publications as a function of authorship order, for each time period. For women, first authorship rose in the second decade, but reverted to the initial figure in the third and fourth decades. Men showed a similar pattern. For both genders, second authorship increased by the third decade and remained steady in the fourth. However, third and late authorship show gender discrepancies. In each case, women’s averages remained steady across the time periods. By contrast, men’s averages have reliably increased by the fourth decade. In concert, Tables 6 and 7 suggest that men counteract their decrease in first authorship publications by networking, resulting in a steady output of JPSP papers over time, whereas women have not used this strategy as effectively.

Gender Differences in Comparison to Control Group

Is the gender gap unique to the 53 authors identified as top publishers by gender? To address this question, a 2 (male/female author) X 2 (original/yoked-control authors) ANOVA tested whether these groups differed in publication success and professional and personal factors. The main effects of author group demonstrated that the original 53 authors had higher values than the control authors on the following variables: handbook citations, textbook citations, number of APS fellows, total articles, total JPSP articles, percent of total articles that appeared in JPSP, length of career, rate of publication in JPSP (in general and adjusted for length of career), 1^st author, 2^nd author, 3^rd author JPSP articles, and recent JPSP articles (published between 2001 and 2004), all Fs(1, 105) > 6.35, p ≤ .01. The original 53 also had earlier dates of 1^st JPSP article and later dates of last JPSP article, Fs(1, 105) > 5.82, p < .05.

Gender main effects revealed that control-group men had higher values than women on the following variables: total articles, total JPSP articles, 1^st, 2^nd, 3^rd and 4^th or later author JPSP articles, recent JPSP articles (published between 2001 and 2004), length of JPSP career, rate of JPSP publication and rate of JPSP publication in the last decade, all Fs(1, 105) > 8.20, p < .01. Men also had earlier Ph.D. dates, and earlier dates for their first JPSP articles, all Fs(1, 105) > 5.58, p < .05.

The following effects were qualified by a significant Group x Gender interaction (means, highest to lowest: Men in Table 1 > Women in Table 2 > Control Men > Control Women): total articles, total JPSP articles, 1^st author JPSP articles, and recent JPSP articles, all Fs(1, 105) > 6.60, p < .01. These interactions all followed the same pattern: the difference between Men in Table 1 and Women Table 2 was greater than the difference between Control Men and Control Women. Interestingly, we observed a slightly different pattern for the per-year rate of publication in JPSP, such that the productive men (Table 1) published at a greater rate than the productive women (Table 2), but that the control women published at a greater rate than the control men (who both published at a slower rate than the productive women in Table 2), F(1,105) = 4.61, p < .05. All of these interactions held when we reanalyzed these data using an ANCOVA controlling for Ph.D. year, indicating these effects are not driven solely by length of career.

We also examined differences in personal variables between the two author groups. The original 53 authors were more likely to be married that than control authors, F(1, 79) = 18.88, p < .01; they also, however, had marginally fewer children than the control authors (M_original = 1.49, SD = 1.24; M_control = 1.93, SD = 0.94), F(1, 79) = 3.33, p < .10. Female authors (in general) tended to have fewer children than male authors (M_female = 1.26, SD = 1.10; M_male = 1.91, SD = 1.12), F(1, 79) = 3.22, p < .10. None of the other personal variables varied as a function of author group or gender.

Collapsing across author groups, their gender, marital status, number of children, and self-reported tendency of negotiating with editors correlated with JPSP publication outcomes: total number JPSP articles, rate of publication in JPSP (adjusted for length of career), number of 1^st author JPSP articles, and number of recent JPSP articles (1995–2004). We were interested in examining the extent to which marital status, number of children, and tendency to negotiate with editors predicted these publication outcomes for men and women separately. We ran separate regressions for the four publication outcomes, for men and women, respectively (Table 8). We always entered the following variables stepwise in this order: number of children, marital status, and tendency to negotiate with editors. For men, being married (controlling for number of children and tendency to negotiate) is related to higher numbers of JPSP articles, greater rate of publication, more 1^st author JPSP articles, and higher numbers recent JPSP articles (1995–2004). Men’s tendency to negotiate with editors (controlling for marital status and number of children) also predicted total number of JPSP articles and marginally predicted number of 1^st author JPSP articles. None of the personal factors significantly predicted JPSP publication outcomes for women.

Table 8.

Personal Factors Predicting JPSP Publication Outcomes

	Men				Women
	Total # JPSP	JPSP Adj. rate	1st Author	Recent JPSP	Total # JPSP	JPSP Adj. rate	1st Author	Recent JPSP
Predictor
Number of Children	−0.37 (1.12)	0.02 (0.05)	−0.79 (0.82)	−0.47 (0.65)	−0.75 (0.83)	−0.05 (0.04)	−0.69 (0.56)	0.25 (0.52)
Marital Status	14.97** (2.53)	0.25* (0.12)	7.61** (1.85)	6.17** (1.47)	0.53 (1.86)	0.01 (0.10)	−1.43 (1.24)	−0.44 (1.17)
Negotiate with Ed.	1.13* (0.52)	0.02 (0.03)	0.70† (0.38)	−0.30 (0.29)	1.03 (0.79)	−0.03 (0.04)	0.63 (0.53)	−0.12 (0.50)

Note. Unstandardized betas, standard error appears in parentheses. Predictors were entered simultaneously; table displays last model including all 3 predictors. Regression includes original 53 authors and yoked-control authors. Recent JPSP = number of JPSP articles 1995–2004. Marital status was coded 1 = single, 2 = married.

^†p < .10.

^*p < .05.

^**p < .01.

Discussion

Clearly, publishing in JPSP matters to a social psychologist’s career. Women’s publication status in JPSP is especially linked to their citation frequency indices, whereas men apparently have other independent sources of scientific impact. As a result, publishing in JPSP is particularly important for women. Taken together, these findings illuminate potential factors underlying the gender gap in JPSP publications. Moreover, there are reasons to suspect that the gender gap is not likely to decrease – at least, not among these productive authors.

First, authors’ age, as indexed by Ph.D. year and year of initial publication in JPSP, predicted both length and rate of JPSP publications, but only for men. Male authors who started later (and are, presumably, younger) not surprisingly have shorter publishing careers, but they also produce at a faster rate than more experienced authors. However, women did not echo these observations. That is, younger women are not necessarily faster, or likely to have shorter JPSP publishing careers, compared with experienced women. This suggests that a “running start” does not predict JPSP publishing length for women, and youth is not advantageous vis-à-vis their speed. Second, marriage is linked to age and a slower productivity rate for women (but not for men), suggesting that older women are likely to marry, but spousal duties may slow women down. Third, recent female JPSP authors tend to be experienced (i.e., success for women breeds success), but they also produce JPSP papers less often, whereas there is no matching maturity or slowing down effect for recent male authors.

Are there signs of burn-out for women? Yes. First, women and men started publishing at approximately the same time in their careers, but men continue to publish longer in JPSP. Second, although length and rate were negatively linked for both genders (suggesting a slower rate for experienced authors; see Table 4), this relationship was stronger for women than for men, z = 1.93, p = .05. Thus, experienced men appear to have more stamina than their female counterparts.

Are there signs that younger women can pick up the slack? No. On the contrary, men who recently started to appear in the journal are publishing fast, but this is not so for younger women. Moreover, men’s career span at JPSP does not predict their more recent publications. By contrast, women with longer track records are more likely to have authored recently in the journal. So while women need JPSP more than men do to ensure scientific impact, the indications are that women have to work longer to achieve a track record at JPSP, including recent authorship, but this achievement slows them down (more so than men), suggesting that the gender gap is not likely to decrease in the future.

In fact, when publications were examined by decade, there were two disturbing trends that buttress this conclusion. First, the gender gap started out as marginally significant, became reliable in the second decade and has remained so up to the most recent time span (1994–2004). In other words, the gap does not show a trend toward closing. Second, men have outpaced themselves in every decade, compared with the first. After an initial publication increase in the second decade, their numbers have remained steady. By contrast, women have declined in their output when the third and fourth decades are compared so that, most recently, their numbers have reverted back to the first decade. A possible reason for this is that men have widened the gender gap in networking (third or later authorship) in recent times, even while women have begun to catch up to men vis-à-vis first authorship. Nonetheless, the most compelling explanation is that men produce papers in the journal for a longer time period and at a faster rate; together, these two variables fully account for the male advantage in JPSP authorship. Our advice to women as individual scholars, then, is to submit early and often to JPSP, to network and collaborate, and above all to persist.

How can the field, as a whole, address the gender disparity? First, we reiterate the following: to the extent that bias is a possibility, we recommend double-blind review for everyone. Subtle bias may operate at every stage of evaluation and targets of discrimination are often unaware of the role it has played in the rejection or acceptance of their work (see Bendick & Nunes, 2012, and Pager & Western, 2012, for empirical approaches that bring subtle discrimination to light). Even among social psychologists—many of whom are dedicated to eradicating bias and discrimination, but who are also responsible for reviewing their peers’ work—knowledge of the existing biases may not be sufficient to inoculate the evaluators (see Kang, 2012, for a discussion of evaluators’ self-analysis with regard to bias). Second, the field needs to recognize multiple indicators of quality, such as textbook and handbook citations, as well as sheer numbers of articles. Current promotion trends acknowledge this by examining journal impact and article citation rates. Finally, as a career strategy, woman may be deciding to submit only their best work to a flagship journal such as JPSP because they perceive their research to be a trade-off with other non-work roles, so it had better be worth the sacrifice. This quality-over-quantity strategy is an adaptive model, and the field needs to recognize it as a viable option. We examined personal factors by a brief questionnaire emailed to all authors highly-published in JPSP and to a control group of authors who also publish in JPSP but not as often.

We asked about authors’ family status, department service, job satisfaction, and success negotiating with editors. We predicted that negotiating with editors ought to increase publication outcomes for both genders, whereas marriage and children might decrease women’s productivity more than men’s (e.g., Eagly, 1987). Surprisingly, negotiation and being married were only reliable predictors for men (not for women). Specifically, men have marginally better self-rated negotiation success with editors, men have published longer in the journal, and men are more likely to be married. Of these predictors uncovered by our analyses that favor men (negotiation success, years publishing in the journal, and being married), only negotiation success accounted for unique variance when predicting JPSP success. We can just speculate as to why this pattern emerged. First, negotiation is often more successful for men than for women (Babcock & Laschever, 2003), and some evidence indicates that this is due to a double standard (i.e., women, but not men, who negotiate are perceived as “pushy” and “demanding”; Bowles, Babcock, & Lai, 2007). Second, having a spouse may afford men more time to concentrate on their careers, given that wives are typically more helpful than husbands regarding domestic responsibilities, even in dual-career families (e.g., Biernat & Wortman, 1991). Future research may determine what accounts for men’s greater effectiveness when negotiating with editors and their success at publishing in JPSP when married.

Several potential explanations for the gender gap were not supported by our research. First, although sex differences in networking appeared, with men appearing as fourth (or later) authors more than women, this variable was not linked to JPSP success. Second, the idea that women’s greater domestic demands undermine success was not supported; marital status and number of children were weak and not unique predictors of women’s JPSP success. Third, men were not likely to receive their degrees from institutions with higher status than women. In fact, the reverse was true, although this advantage was not a predictor of JPSP success. Also surprising was the absence of gender differences in authors’ first and most recent institutional status.

Finally, our analyses showed no sign that the gender gap in JPSP authorship is a function of cohort. Despite a steady increase in female reviewers and editors over time, women remain at a disadvantage in JPSP publications. In fact, when publications were examined by decade, we found that the gender gap started out as marginally significant, became reliable in the second decade, and has remained so up to the most recent time span (1994–2004). That is, there does not appear to be even a trend toward closing the gap.

Whatever the additional explanations, we have evidence of a gender gap in productivity overall, including in the sheer number (though not the proportion) of articles in top-ranked journals. A number of small effects all contribute: years married, length of time publishing in the journal, and negotiation with editors. In addition, we have evidence of a small amount of gender disparity in acceptance rate per article submitted. Absent an independent indicator of article quality, we cannot call this discrimination, but other measures suggest that women’s articles are not of inherently lower quality (if anything, the reverse, considering impact per JPSP article published); the gender discrepancy in articles published does not diminish women’s impact (as measured by textbook and handbook citations). Though our analysis is specific to social psychologists within the U.S., given the pervasiveness of subtle bias, we suspect that a similar pattern of findings might emerge in other scientific disciplines and non-academic domains in which an individual’s success is determined in large part by the sheer quantity of her productivity and visibility in the field. If the disparity is caused by discrimination, it is discrimination by a thousand little cuts. There is no single, quick fix.

Biographies

Mina Cikara is currently a NIH Ruth L. Kirschstein Postdoctoral Fellow at MIT. She received her Ph.D. in Social Psychology and Social Policy from Princeton University in 2010. Cikara is a social cognitive neuroscientist who studies how competition, stereotypes, and prejudice disrupt the processes that allow people to see one another as human and to empathize with others. She uses a wide range of tools—standard laboratory experiments, implicit and explicit behavioral measures, fMRI and psychophysiology—to study how misunderstanding, failures of empathy, and dehumanization unfold in the brain, as well as the behavioral consequences of these processes. She will begin a tenure-track faculty appointment in the department of Social and Decision Sciences at Carnegie Mellon University in Fall 2012.

Laurie A. Rudman is a Professor of Psychology at Rutgers University in New Brunswick, New Jersey. Her research interests are intergroup relations and social justice issues, with a focus on predicting behavior from implicit attitudes and beliefs. The author of over 60 professional publications, she serves on several editorial boards and is currently a Senior Associate Editor of Personality and Social Psychology Bulletin. Her honors and awards include a National Science Foundation Fellowship, a National Research Service Award (National Institutes of Health), and she has twice been awarded (with Eugene Borgida and Julie Phelan) the Gordon Allport Prize for the best paper on intergroup relations, given annually by the Society for the Psychological Study of Social Issues. Dr. Rudman is a fellow of the American Psychological Association, the Association for Psychological Science, and an honorary member of the Society for Experimental Social Psychology (SESP), for whom she served as representative to the Federation of Behavioral, Psychological, and Cognitive Sciences. Her current NSF- funded program of research aims to discover the factors that promote or hinder negative reactions to counterstereotypicality on the part of both perceivers and actors. The downstream consequences of these reactions include cultural stereotype maintenance and human capital issues. The author of The Social Psychology of Gender: How Power and Intimacy Shape Gender Relations (with Peter Glick), Dr. Rudman has frequently served as an expert witness in gender discrimination cases.

Susan T. Fiske is Eugene Higgins Professor of Psychology, Princeton University (Ph.D., Harvard University; honorary doctorates, Université Catholique de Louvain-la-Neuve, Belgium; Universiteit Leiden, Netherlands). She investigates social cognition, especially cognitive stereotypes and emotional prejudices, at cultural, interpersonal, and neuroscientific levels. Author of over 250 publications and winner of numerous scientific awards, she has edited most recently, Beyond Common Sense: Psychological Science in the Courtroom (2008) and the Handbook of Social Psychology (2010, 5/e). Currently an editor of Annual Review of Psychology, Psychological Review, and the forthcoming Behavioral Science and Policy, she wrote Social Beings: Core Motives in Social Psychology (2010, 2/e) and Social Cognition: From Brains to Culture (2008, 3/e). Sponsored by a Guggenheim, her 2011 Russell-Sage-Foundation book is Envy Up and Scorn Down: How Status Divides Us.