A Longitudinal Analysis of the Effect of Alphabetization on Academic Careers

and Tolga Yuret

Abstract

A prevalent belief is that it is advantageous to have surname initials that are placed early in the alphabet (early surname initials) in academic fields in which authors are ordered alphabetically (alphabetic academic fields), because first authors are more visible. However, it is not certain that the advantage is strong enough to affect academic careers. In this paper, the advantage in having such early surname initials is analyzed by using data from 1,345 course catalogs that span a 100 years. We obtained academic titles and surname initials of 19,353 faculty members who appeared 211,816 times in these course catalogs. Two alphabetic academic fields – economics and mathematics – and four other academic fields that are not alphabetic were analyzed. We found that there are some years when faculty members who have early surname initials are more likely to be full professors. However, there are many other years when faculty members who have early surname initials are less likely to be full professors. We also analyzed the career path of each faculty member. Economists who have early surname initials are found to be more likely to become full professors. However, this result is not significant and does not extend to mathematicians.

1 Introduction

There are some academic fields in which authors of an article are sorted according to their surname initials. These academic fields are called “alphabetic academic fields”. It is important to have surname initials that are placed early in the alphabet in the alphabetic academic fields, and these surname initials are referred to as “early surname initials”.

Authors who have early surname initials have more chance to become the first authors in the alphabetic academic fields. Suppose that professors A, Y, and Z write a joint paper. Professor A becomes the first author regardless of his or her contribution to the paper. First authors are more visible and being more visible has advantages. Because of this, the author who contributes most has the right to become the first author in some of the academic fields that are not alphabetic.

There is evidence that researchers who have early surname initials have better publication performance. They are able to publish more papers because their reputation is enhanced by their increased visibility (Van Praag & Van Praag, 2008). Their publications also receive more citations for the same reason (Shevlin & Davies, 1997). They collaborate more often because they are more likely to remain as the first author in collaborated papers (Kadel & Walter, 2015, Ong, Chan, Torgler & Yang, 2018).

Because authors who have early surname initials have better publication performance, it is natural to expect that they are more successful in their academic careers. Full professors who have early surname initials are more prevalent in top-ranked institutions (Efthyvoulou, 2008). Moreover, faculty members who have early surname initials are more likely to get tenure 1 in top institutions (Einav & Yariv, 2006).

A closer look to the previous studies suggests that their results are either barely significant or they contain some specifications that are not significant. For example, researchers who have early surname initials are significantly more likely to collaborate in economics but not in finance (Kadel & Walter, 2015). Researchers who have early surname initials are found to be significantly more productive only when the sample is restricted to those who have above median performance (Van Praag & Van Praag, 2008).

If the effect of alphabetization on publications and citations is subtle, then, it is natural to expect that the effect of alphabetization on academic careers is also subtle. For example, tenured professors are more likely to have early surname initials in top five institutions, but this result becomes insignificant when lower ranked institutions are considered (Einav & Yariv, 2006).

Our preceding paper Yuret (2016) extends the previous literature by considering nine academic fields. To our best knowledge, all previous studies use one alphabetic field and at most one field that is not alphabetic as a control. Using more academic fields serve as a better control for the results. For example, Efthyvoulou (2008) found that early surname initials are more prevalent in top institutions in economics. Likewise, Yuret (2016) also found some evidence that early surname initials are more prevalent in top institutions in economics. However, the study also found that the result does not extend to mathematics, which is another alphabetic field.

Yuret (2016) covered only a single year of data because the data were collected from faculty web pages of universities. This paper extends Yuret (2016) by using a dataset that covers 100 years. This rich dataset is collected by extracting information from “course catalogs” of the universities. Course catalogs are catalogs that list all courses offered by a university. These catalogs are published regularly to help students choose their courses. They contain essential information such as course descriptions and lists of teaching faculty members. We extracted the list of faculty members and their academic titles from course catalogs.

The analysis was done in two parts. In the first part, the ratio of faculty members who have early surname initials in a given program was computed for a given year. We tested whether early surname initials are more prevalent among full professors in alphabetic fields. For example, suppose that 20 percent of faculty members who are not full professors have early surname initials in a given program, but 50 percent of faculty members who are full professors have early surname initials in that program. Then, we said that early surname initials are more prevalent among full professors in that program.

In the second part, we investigated the entire career path of faculty members in a specific program at a specific university and saw whether faculty members who have early surname initials are more successful in that program. For example, suppose faculty members 1 and 2 have worked in the economics program at the university U. Faculty member 1 is promoted to be full professor in the economics program at the university U, but faculty member 2 has left the economics program at the university U while working as an associate professor. Then, we say that faculty member 1 is more successful than faculty member 2 in that program.

In both parts of the analysis, we tested whether faculty members with early surname initials have been more successful in academic life. However, this does not prove a causal relationship. There are other factors such as culture that influence achievement in academic life. There are many faculty members who are born in other countries and their surnames are specific to that culture. For example, Russians are prevalent in science fields in US universities (Yuret 2017), and their surname initials have a different distribution than those of other cultures. To claim a casual relationship, all these factors should be controlled. The lack of proper control is an important problem in the alphabetization literature. Most studies fail to include many controls because of the data restrictions.

2 Data

Universities issue undergraduate course catalogs regularly. 2 All course catalogs are downloaded from internet sources. Web pages of university libraries, registrar’s offices and internet archive sites such as internetarchive.or make the course catalogs available.

From course catalogs, we attain surnames, 3 academic titles, and academic programs of faculty members. An automated procedure is developed to extract information from the course catalogs. First names and academic titles of faculty members from each of the six programs are copied from each course catalog to a text file. We used 1,345 course catalogs to attain information about six academic programs. Therefore, we repeated this procedure 8,070 (1,345´6) times to attain 8,070 text files. The procedure was easy and fast if the printing quality of the course catalogs was high. If the printing quality was low, an optical character recognition (OCR) program was used to copy information from course catalogs. It was easier to record information manually than copying by OCR if the printing quality was very poor.

The next step was to combine all these text files into a single text file. Programs in Perl were used to clean and unite text files. First names of the faculty members were used to separate two faculty members with the same surname. We also checked for possible misrecordings of surnames and/or mistakes.

In the united text file, each faculty member’s surname and complete career path were laid out. For example, the data contained information such as “Faculty member F works in the economics program in university U as an associate professor in years Y1 to Y2 and as a full professor in years Y2 to Y3”.

The data had certain limitations. For example, we were only able to determine career paths within an undergraduate program. If Faculty member F worked in university U1 and then in university U2, then he or she appeared as two separate observations in the data. Likewise, we were unable to follow the faculty members who moved to another undergraduate program within the same university.

The number of universities included in the dataset was restricted because data extraction procedure was time consuming. We focused on top 100 universities from 2016 US News Survey 4 because research activities are more important in promotion decisions in top universities. In other words, the advantage of having early surname initials was more critical in these universities. Unfortunately, course catalogs were not available for all the universities. Therefore, the data were restricted to 34 universities that provided at least 15 years of course catalogs. The first and second columns of Table 1 list these universities and their rankings, respectively.

Table 1

The List of Universities Included in the Data

NameRankRange of catalogsNo. of catalogsNo. of faculty membersNo. of faculty members (complete career information)
Stanford41919–201695919889
MIT71998–201615477234
Duke81934–201683843806
Northwestern121991–201113478279
Cornell151981–201635530347
Notre Dame182001–201616380189
Rice181968–201645428339
UC Berkeley201917–20118011851149
USC231995–201622407250
UCLA231920–20169111771165
Wake Forest271969–201647250201
Boston Coll301980–201637349231
Chapel Hill301997–201620410199
NYU322000–20169420218
Willam & Mary341939–201677564546
Case Western371977–201622307204
UC Irvine391978–201639529420
UC San Diego392001–201616442214
UC Davis411955–201654594564
Wisconsin411995–201511607307
U Washington521949–201435834749
UT Austin521998–201610633327
Maryland571947–20166611991159
Clemson611936–201676657637
Brigham Young661921–201691588581
American721966–201647525446
Rutgers721997–201510572250
Indiana752000–201612407188
Michigan State752000–201614452182
Iowa821970–201628533396
Binghamton891997–201620251151
Colorado891979–201636574386
Vermont891973–201336287207
Florida State961979–201637545379

The number of academic programs was restricted to six because of the difficulties in data extraction. Three social sciences programs – economics, history, and political science – and three natural sciences programs – chemistry, mathematics, and physics – were selected. All selected programs existed in all 34 universities. They were governed by the same college usually referred as College of Arts and Sciences. Therefore, promotion decisions were given by the same dean in the universities. All programs had a stable structure, that is they did not split and merge to the other programs. Consequently, it was easy to follow the faculty members in these programs. Because course catalogs up to 100 years were used, it is important that the programs had a long history.

An important feature of these programs for this study is whether they belonged to alphabetic fields. Each program had a corresponding academic field. For example, economics program can be matched to economics academic field. Therefore, we were able to retrieve the degree of alphabetization of the programs by looking at the alphabetization in their corresponding academic fields.

Three studies that analyzed degree of alphabetization of academic fields mostly agreed upon their measures although they used different datasets and/or methodologies (Levitt & Thelwall, 2013; Waltman, 2012; Yuret, 2016). Two programs – mathematics and economics – are considered to be alphabetic academic fields. The other three programs – physics, political science, and history – are considered to be “partially academic fields”. In partially academic fields, only some of the publications use alphabetic order in sorting authors. For example, publications in particle physics – a subfield in physics – use alphabetic ordering, but this does not extend all subfields in physics. One should be cautious about history because an average historian does not publish many multiauthor papers (Yuret, 2014). Therefore, the effect of alphabetization on historians’ academic careers is expected to be rather limited. The remaining program that is used in this study – chemistry – is not an alphabetic program. 5

Because we span a long-time period, the degree of alphabetization in the past is also important. Frandsen and Nicolaisen (2010) analyzed the trend for economics and information science for the 30-year period from 1978 to 2007. They found an increase in alphabetization in economics and a decrease in alphabetization in information science. However, the change through time is very small. Unfortunately, we could not find any studies that cover more fields and a longer time range that analyze alphabetization of the fields.

Range and number of course catalogs are given in the third and fourth columns of Table 1, respectively. There were missing years between course catalogs mainly because some universities do not issue course catalogs every year. At most two missing years were allowed between catalogs. If there were more than two missing years, then only years after the missing years were included in the data. 6 When there were missing years between catalogs, the number of faculty members and their titles for the missing years were estimated as described in the appendix. Consequently, all universities were included for each of the years within their range.

Some universities had ceased to issue course catalogs in recent years because the online course catalogs are considered to be sufficient for the purpose. Online course catalogs are continuously updated, so they cannot be used on an annual basis. As a result, we were not able to include some universities for recent years as can been seen from the ranges of the universities.

The penultimate column in Table 1 shows the number of faculty members in the data. Faculty members who had assistant, associate, or full professor titles were included. Postdocs, instructors, emeritus professors, and visiting faculty were not considered. At the end, the data contained 19,353 faculty members who appeared 211,816 times in the course catalogs.

The last column of Table 1 gives the number of faculty members who have complete career information in a specific program at a specific university. If we know the academic title of the faculty member when he or she entered the program and exited the program, then we can say that the complete career information is available. If a faculty member is in the first catalog and/or the last catalog of the university, then his or her complete career information is not available. Suppose we find that faculty member F works as an associate professor at the last catalog. Then, we cannot tell whether he or she is promoted to become full professor in that program because we have no access to further course catalogs. The career information is critical in the second part of the analysis. There were 14,789 faculty members who have complete career information.

3 Methodology

Two separate parts of analysis are used to investigate the role of surname initials. In the first part of the analysis, the ratio of faculty members who have early surname initials is computed for each program and for each year. Therefore, we are able to see whether early surname initials are more prevalent among full professors for a given year and a given program. This part is a simple comparison of ratios, and no other statistical methodology is used. We do not look after statistical significance because the results are not consistent through years. That is, faculty members who have early surname initials are more prevalent among full professors in some years, but not in other years. Therefore, a significant result for a given year is not meaningful.

The second part of the analysis has no time dimension. Each faculty member who has complete career information is grouped according to her career achievement. The ratio of faculty members who have early surname initials within each group is computed. We test whether faculty members who have early surname initials are more prevalent among high achievers. A simple T-test is used for statistical significance when a relation in the expected direction is observed.

4 Longitudinal Analysis

There are two main methods to account for early surname initials. In the first method, the ratio of faculty members who have early surname initials to all faculty members is computed. In the second method, a number is assigned to each letter, and the total number (or total of logarithms) of surname initials is computed. We start with the first method and then use the second method at the end of this section.

There are many possible specifications for an early letter. A, B, C, D and E can all be considered as early letters. Although results for other specifications are computed, results for early initials A to C are reported. The results are found to be the same in alternative specifications as well.

The ratio of faculty members who have early surname initials is computed separately for each program in a given year. For example, the ratio that is computed for economics program in 1985 is different from the ratio for political science program in 1985.

For a given program in a given year, following formulas are computed:

RatiooffullprofessorswhohavesurnameinitialsAtoC=NumberoffullprofessorswhohavesurnameinitialsAtoCNumberoffullprofessorsRatioofassistantandassociateprofessorswhohavesurnameinitialsAtoC=NumberofassistantandassociateprofessorswhohavesurnameinitialsAtoCNumberofassistantandassociateprofessors

If the faculty members who have early surname initials are more likely to become full professors, then the ratio of full professors who have early surname initials should be more than the ratio of assistant and associate professors who have early surname initials. In other words, the following ratio should be more than 1:

RatioAtoC:=RatiooffullprofessorswhohavesurnameinitialsAtoCRatioofassiantandassociateprofessorswhohavesurnameinitialsAtoCRatioAtoC:=RatiooffullprofessorswhohavesurnameinitialsAtoCRatioofassiantandassociateprofessorswhohavesurnameinitialsAtoCRatioAtoC:=RatiooffullprofessorswhohavesurnameinitialsAtoCRatioofassiantandassociateprofessorswhohavesurnameinitialsAtoC

The structure of the universities, the degree of alphabetization, and the importance of research have changed a great deal in the hundred years that we covered. Moreover, availability of course catalogs also varies within hundred years. Therefore, we used three separate figures to lay out the ratios by dividing the time interval into three parts.

Figure 1 shows the ratio of A to C for the time period 1917–1969. There are few faculty members in this period because faculty size in the programs is small and there are few universities that have available catalogs. One can deduce from the ranges of the universities that are reported in Table 1 that there is only one university that has a catalog available for 1917. The number of universities that have catalogs available increases to 13 in 1969. The number of faculty members is just 36 for all six programs in 1917. The number increases to 335 in 1940 and 2082 in 1969. Because the number of observations is small, the ratio of A to C is very volatile in this period.

Figure 1
Figure 1

Ratio of A to C, 1917–1969

Citation: Data and Information Management 3, 2; 10.2478/dim-2019-0006

There are some years before 1927 when the ratio of A to C is equal to zero for some programs. In some cases, there are no full professors who have surname initials A to C in this time period. Then, the ratio of A to C is equal to zero. If there are no assistant/associate professors who have surname initials that are between A to C, then the ratio of A to C is indeterminate. We also label indeterminate values as zero in Figure 1.

There are no consistent results for the prevalence of early surname initials among full professors during this time period. The ratio of A to C for economics is more than 1 in 1943 but less than 1 in 1969. The ratio of A to C for mathematics is more than 1 in the early 1930s but the ratio is less than 1 for most of the years after the 1930s. The ratio of A to C for chemistry –not an alphabetic program – is higher than that of mathematics and economics – alphabetic programs – in most of the 1960s.

Figure 2 depicts the ratio of A to C for the time period 1970 to 2000. The number of universities increases from 14 to 32, and the number of faculty members increases from 2,316 to 6,190 during this period. Consequently, the ratio of A to C is much less volatile in this period.

Figure 2
Figure 2

Ratio of A–C, 1970–2000

Citation: Data and Information Management 3, 2; 10.2478/dim-2019-0006

The ratio of A to C does not have consistent values in this period as well. Economists who have early surname initials are less likely to be full professors in the 1980s, but this result does not extend to early 1990s. Mathematicians who have early initials are more likely to be full professors in the 1970s, but the result does not extend to late 1980s. Historians who have early initials are more likely to be full professors compared to economists and mathematicians in most of the years after the 1970s.

Figure 3 covers the period from 2001 to 2016. All 34 universities have catalogs in 2001, but some universities do not have course catalogs in recent years. There are 6,607 faculty members in 2001, but the number decreases to 5,855 in 2016 because of the decreasing number of universities in the data.

Figure 3
Figure 3

Ratio of A to C, 2001–2016

Citation: Data and Information Management 3, 2; 10.2478/dim-2019-0006

Figure 3 also shows that the ratio of A to C is not consistently more than 1. The ratio of A to C for economics is below 1 in all years except for 2016. The ratio of A to C for mathematics is more than 1 in the 2010s, but the ratio

is less than that of chemistry in this period. The ratio of A to C for political science is more than 1 in the early 2000s but falls less than 1 afterward.

The results that are attained from Figures 1 to 3 suggest that this study may inadvertently find the prevalence of early surname initials among full professors in alphabetic fields if it was restricted to data that cover only a few years. A high prevalence of surname initials shows up in some years for some programs. However, there are no consistent relations when multiple programs and years are considered.

The ratios are also computed for the common ranges. For example, we compute the ratios for four universities that have a common range from 1921 to 2011, for five universities that have a common range from 1949 to 2014, etc. The ratio of A to C is sometimes more than 1 and sometimes less than 1 for the same program when common ranges are used as well.

From Table 1, we see that there are ten universities that are among top 25 in US News rankings. Figure 4 shows the ratio of A to C for these universities only. Economics programs have the ratio of A to C greater than 1 for a single year. Mathematics programs have the ratio of A to C more than 1 in the earlier and later years. Only physics programs have the ratio of A to C more than 1 between 2006 and 2011. Chemistry is one of the two programs that have the ratio of A to C more than 1 after 2013. Therefore, there are no consistent values for the ratio of A to C in top universities either.

Figure 4
Figure 4

Ratio of A to C. Ten universities from top 25, US News, 2001–2016

Citation: Data and Information Management 3, 2; 10.2478/dim-2019-0006

Figures 14 use the ratio of faculty members who have early surname initials in the alphabet. Similar measures are widely used in the literature. This is because the advantage of surname initials is especially important for the very first few letters in the alphabet. However, the ratio has a disadvantage because it groups all letters into just two groups. That is, A, B, and C are treated the same because they are grouped in the early initials group. Likewise, there is no difference between D and Z.

In the literature, two methods are suggested to overcome this problem. In both methods, numbers are assigned to letters in an ad hoc fashion as follows: A=1, B=2, ¼, Z=26. In the first method, the numbers that are assigned to surname initials are simply added up. In the second method, the logarithms of the numbers that are assigned to surname initials are added up. Because the logarithm is a concave function, difference between letters early in the alphabet are more pronounced than difference between letters late in the alphabet. This is reasonable because difference between surname initials A and B is important in terms of being visible but both surname initials Y and Z are almost equally obscure. The outcomes for both methods are computed. However, the results are very similar, so the results from the second method are reported.

The average of the logarithm of surname initials for faculty members for each program in a given year is computed as follows:

Averagelogletterforfullprofessors=TotallogletteroffullprofessorsNumbersoffullprofessors.

Suppose faculty members who have surnames Atkins, Clemens, and Eaton work as full professors in a specific program. Then, average log letter for full professors in that program is (log1+log3+log5)/3.

Averagelogletterforassistantandassociateprofessors=TotallogletterofassistantandassociateprofessorsNumberofassistantandassociateprofessors

Because the logarithm is an increasing function, a higher average means that the distribution of initials is skewed toward the end of the alphabet. Consequently, full professors who have early surname initials are more prevalent if average log letter for full professors is less than average log letter for assistant and associate professors. Therefore, “ratio log letter”, which is defined

in the following, is expected to be more than 1 if faculty members who have early surname initials are more prevalent among full professors:

Ratiologletter=AveragelogletterforassistantandassociateprofessorsAveragelogletterforfullprofessors

Figure 5 gives similar results as Figure 3. For example, both Figures 3 and 5 show that early initials in physics are more prevalent among full professors in most of the years. Likewise, both figures show that early surname initials in mathematics are more prevalent among full professors in earlier and later years. There are also differences. For example, early surname initials in history are less prevalent among full professors in all years in Figure 3, whereas this is not the case in Figure 5. However, the main result from both figures is the same: Early surname initials among full professors are not more prevalent in any of the programs for all the years that we considered.

Figure 5
Figure 5

Ratio log letter, 2001–2016

Citation: Data and Information Management 3, 2; 10.2478/dim-2019-0006

5 Analysis of Faculty Members with Complete Career Information

This part of the analysis does not have a time dimension. We investigated the career paths of each faculty member in a specific program at a specific university. For example, we determined the academic titles that a faculty member from economics program at Stanford University obtained. Career information of each faculty member was grouped into three categories:

  1. 1)“Never full”: Faculty members who had only been assistant or associate professors in the program.
  2. 2)“Promoted”: Faculty members who started as assistant or associate professors in the program, but they had become full professors in later years. In other words, they had been internally promoted within the program.
  3. 3)“Only full”: Faculty members who started as a full professor in the program. Consider a full professor who was in an economics program at MIT but moved to a Stanford economics program as a full professor. Then, we included this faculty member to the “only full” category when Stanford course catalogs were analyzed.

It was necessary that the faculty member was not included in the first or last catalog that was available from a specific program. For example, 1934 was the first year that the Duke course catalog was available. If a faculty member was in the 1934 course catalog as a full professor, then it was not certain whether he or she had been internally promoted within Duke or moved to this university from another university. Consequently, this faculty member was assigned to both “only full” and “promoted” depending on his or her career path before 1934.

The difference between “only full” and “promoted” was important because there is evidence that inside promotion is easier than moving to that program from outside as a full professor (Oyer, 2007). For example, the requirements of a Duke associate professor to become a full professor in that university are less strict than a Brown University’s associate professor to become full professor at Duke.

In this section, we tested whether surname initial distributions were consistent with the assumed difficulty of different career paths. Faculty members in the “never full” category were assumed to have the lowest career achievement. Faculty members in the “only full” category were assumed to have a higher career achievement than those in the “promoted” category because academic requirements for inside promotion were assumed to be lower. Consequently, we tested the prevalence of the early surname initials in the categories in the following ascending order: “never full”, “promoted”, “only full”

Table 2 compares the ratio of faculty members who have surname initials A to C for three categories of faculty members. For example, faculty members who have early surname initials and who are in the “never full” category is 19.37 percent of all the faculty members who are in the “never full” category. A higher ratio means that faculty members who have early initials are more prevalent in that category. Therefore, we predicted that the “never full” category would have the lowest ratio, the “only full” category would have the highest ratio, and the “promoted” category would have the middle ratio.

Table 2

Ratio of Faculty Members Who Have Surname Initials A to C by Academic Achievement

ProgramRatio of faculty members with early surname initials
Never fullPromotedOnly full
Economics0.19370.20110.1980
Mathematics0.17380.18510.1649
Physics0.19070.18820.1757
Political Science0.18090.15840.1744
History0.18410.17790.1991
Chemistry0.19980.18510.1925

The lowest ratio in economics was for those in the “never full” category as expected. We predicted the ratio in the “only full” category to be the highest, which is not satisfied. Moreover, the differences in ratios among the three categories in economics are statistically insignificant. The ratio of the “only full” category in mathematics was the lowest, which is the exact opposite of our prediction.

It can be the case that the advantage of early initials is a current phenomenon because the publication pressure has been felt with more intensity in recent decades. To account for this, Table 3 considers only the faculty members who are first observed after 1980 in our data. The results are even worse for mathematics. The ratios are “never full”, “promoted”, and “only full” in the descending order, whereas we expected this order to be ascending.

Table 3

Ratio of Faculty Members Who Have Surname Initials A to C by Academic Achievement (Restricted to Faculty Members Who Started After 1980)

ProgramRatio of faculty members with early surname initials
Never fullPromotedOnly full
Economics0.19300.20860.2030
Mathematics0.18790.17660.1689
Physics0.19290.18500.1878
Political Science0.17840.15230.1836
History0.17790.13870.1821
Chemistry0.20830.18830.1842

Another possibility is that the importance of early surname initials may be more pronounced in the top institutions. Table 4 considers ten universities in our sample that are top 25 in US News rankings. As expected, the ratio in the “never full” category has the lowest ratio in economics and mathematics. However, the difference between ratios is not statistically significant. Nevertheless, the evidence is suggestive because a similar order is not observed for the other four programs.

Table 4

Ratio of Faculty Members Who Have Surname Initials A to C by Academic Achievement (10 Universities in Top 25)

ProgramRatio of faculty members with early surname initials
Never fullPromotedOnly full
Economics0.19840.20410.2000
Mathematics0.16070.18240.1800
Physics0.17760.19270.1594
Political Science0.19590.15450.1493
History0.20770.18080.2049
Chemistry0.19880.19700.1818

Table 5 compares the average of the logarithm of surname initials for faculty members in different categories of academic achievement. Because higher values are assigned to letters late in the alphabet, a higher average means that faculty members have surname initials skewed toward the end of the alphabet. Consequently, we expected the average to be the highest among faculty members in the “never full” category.

Table 5

Average Log of Surname Initials for all Faculty Members

ProgramAverage log letter
Never fullPromotedOnly full
Economics2.1842.1302.111
Mathematics2.2192.1952.223
Physics2.1682.1622.235
Political Science2.2042.2542.198
History2.1752.1862.142
Chemistry2.1372.1912.147

The results of Table 5 are consistent with those of Table 3. The relation is in the expected direction for faculty members in the economics program. However, the average is the highest for the faculty members who are in the “only full” category in mathematics, which is the opposite of our prediction.

6 Conclusion

A novel dataset was used to investigate the distribution of early surname initials according to career achievement of faculty members. Using course catalogs allowed us to attain surnames and titles of faculty members from many programs and for longer time periods.

We could not find any consistent results. Full professors were more likely to have early surname initials in some years, but the reverse was true for other years. This result shows that it is important to use a longitudinal data for analyzing alphabetization. If this study was restricted to a short time period, then one of the two opposite results could be found. By using a long-time period, we are able to show that faculty members who have early surname initials are not more prevalent in all the years that we studied.

Another advantage of having a longitudinal analysis is that we are able to differentiate between different career paths. A 1-year data is not able to differentiate between an associate professor who will be promoted to a full professor and an associate professor who will not be promoted to full professor. We find that faculty members who have early initials have better academic achievement in economics. However, this result is not significant. Moreover, we also see that using data from multiple programs is important. The faculty members who have early surname initials have no advantage in their academic careers in mathematics, which is also an alphabetic academic field.

This paper does not claim that there is no effect of alphabetization on academic careers. However, the effect is subtle and cannot be observed consistently through time. It can be the case that the visibility advantage is not large enough to increase publications and citations of the researchers to a large extent. Previous research found a positive but small effect of alphabetization on productivity. A small effect of alphabetization on publication performance may not be enough to affect promotion decisions.

Another possible reason that we do not find a positive effect may be because of culture. Yuret (2017) showed that many academics in elite institutions in the United States have a foreign origin. Moreover, academics from different countries are specialized in different academic fields. These cultural factors may hinder a consistent effect of alphabetization on academic careers.

The analysis is done only by using US data in this study. This is a major handicap because different countries have different systems and have different effects of alphabetization. For example, Efthyvoulou (2008) found a positive effect of alphabetization on academic careers in the United States but not in the United Kingdom. If we could do analysis for another country, it is possible that our conclusions about the effect of alphabetization may have changed.

The data collection from course catalogs is advantageous because we can consistently collect information from many programs and from many years. However, it should be noted that being a full professor in a prestigious university is not the only measure of academic achievement. A different conclusion may have been drawn if we redo the analysis by using academic rewards, citation performance, or project funding.

Despite its subtlety, alphabetization effect is popular and widely known. For example, Times Higher Education issued a news report about Yuret (2016), which is our previous paper on the effect of alphabetization on academic careers. 7 Anectodal evidence suggests that colleagues in alphabetic fields are well aware of the alphabetization issue. However, the magnitude of the effect is not well known.

A hiring committee may be too knowledgeable to be concerned about surname initials. Consequently, discrimination against faculty member candidates who have late surname initials is not likely the case. However, a new graduate from a college who has not selected his or her field for graduate studies may be concerned for his or her visibility because of the alphabetization of the field. This study helps him or her by showing that there is no clear effect of visibility on academic careers. Hence, a late surname initial should not be a serious concern.

Computing the probability that the faculty member is in the program in the missing years (used in all figures and tables except for Table 1)

In the longitudinal part of the analysis, the ratio of surname initials A to C and average log letter is computed for all years including the missing years between the catalogs. First, we compute the probability that the faculty member is in the program. Second, the probability is used as weight when the ratio and the average are computed.

Suppose that a faculty member who has surname initial A is in the program with 1/2 probability. Then, half a faculty member is added to the number of faculty members who have surname initials A to C. Probability times log letter (1/2* log(2)) is added to the total log letter. For both the average and the ratio, half a faculty member is added to the total number of faculty members.

The probability is computed as follows:

  1. a)There is one missing year between course catalogs
  2. i)Faculty member is in both course catalogs: Faculty member is assumed to be in the program in the missing year with probability 1.
  3. ii)Faculty member is in one of the course catalogs: Faculty member is assumed to be in the program in the missing year with probability 1/2.
  4. b)There are two missing years between course catalogs
  5. i)Faculty member is in both course catalogs: faculty member is assumed to be in the program in both missing years with probability 1.
  6. ii-)Faculty member is in the first course catalog but not in the second course catalog. Following three scenerios are assumed to have equal probability of 1/3: faculty member is in the program in both missing years, faculty member is not in the program in neither of the missing years, and faculty member is in the program in the first missing year but not in the second year. The fourth possibility – the faculty member is not in the program in the first missing year but he or she is in the program in the second missing year – is possible but not likely to occur and assigned 0 probability. Therefore, the faculty member is assumed to be in the program with probability 2/3 in the first missing year and 1/3 in the second missing year.
  7. iii)Faculty member is not in the first course catalog but he or she is in the second course catalog: the same reasoning applied as in the previous case. Consequently, the faculty member is assumed to be in the program with probability 1/3 in the first missing year and 2/3 in the second missing year.

Deciding the title of the faculty member in the program in the missing years (used in all figures)

  1. a)Faculty member is only in one of the two catalogs: the faculty is assumed to have the same title in the missing year(s) as his or her title in the catalogs.
  2. b)Faculty member is in both catalogs:
  3. i)Faculty member is a full professor in both catalogs: he or she is assumed to be a full professor in the missing year(s).
  4. ii)Faculty member is not a full professor in neither of the catalogs: he or she is assumed to not to be a full professor in the missing year(s).
  5. iii)Faculty member is not a full professor in the first catalog but a full professor in the second catalog: he or she is dropped out of the data for the missing year(s).

References

  • Efthyvoulou, G. (2008). Alphabet economics: The link between names and reputation. Journal of Socio-Economics 37(3), 1266–1285.

  • Einav, L. & Yariv, L. (2006). What is in a surname? The effects of surname initials on academic success. The Journal of Economic Perspectives 20(1), 175-188.

  • Frandsen, T. F. & Nicolaisen, J. (2010). What is in name? Credit assignment practices in different disciplines. Journal of Informetrics 4(4), 608–617.

  • Kadel, A. & Walter, A. (2015). Do scholars in economics and finance react to aphabetical discrimination? Finance Research Letters 14, 64–68.

  • Levitt, J. M. & Thelwall, M. (2013). Alphabetization and the skewing of first authorship towards last names early in the alphabet Journal of Informetrics 7(3), 575–582.

  • Ong, D., Chan, H. F., Torgler, B. & Yang, Y. (2018). Collaboration incentives: Endogenous selection into single and coauthorships by surname initial in economics and management. Journal of Economic Behavior & Organization, 147, 41–57.

  • Oyer, P. (2007). Is there an insider advantage in getting tenure? The American Economic Review 97(2), 501–505.

  • Shevlin, M. & Davies, M. N. O. (1997). Alphabetical listing and citation rates. Nature, 388(6637), 14.

  • Van Praag, C. M & Van Praag, B. M. S. (2008). The benefits of being economics professor A (rather than Z). Economica 75(300), 782–796.

  • Waltman, L. (2012). An empirical analysis of use of alphabetical authorship in scientific publishing. Journal of Informetrics 6(4), 700–711.

  • Yuret, T. (2014). Why do economists publish less? Applied Economics Letters 21(11), 760-762.

  • Yuret, T. (2016). Does alphabetization significantly affect academic careers? Scientometrics 108 (3), 1603–1619.

  • Yuret, T. (2017). An analysis of the foreign-educated elite academics in the United States. Journal of Informetrics 11(2), 358–370.

  • Yuret, T. (2018). Tenure and turnover of academics in six undergraduate programs in the United States, Scientometrics, 116(1), 101–124.

Footnotes

1

Tenure is a contractual right that grants faculty members a permenant position, and this position is prevalent in research universities in the United States. A tenured professor has more academic freedom because his job in the university is safe.

2

The data for this paper and Yuret (2018) are collected at the same time. Both papers use information about the same group of faculty members. Naturally, there are similarities in data description sections of two papers.

3

Faculty members who have one surname in the course catalogs are assumed to use that surname in their publications. The publication records are checked for those who have two surnames. Most faculty members are observed to use both their surnames in their publications.

4

Available online from

5

Among three studies, Yuret (2016) is the only study that analyzes chemistry.

6

Consider a university that has N course catalogs. Then, there are N-1 intervals between these catalogs. Because our data contain 1345 course catalogs and 34 universities, there are 1311 intervals (1345-34). 1310 intervals (86.2 %) do not contain any missing years, 173 intervals (13.2 %) contain one missing year, only 8 intervals (0.6 %) contain two missing years.

7

Available in

If the inline PDF is not rendering correctly, you can download the PDF file here.

  • Efthyvoulou, G. (2008). Alphabet economics: The link between names and reputation. Journal of Socio-Economics 37(3), 1266–1285.

  • Einav, L. & Yariv, L. (2006). What is in a surname? The effects of surname initials on academic success. The Journal of Economic Perspectives 20(1), 175-188.

  • Frandsen, T. F. & Nicolaisen, J. (2010). What is in name? Credit assignment practices in different disciplines. Journal of Informetrics 4(4), 608–617.

  • Kadel, A. & Walter, A. (2015). Do scholars in economics and finance react to aphabetical discrimination? Finance Research Letters 14, 64–68.

  • Levitt, J. M. & Thelwall, M. (2013). Alphabetization and the skewing of first authorship towards last names early in the alphabet Journal of Informetrics 7(3), 575–582.

  • Ong, D., Chan, H. F., Torgler, B. & Yang, Y. (2018). Collaboration incentives: Endogenous selection into single and coauthorships by surname initial in economics and management. Journal of Economic Behavior & Organization, 147, 41–57.

  • Oyer, P. (2007). Is there an insider advantage in getting tenure? The American Economic Review 97(2), 501–505.

  • Shevlin, M. & Davies, M. N. O. (1997). Alphabetical listing and citation rates. Nature, 388(6637), 14.

  • Van Praag, C. M & Van Praag, B. M. S. (2008). The benefits of being economics professor A (rather than Z). Economica 75(300), 782–796.

  • Waltman, L. (2012). An empirical analysis of use of alphabetical authorship in scientific publishing. Journal of Informetrics 6(4), 700–711.

  • Yuret, T. (2014). Why do economists publish less? Applied Economics Letters 21(11), 760-762.

  • Yuret, T. (2016). Does alphabetization significantly affect academic careers? Scientometrics 108 (3), 1603–1619.

  • Yuret, T. (2017). An analysis of the foreign-educated elite academics in the United States. Journal of Informetrics 11(2), 358–370.

  • Yuret, T. (2018). Tenure and turnover of academics in six undergraduate programs in the United States, Scientometrics, 116(1), 101–124.

OPEN ACCESS

Journal + Issues

Search