Reforming Doctoral Education through the Lens of Professional Socialization to Train the Next Generation of Chemists

Abstract

Several national organizations in the United States have questioned the effectiveness of doctoral education in chemistry in preparing and training doctoral students for their desired professional pathways outside of academia. This study investigates the knowledge and skills that chemists with a doctorate across academic and nonacademic job sectors perceive to be necessary for their careers and the ways that these chemists require and/or value certain skillsets over others depending on their job sector. Based on a prior qualitative study, a survey was distributed to gather knowledge and skills needed by chemists with a doctorate in different job sectors. Findings based on 412 responses provide evidence that many 21st century skills beyond technical chemistry knowledge affect success in many types of workplaces. Further, academic and nonacademic job sectors were found to require different skills. The findings question the learning goals of graduate education programs that exclusively target technical skills and knowledge mastery versus those programs that incorporate concepts from professional socialization theory to broaden their scope. The results of this empirical investigation can be used to illuminate those learning targets that currently are less emphasized, to provide all doctoral students with the greatest opportunities for career success.

Introduction

Doctoral education in chemistry has played a critical role in fueling the economy and solving some of the world’s most challenging issues by providing robust training for the next generation of chemists.¹ Such training plays a pivotal role in the improvement of national innovation and international competitiveness.² However, several national organizations in the United States (U.S.) have questioned the efficacy of the current status of doctoral education in chemistry,³⁻¹⁴ suggesting that it is outdated. The American Chemical Society (ACS) and Chemical & Engineering News have devoted attention and discussion to issues surrounding doctoral education over the decades,¹⁵⁻²² and many educational and governmental bodies have called for changes in doctoral education in the field of chemistry specifically. In 2012, the National Research Council (NRC)¹⁰ and ACS¹ reported that current graduate programs do not provide sufficient preparation for chemistry students’ future careers. Such criticism has been echoed by other national agencies and institutions,^{8−13,23−25} as well as the Obama administration,²⁶ thereby suggesting that changes in educational programs are long overdue to better prepare students for today’s workforce. Details about the primary challenges to chemistry doctoral education are available in a prior literature review.²⁷ Despite the urgent calls to reform doctoral education in chemistry, most of the reports available are not derived from robust, empirical investigations.²⁷ The empirical research undertaken in this study is aimed to fill this gap in the literature and provide direction for updating doctoral education in chemistry to prepare the next generation of chemists with the skills and knowledge they need in the profession.

A common criticism of doctoral education in the U.S. is its lack of attention to professional socialization.²⁸⁻³³ Professional socialization is the process through which doctoral students acquire knowledge and skills as a means to integrate successfully into their chosen profession. As such, professional socialization provides the direction doctoral students need to secure and succeed in a specific job, which is a central objective of doctoral programs. However, doctoral education in chemistry is fundamentally rooted in the cognitive apprenticeship model, whereby students gain knowledge and skills directly from their advisors.³⁴ Research^35,36 has revealed limitations of this model in that it does not sufficiently characterize students as they pursue increasingly diverse career pathways. Additionally, changes in the industrial landscape over the past several decades have led to changes in the chemistry job market and increased demand for chemists in a variety of careers.^27,37 Specifically, the last several decades have witnessed an increasing trend toward nonacademic positions in the chemistry job market.¹⁰ According to U.S. national data, most doctoral recipients in chemistry desire careers in industry versus academia and go on to attain positions in industry after graduation.^38,39,72 However, according to the ACS,³⁹ only about a quarter of chemistry graduate students perceive that they received adequate information about nonacademic career paths and 40% of these students did not feel that their advisors supported their choice of career.³⁹ These, and other sources, have led to the persistent viewpoint over decades⁴³⁻⁴⁷ that doctoral education is primarily or exclusively focused on the development of knowledge and skills that will directly benefit academic research positions over other more prevalent chemical professions.^1,10,40−42

Multiple national reports published over the past two decades point to a mismatch between the knowledge and skills acquired by doctoral students in graduate school and the knowledge and skills needed for the workplace,⁴⁸⁻⁵⁴ but details regarding this phenomenon that are based on empirical evidence are sparse. The consequences of such a disparity in training and work function have resulted in chemists in industry and government anecdotally reporting that graduate schools need to teach more so-called “soft skills”,^55,56 or “21st Century Skills (21CS)”, according to the NRC.⁶⁸ In the general public employment literature, cognitive skills alone have a relatively low predictive ability of job performance in that only 6% of job performance is explained by cognitive skills.⁶⁹ Calls for reform in doctoral training in chemistry as well as science, technology, engineering, and math (STEM) fields support this sentiment.^1,13

A qualitative study conducted previously by our group³⁵ identified 12 broad themes about the knowledge and skills required by doctoral chemists for positions in academia, industry, and government. This earlier work provides empirical evidence from chemists in academia, industry, and government that the requisite knowledge and skills for all professions are heavily defined by 21CS as opposed to technical skills and indicates that chemists in different job sectors appear to prioritize 21CS differently. Specifically, we noted several differences between chemists from academic and nonacademic positions in terms of teamwork, organizational, teaching, and problem-solving skills as well as personal values and growth. Because the results of the earlier study are not generalizable, a quantitative instrument is warranted that aims to draw a national-level picture of the knowledge and skills required by chemists with a doctorate in different job sectors.

In short, the current doctoral education system is preparing doctoral students for a journey using a learning system that does not effectively represent the multitude of potential routes that students might take to reach their desired professional destination. Therefore, updating doctoral education to help the next generation of chemists navigate their career paths by offering equitable doctoral training that incorporates professional socialization theory is essential. In this study, we focused on two primary research questions: (1) What are the knowledge and skills that chemists with a doctorate perceive to be necessary for their careers? and (2) How does the job sector affect the way that academic versus nonacademic chemists require and/or value certain knowledge and skills over other skills?

Theoretical Framework

21CS can be classified into three broad competencies: cognitive (e.g., reasoning, problem-solving, etc.), intrapersonal (e.g., self-regulation, motivation, etc.), and interpersonal (e.g., group behaviors, teamwork, etc.).⁶⁸ Also, in Appendix A of the NRC Report,⁶⁸ the NRC further delineates 15 “specific” skills and competencies that are purported to describe deeper learning, that is, more than knowledge only. However, although the chemistry community is broadly, but not specifically, calling for the focus on 21CS/soft skills in graduate education, the NRC is quick to point out a lack of consensus on any formal definitions of such skills, even for the 15 specific skills it lists as necessary. Despite the lack of formal definitions of skills, we adopted the NRC framework for 21CS in this study because it provides (1) a foundation for the survey in this study that focuses on 21CS as opposed to strictly technical skills, (2) a theoretically, empirically grounded rationale that is largely absent from prior anecdotal calls for reform in chemistry graduate education, and (3) a theoretical validation check of our developed instrument to align reasonably with the 21CS that are loosely defined in the NRC framework. However, we chose not to adopt the NRC framework deductively; that is, we did not use the 21CS as strict definitions, survey items, or reported factors because (1) the NRC framework lacks precise definitions of 21CS, which limits the value of this study’s findings for designers of chemistry doctoral programs and (2), whereas the 21CS framework is intended for the general population, the current study population of doctoral chemists is specific and is a population for whom mastery of technical content is traditionally valued.

Finally, we interpret our findings under the assumption that socialization theory as opposed to the cognitive apprenticeship model will offer graduate students greater opportunities to develop into professional scientists. Socialization is the dominant theoretical framework for doctoral education research in the U.S.⁵⁷ It is the process through which doctoral students acquire knowledge and skills as a means to integrate into a profession, in other words, the process of acquiring a professional identity.^32,58−60 Socialization theory conceptualizes doctoral student development as occurring in four stages: anticipatory, formal (i.e., coursework, guidelines, instruction), informal (i.e., interacting with advisors, peer students, research literature), and personal (i.e., seeing oneself as an independent researcher).⁶¹ It provides a holistic view of ways that doctoral students understand and navigate the program in which they are a part as they progress through graduate school, thereby accounting for all sources of knowledge and skills. Socialization theory involves both cognitive and affective growth as opposed to a strict transfer of knowledge and skills from advisors that is reflected in the cognitive apprenticeship model, which is the traditional model employed in graduate schools in chemistry. Socialization theory collectively incorporates the roles of advisors, peers, and programs as key socialization agents. As a result, identifying these knowledge and skills is critical, as they are central outcomes of the professional socialization process.

Methods

The study was conducted in three stages. Stage 1 is the survey development stage in which survey items that represent the knowledge and skills of chemists with doctorates were developed for a quantitative instrument. Stage 2 is the pilot study in which the instrument was tested and evidence of validity and reliability was gathered. Stage 3 is the deployment of the developed survey to obtain a national-level picture of the knowledge and skills required by chemists with doctorates in different job sectors. All three stages of this study were conducted with approval from the Auburn University Institutional Review Board.

Stage 1: Survey Development

The developed online survey has two parts. The first part includes 73 items to measure the knowledge and skills that chemists with doctorates perceive to be necessary for their careers. The second part contains demographic questions. The survey items in the first part were derived from the literature and the results of our group’s previously conducted qualitative study.³⁵ We used the 12 themes generated from that previous work as the basis to develop survey items within each theme. We combined the skills and subskills identified in our earlier study and the skills reported in other studies to create the survey items. For example, the item “manage time by consciously prioritizing tasks” was supported broadly by the “time management” theme in the qualitative study, and the specific wording was crafted to leverage-specific quotes from this theme, such as “one of the biggest things is actually prioritizing the thing that you have to do”. For item-level validation, we verified that each item could be classified into at least one of the three broad categories (cognitive, intrapersonal, and interpersonal) and 15 specific 21CS skills in the NRC framework,⁶⁸ but most items could be reasonably classified into more than one category. For example, the item “identifying potential sources of funding” could reasonably be classified as cognitive (knowledge of existing sources), intrapersonal (motivation to exhaustively identify sources), and interpersonal (seeking advice from others to learn about sources) and therefore implicates many of the 15 specific classifications as well.

The survey responses are on a six-point Likert scale with a seventh separate option, Not Applicable, as follows: 1 = Strongly Disagree, 2 = Disagree, 3 = Somewhat Disagree, 4 = Somewhat Agree, 5 = Agree, 6 = Strongly Agree, and 0 = Not Applicable. The format for the demographic items is multiple choice for types of job sector (Academia, Industry, Government, Other), gender (Male, Female, Prefer not to answer, and a Self-Identify option), and race/ethnicity (Hispanic/Latino, Black/African American, Native Hawaiian/Pacific Islander, Asian, American Indian/Alaska Native, Other). After we had developed a draft of the survey, we sent it to three chemistry education researchers to gather evidence based on their expert review. We then revised the survey items according to the reviewers’ comments and suggestions.

Stage 2: Pilot Test

To determine whether the survey was well-designed and could achieve the desired results, we piloted it with 60 chemists who had a doctorate to identify any grammatical or syntactical mistakes or other potential problems that needed to be corrected. The respondents selected for the pilot study were broadly representative of different types of chemists in various academic, government, industry, and other positions to attain a purposeful sample.

A total of 19 pilot respondents then participated in a think-aloud validation interview via Zoom (about 30 min) to explain the rationale for their choices. The survey items were reviewed during the interview on the basis of whether they were clear and the extent to which the response options aligned with participants’ beliefs.⁶² Items with wording issues were either rewritten or eliminated from the survey. A total of 70 items were retained from the original 73-item survey, with one modification made based on the interview responses. Appendix B in the Supporting Information provides the full interview protocol.

To evaluate temporal stability, we invited the participants to take the survey again 2 weeks later to determine test–retest reliability, and we obtained 12 retest responses. We then calculated test–retest reliability at the total score level and item score level. At the total score level, all of the survey responses from a single participant were summed as a single score. We used Pearson’s correlation coefficient to assess the test and retest scores. The high correlation coefficient of 0.94 indicates good consistency for the scores over replicate test administrations. At the item score level, we compared each survey item response from the first test to each response to the same survey item from the second test; see Figure S1. Pearson’s correlation coefficient at the item score level was 0.69, which indicates the satisfactory temporal stability of this instrument.

Stage 3: Survey Deployment

We invited chemists from academia, industry, government, and a few other job sectors to participate in the online survey via Qualtrics. Only chemists who held a doctorate degree in chemistry or a related field were included in the project. The survey also was posted on social media via Twitter and LinkedIn. A $10 Amazon gift card was given to participants who completed the survey and validation interview, and a $20 Amazon gift card was given to participants who completed the survey twice as well as the follow-up interview. In addition, all participants were given the 1:500 chance to win $100. The average survey completion time was approximately 10 min.

Sample Plan and Data Collection

We compiled a master list of universities and/or colleges, companies, government agencies, and individuals for data collection. For academic faculty at universities or colleges, we collected the names of 5450 faculty members from 207 institutions for the master list. We selected faculty members from various types of institutions for this study, including 2-year colleges, community colleges, 4-year liberal arts colleges, 4-year comprehensive colleges/universities, and 4-year research colleges/universities. In addition, we collected the names of 297 chemists with doctorates from 11 national laboratories for the master list. However, contact information of chemists in industry, government, and other positions was more difficult to find online, so we contacted 82 representatives from a list of prospective companies that hire chemists with doctorates, including Shell Chemicals, ExxonMobil, Eastman, Dow, Eli Lilly, Pfizer, BASF, and Merck, via LinkedIn, email, or Twitter with a letter that described the research aim and invited participation in the survey. We asked these representatives to distribute the letter of invitation to colleagues and collaborators, which constituted snowball sampling for this study.⁶³

Power analysis revealed that a sample size of 21 per group (job sector) would be required to detect large effects (a = 0.05, h² = 0.4). As we hypothesized large effects, we set this sample size as the target sample size. Table S1 presents the target sample size versus the actual sample size according to the job sector. As the participants should represent the national distribution of broad job sectors, the target number of participants from each sector was determined according to the national distribution of chemists in different sectors and power analysis. The results suggest a total minimum target sample size of 210. Note that our sample is oversampled for academic faculty. Different data analysis methods to handle imbalanced data are described within the specific statistical technique in the Data Analysis section.

Participants

Table S2 shows that a total of 412 valid responses to the survey were received. Surveys that were returned with incomplete or invalid responses were eliminated. Of the 412 respondents, 75% were academic chemists, 17% were chemists from industry, and 8% were chemists from other sectors. Compared to the national distribution of chemists in terms of demographics,^38,64 our sample approximately represents the population of chemists with doctorates, with the exception that it is oversampled for chemists in the academic sector (Table S2). Due to small sample sizes among the individual nonacademic populations (i.e., chemists in industry, government, etc.), we combined the nonacademic populations. However, we recognize the limitations of treating all nonacademic positions as monoliths.

Data Processing

The Qualtrics system automatically recorded the duration of each respondent’s survey completion time with a starting date/time and an ending date/time. Participants were removed if they completed the survey in less than 5 min (1), responded in obvious patterns or completely homogeneously (1), had greater than 90% missing values (6), or failed to respond correctly to two focus check questions (29). After cleaning, a total of 47 missing values (0.17% of the data corpus) remained across all items, seemingly at random. Due to the insignificant amount of missing data, we imputed these responses using the median, despite this approach generally being considered a limited-imputation method.

Data Analysis

Initially, we adopted exploratory factor analysis to elucidate the relationship between different items and constructs and to explore the dimensionality of the developed instrument. We performed Bartlett’s test of sphericity to confirm that the correlation matrix was not an identity matrix (χ² = 12,156.65, p < 0.05), and therefore, that the data were appropriate for factor analysis. We calculated the correlation coefficient to test for multicollinearity in the data. We conducted Mardia’s test to check multivariate normality and the Shapiro–Wilk test to check univariate normality for each survey item. Multivariate and univariate nonnormality (Tables S6 and S7) imply that robust estimation methods should be employed in subsequent factor analyses. We used the principal axis factoring extraction method with oblique rotation for nonnormally distributed ordinal data. We performed the parallel analysis that compared the eigenvalues obtained from the dataset to the eigenvalues generated from simulated data to determine the optimum number of factors to be extracted. After the factor structure was established, we computed Cronbach’s α and McDonald’s omega to determine the internal consistency of each construct. After finalizing the model, we performed measurement invariance tests of the configural, metric, and scalar models to ensure legitimate comparisons between groups.⁶⁷ We used the robust estimator, weighted least-squares mean, and variance adjusted (WLSMV) to handle the categorical data with nonnormality issues.

We performed one-way multivariate analysis of variance (MANOVA) to compare the factor scores of the factors between the job sectors. Specifically, we employed a resampling method based on the wild bootstrap approach using Rademacher weights to handle the imbalanced data and violated MANOVA assumptions. We also conducted linear discriminant analysis to identify factors that could differentiate groups and adjusted prior probabilities of academia and nonacademia to 0.5. To identify the specific factors that contributed to the significant global effect, we conducted one-way analysis of variance (ANOVA) using job sector as the independent variable and the broad knowledge or skill factor score as the dependent variable. Because the ANOVA assumptions were not met, we conducted the nonparametric Kruskal–Wallis test followed by Games–Howell post hoc tests to determine group differences. We Bonferroni-adjusted the significance values to control for multiple comparisons and calculated the Wilcoxon effect size r to determine the magnitude of these significant differences. All statistical analyses were performed using R Version 4.2.1.

Results

Research Question 1

The mean values presented in Table 1 indicate the dominance of high agreement, suggesting that the participants required and/or valued most of the knowledge and skills items listed in the survey. Chemists across academic and nonacademic careers generally felt that most of the skills listed would be required to carry out their daily jobs. We performed factor analysis to elucidate ways that different survey items and constructs relate to one another and to explore the dimensionality of the instrument. Table 1 presents the seven factors that were retained from the parallel analysis (Figure S2); each factor is labeled as a broad knowledge factor or skill category. The factor structure indicates that this instrument is a multidimensional construct. We performed measurement invariance testing to ensure legitimate comparisons between academic and nonacademic job sectors. We also tested two baseline models to ensure that an individual group had a reasonable fit to the model before constraining any parameters to be equal across the two groups. Configural, metric, and scalar invariance were established across academic and nonacademic job sectors, as indicated by the change in fit statistics when moving from one step to another; see Table S3. To determine the internal consistency of each subscale, we calculated Cronbach’s α and McDonald’s omega. Both coefficients show good internal consistency for this instrument (0.70–0.89). Table S4 provides the full survey items within each factor. Table S5 presents the correlation between factors.

Table 1. Cronbach’s α, McDonald’s Omega, Mean, and Standard Deviation for Each Job Sector per Construct (Factor).

factor	loading items with examples	Cronbach’s α	McDonald’s omega	job sector	mean (SD)
securing funding and reputation	items: 44, 47, 41, 45, 2, 3, 46, 6, 48	0.87	0.89	all	4.77 (0.40)
	identify potential sources of funding			academia	4.87 (0.38)
	establish reputation in my field			nonacademia	4.45 (0.55)
group/multiple project management	items: 49, 40, 14, 35, 34, 32, 33, 38, 36, 19, 39, 25, 52	0.78	0.78	all	5.10 (0.68)
	set goals within a group for a project or issue			academia	5.04 (0.77)
	plan out and track the progress of multiple projects			nonacademia	5.29 (0.40)
presenting information	items: 62, 58, 57, 64, 60, 61, 59, 66, 68, 63	0.83	0.82	all	5.53 (0.12)
	communicate topics effectively to different audiences			academia	5.55 (0.14)
	use precise and clear language when conveying information to someone else			nonacademia	5.48 (0.12)
intrapersonal skills	items: 30, 4, 28, 7, 29, 1, 22, 10, 11, 69, 65, 18	0.80	0.81	all	5.03 (0.30)
	do my work in an organized fashion			academia	5.02 (0.27)
	approach new areas and challenges with fearlessness			nonacademia	5.06 (0.46)
technical skills	items: 13, 12, 54, 17, 15, 51, 50, 16, 53, 21, 31, 8, 70, 9	0.82	0.82	all	5.23 (0.31)
	independently design and carry out projects			academia	5.23 (0.32)
	effectively search, retrieve, interpret, evaluate, and synthesize scientific literature			nonacademia	5.21 (0.31)
interpersonal skills	items: 27, 37, 23, 42, 20, 43, 26, 24, 67	0.79	0.80	all	5.13 (0.30)
	actively listen to others			academia	5.25 (0.29)
	be aware of how others are feeling			nonacademia	4.72 (0.59)
ethics and safety	items: 55, 56, 5	0.70	0.70	all	5.35 (0.06)
	minimize safety risks			academia	5.35 (0.04)
	understand ethics as they relate to my projects			nonacademia	5.35 (0.14)

The seven factors that were retained are “securing funding and reputation”, “group/multiple project management”, “presenting information”, “intrapersonal skills”, “technical skills”, “interpersonal skills”, and “ethics and safety”. The 21CS framework influenced the nomenclature of the factor solution and, if followed strictly, all three of the broad competencies (cognitive, intrapersonal, and interpersonal) are present in our factor solution, with four additional factors. Two of the four additional factors (“presenting information” and “group/multiple project management”) can arguably be categorized primarily as interpersonal skills, but the chemist population appears to see them slightly differently than the other factors that were classically identified as interpersonal. With regard to the other two of the four additional factors, the “ethics and safety” factor is not discussed in the 21CS framework; yet, it is important within the chemistry community, and “securing funding and reputation” likely represents an amalgam of all 21CS categories. Generally speaking, the results of the factor analysis highly support our notion that broad categorizations of 21CS do not provide adequate detail in describing how the specific population of chemists sees the importance of requisite skills, but the results do support that the broad categories in the NRC’s 21CS framework are applicable to the chemistry community.

A preliminary interpretation of the average scores is as follows: For the factor “securing funding and reputation”, the subskills include financial management and professional relationship development. The subskills for the factor “group/multiple project management” include goal setting, time management, and people management. The degree of demand for the subskills in these two factors is reversed for chemists from academic and nonacademic careers. For the factor “presenting information”, the mean value is the highest among all of the factors, which indicates that effective oral and written communication skills are deemed essential across both job sectors. However, the audience for researchers in academia (primarily other scientists/chemists), instructors in academia (primarily students), and nonacademic chemists (chemists with various types of academic training) are quite different, likely implicating different communication subskills in different groups. For the factor “intrapersonal skills”, the mean value is also high, which is not surprising, given that most careers in chemistry are deemed professional. Intrapersonal skills are related to introspection and self-management, which include managing behaviors and emotions, weathering challenges, and working toward goals. For the factor “technical skills”, the mean value is very high. The subskills include data analysis, research design, literature searches, and problem-solving, which align with the primary goals of doctoral education in chemistry. For the factor “interpersonal skills”, the subskills are tactics a person needs to interact effectively with others. Compared to the other factors, it elicits the largest difference in average scores between academic and nonacademic chemists. The factor “ethics and safety” includes subskills such as safety management. Participants from academic and nonacademic job sectors equally report that they require the knowledge and skills within this construct, but common consensus suggests a disparity in safety culture between the sectors.⁷⁰

Research Question 2

Rather than seeking an absolute interpretation of the results, the intended purpose of the survey is to compare the degree to which respondents in different job sectors agree that a particular skill is necessary. We performed one-way MANOVA to compare the mean factor scores of the factors/broad knowledge and skills between the academic and nonacademic job sectors. The MANOVA results show a statistically significant overall difference between academia and nonacademia (p < 0.001).

To further explore the specific factors that contribute to the significant global effect, we conducted one-way ANOVA tests. Table 2 presents these results that show significant differences in the factors “interpersonal skills” (r = 0.46), “securing funding and reputation” (r = 0.30), and “group/multiple project management” (r = 0.23) between the academic and nonacademic job sectors. However, for the factors “presenting information”, “intrapersonal skills”, “technical skills”, and “ethics and safety”, no significant differences are evident between academia and nonacademia.

Table 2. One-Way ANOVA Results^a.

factor	F	p.adj	significance	effect size r
interpersonal skills	82.2	8.68 × 10–19	*	0.46
securing funding and reputation	31.8	1.03 × 10–7	*	0.30
group/multiple project management	19.6	4.66 × 10–5	*	0.23
intrapersonal skills	4.88	1.09 × 10–1		0.11
presenting information	3.41	1.94 × 10–1		0.09
ethics and safety	1.61	4.10 × 10–1		0.06
technical skills	0.38	5.39 × 10–1		0.03

^aNote: F is the F value; p.adj is the p-value with Bonferroni adjustment; and the asterisk under “Significance” indicates p-value < 0.05.

The boxplots shown in Figure 1 are grouped by factors, with the two colors representing the academic and nonacademic job sectors, respectively. For the factors “presenting information”, “intrapersonal skills”, “technical skills”, and “ethics and safety”, chemists from both academic and nonacademic careers perceived the knowledge and skills within these four constructs with nearly identical priority. For the factor “securing funding and reputation”, academic faculty gave significantly higher scores than participants from nonacademic job sectors, likely reflecting the necessity of both job security and status in academia.⁷² For the factor “group/multiple project management”, chemists in nonacademic settings gave significantly higher scores for the design and execution of a specific project than participants in academia, likely reflecting the typical multidisciplinary team structure of nonacademic organizations.⁷³ For the factor “interpersonal skills”, academic faculty rated these subskills significantly higher than respondents in nonacademic careers. Note that academic chemists include teaching- and research-emphasis faculty in this study because, in our prior qualitative study, teaching-emphasis faculty members highly valued interpersonal skills.³⁵

Figure 1.

Comparisons of factor scores between academic and nonacademic job sectors.

We conducted the discriminant analysis to identify the relative importance of the factors that best distinguish academic and nonacademic job sectors (Table 3). The discriminant function coefficients indicate the contribution of each factor to differentiate the two job sectors controlling for all other factors. Based on the order of the absolute value of the coefficients, the factor “interpersonal skills” contributes the most distinction between the groups, followed by “group/multiple project management” and “securing funding and reputation”. The factors “ethics and safety”, “presenting information”, and “intrapersonal skills” contribute similarly to group separation. The factor “technical skills” contributes the least to group distinction.

Table 3. Discriminant Function Coefficients.

factor	discriminant function coefficients
interpersonal skills	–1.26
group/multiple project management	0.84
securing funding and reputation	–0.60
ethics and safety	0.26
presenting information	–0.23
intrapersonal skills	0.21
technical skills	0.02

The sign of discriminant function coefficients indicates the direction of the relationship. For factors with negative coefficients (i.e., interpersonal skills, securing funding and reputation, and presenting information), when other factors are held constant, the increase in these factors suggests that the chemist is more likely to be in academia as their average discriminant scores are lower (see Figure 2). For factors with positive coefficients (i.e., group/multiple project management, ethics and safety, intrapersonal skills, and technical skills), the increase in these factors suggests that the chemist is more likely to be in nonacademia as their discriminant score is higher than that of the chemists from academia.

Figure 2.

Comparison of discriminant scores between academic and nonacademic job sectors.

Discussion and Implications

The survey conducted in this study was developed to measure chemists’ perceptions of the knowledge and skills they need for their careers. The results reveal that chemists with doctorates need almost all of the 21CS included in the survey to be successful in their careers. Our study also confirms the findings of general-population research⁶⁸ that shows that employers tend to emphasize 21CS at least as much, if not more, than technical skills. However, doctoral education goals, as articulated in the literature, are centered almost entirely around technical knowledge and the gaining of chemistry mastery, with only a secondary emphasis on 21CS.^1,10 The tension between technical and nontechnical goals is evident in the ACS’s own commissioned report, which states that doctoral education in chemistry “does not provide sufficient preparation for [graduate students’] careers” (pg. 2) and immediately after, reaffirms that doctoral education in chemistry “must manifest traditional depth and must maintain a focus on mastery” (pg. 2).

Whereas our previous study³⁵ suggests that the skills required by chemists from different job sectors might differ, this study provides a more comprehensive assessment tool to quantify the differences between the knowledge and skills needed in academic and nonacademic workplaces. Significant differences are evident between academic and nonacademic job sectors for “securing funding and reputation”, interpersonal skills, and “group/multiple project management”. Combined with the long-standing view that doctoral education has neglected the needs of nonacademic sectors,⁷⁴ our findings paint clear targets for doctoral education reformers to place more focus on learning goals that are valued in nonacademic positions.

The findings in this study provide evidence that chemistry faculty should be cognizant of many other skills and knowledge that affect success in diverse workplaces beyond technical chemistry knowledge and should acknowledge that the learning targets in graduate programs may not currently be aligned with these required skills. This empirical investigation can be used to illuminate learning targets for doctoral education in chemistry that are vital for chemistry careers but that are possibly underemphasized in current doctoral programs. According to our results, the learning goals of doctoral education programs should be broadened and achieved via the mechanisms theorized by professional socialization. The findings imply that graduate programs should provide more opportunities for students to develop 21CS such as communication, project management, and interpersonal skills. For example, the NRC¹⁰ and ACS¹ recommend specific innovations that are not commonly implemented in most graduate programs, many of which would increase the focus on 21CS.

The results also provide evidence that, although chemists in different job sectors require and/or value most of the knowledge and skills identified in this study’s survey, certain job sectors require and/or value specific skillsets. For example, academic faculty members reported the need to secure funding and uphold their reputation through financial management and professional relationship development significantly more than chemists in nonacademic job sectors reported such need. However, these academic faculty members were likely to be research-based academics as opposed to those with exclusively instruction-based jobs. Chemists in nonacademic workplaces reported the need for significantly more group/multiple project management skills, including goal setting, time management, and people management skills, compared to chemists in academia. The findings imply that faculty members who design chemistry doctoral programs should acknowledge that a specific career path may require a specific set of subskills to succeed in that career and should tailor curricula accordingly. Doctoral students will benefit from a curriculum that better prepares them for a wide range of chemistry-related positions. Differences in the ways that chemists with doctorates perceive knowledge and skills as necessary for different job sectors suggest that doctoral education is not one-size-fits-all. Rather, such differences indicate that doctoral education should be reasonably tailored to the individual student’s career development plan and should incorporate concepts from the socialization theory to promote equitable doctoral training for students.

In light of this study, those involved in chemistry doctoral education might wish to consider their Ph.D. programs through the lens of the socialization model³² and incorporate the four stages of the professional socialization process (i.e., anticipatory, formal, informal, and personal) to train the next generation of chemists. The elements of graduate programs must recognize the various ways that students become aware of and integrate into a culture of a given profession. Additionally, the passengers in this doctoral education journey should be acknowledged. That is, input from graduate faculty, chemistry graduate programs, graduate schools, as well as students themselves is vital for doctoral education innovation as they are all key elements in the different stages of the socialization process. The various ways that they communicate and shape students’ values and attitudes about learning different skillsets for different vocations should be considered when (re)designing doctoral programs.

Implications for Doctoral Students

Doctoral students can acquire knowledge and skills explicitly or implicitly from various sources during the four stages of socialization (i.e., anticipatory, formal, informal, and personal) through graduate programs in chemistry. Therefore, faculty should be cognizant of the formal and informal interactions that take place within graduate programs so that students can obtain the required knowledge and skills for specific careers. Additionally, doctoral students should be encouraged to seek broad professional development opportunities to build skills for their future professional environments, especially if these opportunities are not emphasized as part of their given program. Graduate student organizations can sponsor panel discussions or host seminars that feature employers from industry, government, or other job sectors to learn more about the skills students will need and ways to develop them. Connections with employers from diverse job settings also could lead to mentoring programs where students are matched with potential employers.

Implications for Graduate Faculty

As students tend to consult graduate faculty members for career advice, chemistry faculty in doctoral programs should be equipped to guide their students in a fully informed career pathway decision-making process, which would include the types of knowledge and skills that students will need to acquire for their desired careers. Faculty should support their students in pursuing professional development opportunities, even if that means allowing them time away from research activities. Chemistry faculty members can inform students about existing professional society resources, such as the ACS career portal and American Association for the Advancement of Science career training sessions. These resources provide venues for students to grow their careers with training opportunities, networking, and events. Faculty members also should follow individual development plans (ChemIDP)⁶⁵ to ensure that students develop these required skills. Moreover, graduate faculty members in chemistry departments can design their doctoral programs and propose their innovations via seminars, informal retreats, or individual mentoring to promote the knowledge and skills highlighted in this study.

Implications for Chemistry Graduate Programs

Of particular interest and importance to chemistry departments is providing structured opportunities for doctoral students to practice skills and experience different work environments to build their skills in a particular job sector. Graduate program administrators can help doctoral students explore possible career paths and establish networking by incorporating internships into the program and providing informal opportunities to interact with professionals from nonacademic job sectors. In addition, developing cocurricular career exploration and professional development workshops would support chemistry doctoral students. For example, the existing program, Entering Research Curriculum,⁶⁶ provides evidence-based activities for trainees to help them successfully navigate the research environment. Chemistry departments can implement this program or design new programs to accommodate different career needs. Graduate program administrators also should consider more formal connections and relationships with prospective employers. For example, industry career training partnerships, such as the National Science Foundation-funded Accelerate to Industry (A2i) week-long immersion program, can help students develop professional skills and find satisfying careers.

Implications for Graduate Schools

Graduate schools that function in an administrative capacity at universities are in a better position to provide some of the oversight and program support that graduate faculty or chemistry departments alone cannot offer. Graduate schools can provide formal mentoring programs with industry or government partners as well as career counseling or career placement services to doctoral students.

Limitations

Our sample roughly represents the U.S. national distribution of chemists with doctorates. However, academic faculty are overrepresented in the sample because obtaining survey responses from chemists in industry was difficult, as data were collected during the COVID pandemic and lockdown and many individuals likely ignored requests that would add to their work and time burdens. Future study should include perspectives from more chemists with doctorates who are working in industry and other job sectors. Also, we recognize that aggregating all academic positions and all nonacademic positions into respective conglomerates is a limitation as there are undoubtedly differences in the requisite skills among subsectors, organizations, and individual roles.

While this study can serve as the basis for graduate education reform, we did not study graduate programs directly. Future studies including the direct study of the knowledge and skills that doctoral students acquire in graduate programs should be conducted. Additionally, this study is situated in the chemistry discipline; however, other professional areas where PhDs are employed in a range of job sectors could investigate doctoral education and integrate professional socialization into graduate programs.

The Likert-type scale used in the survey treated the variables as continuous rather than categorical. The survey response distribution indicates a dominance of high ratings, leading to multivariate and univariate nonnormality in the dataset. Thus, variables with log transformation or generalized latent variable models may help solve potential issues. Also, the factor analysis used in this study was built upon the analysis of the covariance matrix in the data and assumed linear relationships between items and factors. Factor analysis has limitations, which suggests the use of nonlinear factor analysis in the future.

Conclusions

In this empirical study, we explored the knowledge and skills that chemists with a doctorate across different job sectors perceive to be necessary for their careers. The results show that chemists need most of the knowledge and skills included in the developed survey, and those knowledge and skills reach beyond solely research-related technical skills. Further, chemists in certain job sectors require and/or value specific skillsets. The findings can serve as the basis for improved graduate curriculum design that recognizes different ways to provide students with equitable doctoral training for different career interests. Everyone involved in chemistry doctoral education should consider programs through the lens of the socialization model and incorporate the professional socialization process to train the next generation of chemists.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.2c00561.

• 21st Century skills and competencies, validation interview protocol, test–retest responses, measurement invariance testing results, and factor loading from the factor analysis (PDF)

Author Present Address

Division of Biological Sciences, University of California, San Diego, CA, United States

The authors declare no competing financial interest.