Career Pathways in Diversity, Equity, and Inclusion: A Comparative Analysis of Australia and the United States

This study examines career pathways within the Diversity, Equity, and Inclusion (DEI) profession across Australia and the United States, combining descriptive mapping with predictive modelling. Using a dataset of 1000 Australian and 915 United States LinkedIn profiles, we analyse sectoral flows, tenure patterns, and credentialing to identify structural features of DEI careers. Results reveal distinct national contexts: in Australia, DEI careers are anchored in NGO and government sectors, while in the United States, they are concentrated in corporate environments with strong exchanges between firms and academia. Across both countries, careers demonstrate strong intra-field persistence but rely heavily on adjacent entry points from Human Resources, Organisational Development, and Project Management. Predictive modelling using Logistic Regression, Random Forest, and XGBoost demonstrates exceptionally high performance (ROC-AUC 0.99 - 1.0; Average Precision 0.98 - 1.0), highlighting tenure stability, sectoral context, and professional credentialing (e.g., Certified Diversity Executive, Inclusive Leadership) as the most salient predictors of persistence. These findings suggest that DEI careers are not random but institutionally embedded and analytically predictable. Limitations include LinkedIn selection bias, classification sensitivity, and cross-sectional data. Future research should expand to longitudinal and cross-regional datasets, incorporating qualitative perspectives to complement predictive insights. This study contributes both comparative evidence and methodological innovation to the emerging scholarship on DEI careers.

• Diversity
• Diversity and Inclusion
• DEI
• Career Pathway
• Comparative Analysis
• Diversity Equity Inclusion Careers
• DEI Professional Pathways
• Workplace Diversity Careers
• Diversity Careers Australia United States
• Inclusion Career Trajectories

1. Introduction

The rise of Diversity, Equity, and Inclusion (DEI) as a unique area of professional focus represents one of the most notable changes in modern workplace management ([11]). What started as compliance-oriented human resources activities has transformed into strategically integrated practices bolstered by dedicated roles, organisational competencies, and resource allocations that promote successful diversity, equity, and inclusion initiatives ([23]). This formalisation is also evident in the ways organisations publicly demonstrate inclusivity; for example, via leadership biographies, “About Us” sections, and recruitment messaging found on their websites ([12]). This shift has resulted in a completely new professional environment, but the career trajectories within this developing sector are still not well defined.

The growth of DEI as both a central organisational focus and a distinct professional area prompts important inquiries regarding the career paths available within this emerging field. In contrast to well-established professions that have clear entry routes and advancement opportunities ([1]), the career landscape for DEI remains largely ambiguous, especially in various national contexts. This uncertainty carries significant consequences not just for personal career development, but also for the enduring success and impact of diversity efforts within organisations ([17]).

The comparative aspect of DEI careers remains notably under-explored. Much of the current research tends to concentrate on specific national contexts, mainly the United States ([14]), leaving a lack of understanding regarding how various national institutional settings including regulatory policies, cultural values, and industry structures which impact professional paths within DEI. According to [6], the implementation of diversity management practices and their effects are significantly influenced by regional sociocultural and institutional standards, indicating that DEI career trajectories may differ considerably from one region to another. The absence of understanding is especially significant considering that DEI initiatives are expanding their reach globally. However, the way these initiatives are implemented and the career opportunities they generate are frequently shaped by varying institutional pressures, legal structures, and demands for legitimacy, as emphasised in the global case studies from the Diversity, Equity and Inclusion Lighthouses report ([25]).

This study addresses these gaps by conducting the first comparative analysis of DEI career pathways in Australia and the United States—two countries that share linguistic and economic characteristics but operate under different institutional structures for managing diversity ([3]; [9]). By utilising a robust dataset comprising of 1915 LinkedIn profiles, we merge descriptive analysis of career trends with predictive modelling to uncover the structural elements influencing DEI career paths.

Our investigation is framed by four main inquiries:

i) In what ways do career trajectories in DEI differ between Australia and the United States?

ii) Which sectors retain and promote DEI professionals most effectively?

iii) What educational pathways and certifications are associated with long-term DEI careers?

iv) Can predictive analytics identify professionals most likely to enter or exit DEI roles?

This research enhances both theoretical and practical insights into emerging professions. From a theoretical perspective, it supports [19] assertion that professions act as agents of institutional change by showing how DEI roles are increasingly integrated into organisational structures. On a practical level, the results challenge the perception that DEI work is sporadic or temporary, instead uncovering consistent career development patterns and identifiable institutional influences that affect advancement. These findings provide a more structured and evidence-based perspective on DEI as a distinct profession. To situate these developments, the subsequent section reviews the current literature on DEI career pathways, particularly focusing on the national contexts of Australia and the United States.

2. Literature review: DEI Career Pathways across Australia and U.S.

The rise of Diversity, Equity, and Inclusion (DEI) as a distinct professional field reflects broader transformations in workplace culture, governance, and social expectations. As reported by [21], the global DEI market was estimated to be worth around USD 10.5 billion in 2023, with forecasts suggesting a compound annual growth rate of 12.20% until 2032, highlighting an exceptional organisational commitment to diversity initiatives increasingly viewed as key factors for innovation and financial success ([13]; [18]).

This shift in professionalism signifies a crucial change from compliance-focused tasks within human resources to strategic capabilities that necessitate specific expertise and dedicated positions. The formalisation of DEI initiatives has been paralleled by the establishment of professional networks, certification programs, and targeted educational routes, indicating its evolution from a secondary function to a recognised career field ([23]).

The journey into DEI work rarely follows a straight line in either Australia or the U.S. Most professionals find themselves drawn to the field from other areas—perhaps they’ve been working in human resources, organisational development, or corporate social responsibility when an opportunity presents itself ([2]; [10]). While this sideways career shift brings a wealth of diverse experience and skills to DEI roles, it also creates some real challenges. Without clear pathways or standardised job descriptions, practitioners often find themselves navigating a landscape where their job title might be “Diversity Officer” in one organisation and “Inclusion Manager” in another, with vastly different levels of authority and influence ([15]). This uncertainty makes it difficult for professionals to establish their identity within the field and chart a clear course for their career development. Adding to these challenges is the absence of recognised accreditation or professional certification, leaving many DEI practitioners to build their credibility through experience alone rather than formal qualifications that employers and colleagues immediately understand and respect.

By contrast, Australia’s DEI trajectory has been shaped less by corporate activism and more by regulatory and public accountability frameworks. The Workplace Gender Equality Act, which mandates reporting obligations for organisations with over 100 employees, has created a compliance-driven demand for DEI practitioners ([24]). What this means in practice is that Australia’s approach to DEI has grown up largely outside the corporate world. Instead of being driven by business imperatives from company boardrooms, much of the early DEI work has been championed by government departments, universities, and non-profit organisations who saw it as fundamental to creating a fair and inclusive society ([8]; [22]). This has shaped how Australians think about diversity and inclusion, it’s often viewed through the lens of social justice and community building rather than simply as a business strategy. You’re more likely to find seasoned DEI practitioners who cut their teeth working on multicultural policy in local councils, developing inclusive programs at universities, or running community engagement initiatives for NGOs, rather than those who started in corporate human resources departments focused on bottom-line results.

Despite the growing literature, significant gaps remain. Existing studies are largely descriptive, mapping practitioner demographics, role types, and tenure lengths. Far less attention has been paid to how sectoral contexts (e.g., public vs. private vs. NGO vs. academia) shape DEI careers, or how national regulatory frameworks influence professional ecosystems. Even fewer studies have employed predictive or computational approaches to assess determinants of DEI career persistence and exit. This study addresses these gaps by combining descriptive mapping with predictive modelling, providing both comparative insights across the United States and Australia and methodological innovation for DEI scholarship.

Having formal qualifications does seem to help DEI professionals build more stable careers. Australian research by [15] paints a picture of a field where people typically stay in individual DEI roles for about two years before moving on, with most practitioners spending around four years in DEI work altogether before potentially shifting to other areas. To build credibility and sharpen their skills, many professionals are turning to specialised training—courses in leadership development, unconscious bias awareness, and other relevant areas. Yet the field still feels quite disjointed, with no clear pathway for newcomers and surprisingly little involvement from certain academic disciplines that could really strengthen the profession.

The American experience tells a different story, marked by dramatic highs and lows. Following George Floyd’s murder in 2020, there was an unprecedented surge in DEI roles—job listings shot up by more than 120% in just a few months as organisations scrambled to respond ([16]). The numbers were staggering; DEI positions in Russell 3000 companies more than doubled from around 6000 in 2017 to over 13,000 by mid-2022. But what went up quickly also came down. By mid-2024, those numbers had dropped back to about 11,000, and job postings for DEI roles fell by a striking 43% from their August 2022 peak through to July 2024 ([7]; [20]). It’s a sobering reminder of how external events can create both opportunities and instability in this emerging profession.

Recent research has increasingly adopted machinelearning approaches to model career and workforce outcomes, such as predicting employee turnover using supervised algorithms including random forests and gradient boosters, and framing career stage progression as an AI-augmented prediction problem across heterogeneous labour contexts ([4]; [5]).

3. Methodology

3.1. Research Design

This study adopted a mixed-methods computational design to investigate career pathways in the Diversity, Equity, and Inclusion (DEI) profession across Australia and the United States. The research combined descriptive analysis of sectoral flows and tenure dynamics with predictive modelling of career persistence versus exit. The dual approach was intended to

i) capture macro-level patterns of sectoral mobility, and

ii) assess the extent to which DEI career outcomes can be forecasted based on professional, educational, and experiential features.

3.2. Data Collection

Data were collected from publicly available LinkedIn profiles of DEI professionals. Using keyword searches for role titles containing “diversity,” “equity,” “inclusion,” and related variants, profiles were extracted and filtered to retain only those with sufficient longitudinal data (at least one prior and one current role clearly documented, with employment dates and sector information). The final analytical sample comprised 1000 profiles for Australia and 915 profiles for the United States. Personally identifiable information (PII) was excluded, with the dataset anonymised to protect individual privacy.

The final dataset included structured information on:

Current and previous job roles (titles, descriptions, sectors, companies, durations).Education (field of study, highest degree attained).Certifications and professional training.Skills (self-reported LinkedIn skill tags).

From these, derived features such as tenure in months, sector transitions, and role-level DEI classification were constructed.

3.2.1. Representativeness and Coverage Limitations

LinkedIn data introduces systematic sampling bias by over-representing corporate, academic, and formal organisational roles while potentially under-sampling grassroots activists, community-based practitioners, and those in smaller NGOs. LinkedIn adoption is higher among white-collar professionals and varies by sector and geography, meaning the findings may better capture formalised, credential-oriented DEI pathways while underestimating alternative routes through activism or community organising. These limitations affect the generalisability of observed career patterns.

In addition, LinkedIn’s user base reflects uneven demographic and sectoral participation. Platform adoption is higher among professionals in corporate, academic, and public-sector roles, and among individuals with formal qualifications and stable employment histories. As a result, DEI practitioners working in grassroots activism, community organising, informal advocacy, or small community-based organisations are likely to be under-represented in the dataset. This bias may systematically privilege institutionalised forms of DEI work while under-capturing careers grounded in lived experience, volunteerism, or movement-based leadership. Accordingly, observed career pathways should be interpreted as reflecting formal, organisationally embedded DEI professions rather than the full spectrum of diversity and inclusion practice.

3.2.2. Definition of Persistence and Exit

For the purposes of this study, career persistence and exit in Diversity, Equity, and Inclusion (DEI) were operationalised as a binary outcome based on observed role transitions at the time of data collection. Persistence was defined as an individual’s current role being classified as a DEI position, operationalised through (i) the presence of DEI-related keywords in the job title (e.g., “diversity,” “equity,” “inclusion”), and (ii) explicit responsibility for DEI-related functions as described in the role summary. Exit was defined as a transition from a prior DEI-classified role into a current role without an explicit DEI mandate—including general human resources, project management, or other non-DEI functions—or apparent departure from the labour market where no subsequent role was listed.

No minimum tenure threshold was imposed for this classification. Instead, tenure length was retained as a continuous predictor within the modelling framework, allowing persistence and exit to be assessed independently of arbitrary duration cut-offs.

3.3. Feature Engineering

Several preprocessing steps were performed to prepare the data for analysis:

1) Tenure Calculation: Employment durations were standardised into months. Missing values were imputed with medians within sector categories.

2) Sector Mapping: Companies were mapped to one of four primary sectors: Corporate, NGO, Government, Academia, using a manually validated dictionary. This enabled analysis of sectoral inflows, outflows, and retention.

3) Transition Coding: Career moves were classified as DEI → DEI, Non-DEI → DEI (entry), DEI → Non-DEI (exit), or Non-DEI → Non-DEI.

4) Text Features: Current and previous job titles were vectorised using TF-IDF with bigram/trigram settings (ngram_range = (1,2)), capped at 2000 features. Dimensionality reduction was applied via TruncatedSVD (100 components).

5) Education & Certifications: Encoded using one-hot vectors for the highest degree and major certifications (e.g., Certified Diversity Professional, SHRM, Inclusive Leadership).

6) Numeric Features: Current and previous tenure lengths were retained as continuous predictors.

The use of supervised machine learning algorithms for forecasting career outcomes aligns with recent advances in workforce analytics and career trajectory modelling, including systematic reviews of turnover prediction using random forests and other classifiers, and interdisciplinary research into AI’s role in career stage prediction and transition analysis.

3.4. Predictive Modelling

Three supervised machine learning algorithms were implemented to model the likelihood of persistence in DEI careers:

Logistic Regression (L2-regularised): interpretable baseline for text and numeric features.Random Forest Classifier: nonlinear model capturing interaction effects, with 300 estimators and balanced class weights.XGBoost: gradient-boosted trees optimised for tabular + text data, evaluated with logloss.

Models were trained on an 80/20 train-test split with stratification to preserve class balance. The target variable distinguished DEI persistence versus exit.

3.4.1. Data Leakage Prevention

To minimise the risk of data leakage, all preprocessing and feature construction steps were performed strictly within the training data during model evaluation. Train-test splits were created prior to model fitting, and text-based features derived from job titles were vectorised using TF-IDF parameters learned exclusively on the training folds. Dimensionality reduction via TruncatedSVD was likewise fitted only on training data and subsequently applied to held-out observations.

Outcome labels (persistence versus exit) were constructed solely from observed role transitions and did not incorporate tenure length, sectoral retention statistics, or aggregate flow information that could trivially encode the target. Tenure and sector variables were retained as independent predictors rather than classification criteria, reducing the risk of target leakage through definitional overlap.

We note that tenure variables are structurally related to the outcome, as career exit is frequently preceded by shorter current-role tenure. Accordingly, high separability reflects near-perfect discrimination driven primarily by tenure-based temporal signal rather than trivial target encoding.

3.4.2. Hyperparameter Tuning

Hyperparameters for Random Forest and XGBoost models were selected using grid-based search with cross-validation on the training data only. For Random Forest, the number of trees (n_estimators), maximum tree depth (max_depth), minimum samples per split (min_samples_split), and minimum samples per leaf (min_samples_leaf) were tuned to balance predictive performance and overfitting. The final Random Forest model used 300 trees with balanced class weights, while depth and split parameters were constrained to avoid overly complex trees.

For XGBoost, tuning focused on the number of boosting rounds (n_estimators), learning rate (eta), maximum tree depth, subsampling ratio, and column subsampling ratio. Early stopping was applied based on validation-set log loss to prevent overfitting. Hyperparameter selection prioritised stable performance across cross-validation folds rather than maximal in-sample accuracy. All tuning procedures were conducted exclusively within the training data to preserve the integrity of the held-out test set.

3.5. Evaluation Metrics

Model performance was assessed using:

Accuracy.Precision, Recall, F1-score.ROC-AUC (Receiver Operating Characteristic – Area Under Curve).Average Precision (AP) from Precision-Recall curves.

Both ROC and PR curves were plotted for each model. To ensure robustness, performance was also evaluated via k-fold cross-validation.

3.6. Ethical Considerations

The study relied exclusively on publicly available LinkedIn data. Profiles were anonymised, and analysis focused on aggregate patterns rather than individual career trajectories. No personal or sensitive data were retained, and the study adhered to ethical standards of secondary data use in computational social science.

4. Results

4.1. Cross-Country Overview

The comparative analysis highlights distinct centres of gravity in DEI career flows. In Australia, DEI careers are anchored in the non-governmental and public sectors, reflecting the country’s regulatory and civic orientation. In contrast, the United States demonstrates a corporate-centred system in which business functions serve as the primary hub for both entry and exit, with a strong pipeline to and from academia. Despite these national differences, both ecosystems demonstrate high intra-field persistence (DEI → DEI) and adjacency to neighbouring roles in Human Resources (HR), Organisational Development (OD), and Project Management (PM).

4.2. Australia: Sectoral Transitions, Retention, and Stability

4.2.1. Sector Transitions

Analysis of 954 observed transitions indicates that the corporate sector is the largest exporter of talent, while NGOs represent the principal destination. The five most frequent transitions (Table 1) accounted for over 60% of all observed moves.

Table 1. Top career transitions in Australia (n = 954).

Table 1

Transition Path	Count	Percentage
Corporate to NGO	257	26.9%
Corporate to Government	134	14.0%
Corporate to Academia	87	9.1%
Government to NGO	71	7.4%
Corporate to Corporate	56	5.9%

4.2.2. Retention and Net Flow

Sectoral retention rates varied significantly (Figure 1). NGOs demonstrated the highest internal retention (42.5%) and the largest net gain (+309 professionals), while corporate roles exhibited the greatest net loss (−485). Healthcare retention was effectively zero in this snapshot, an artefact of limited sample size and batch profile updates.

4.2.3. Tenure Stability

By dispersion measures (coefficient of variation, CV), corporate careers were the most stable (CV = 7.53), government and academia moderately variable (CV = 10.56 and 9.46), and NGOs the most heterogeneous (CV = 19.05). Academia registered the longest single tenure (102 months), underscoring its potential for career longevity, though corporate remained the most predictable overall (Figure 2).

Interpretation

Australia:Corporate = most stable but median shorter (~24 months).Academia = longest maximum tenure (102 months).NGOs = most variable (short and long roles).United States:Corporate and NGOs both have the highest medians (~35 months).All four sectors converge around a 45-month 75th percentile (≈3.8 years).Corporate shows the lowest variability (CV = 0.55).

Figure 1

Figure 1. Percentage distribution of inflows and outflows of DEI professionals across sectors in Australia and the United States. NGOs in Australia show higher inflow share relative to outflows, producing a strong net gain, while corporate records a large net loss. In the U.S., corporate dominates both inflows and outflows (~50% each), underscoring its role as the systemic hub.

Figure 2

Figure 2. Median and maximum tenure of DEI professionals by sector in Australia (left) and the United States (right). Australian academia supports the longest observed careers, while U.S. corporate and NGO roles provide the most consistent medium-term tenures.

4.2.4. Timing Effects

The majority of records (85% - 100% depending on sector) were coded as “new starters.” This reflects LinkedIn update behaviour rather than actual employment commencements, introducing a systematic bias. Accordingly, emphasis is placed on dispersion and maximum tenure measures rather than raw distributions.

4.3. United States: Sectoral Dynamics and Tenure

4.3.1. Sector Transitions

Among 908 observed transitions, the corporate sector again dominated. Corporate-to-corporate moves comprised a quarter of all flows, followed by a pronounced two-way exchange between corporate and academia (Table 2).

Table 2. Top career transitions in the United States (n = 908).

Table 2

Transition Path	Count	Percentage
Corporate to Corporate	226	24.9%
Corporate to Academia	103	11.3%
Academia to Corporate	92	10.1%
Government to Corporate	72	7.9%
Corporate to NGO	67	7.4%

4.3.2. Inflow and Outflow Balance

The corporate sector accounted for 51% of all exits and 50% of all entries, illustrating its role as a systemic hub. Academia and NGOs balanced at ~15% - 20% each, while government remained marginal (~14%). The net outcome was slightly negative for corporate (−1.1%) and slightly positive for academia (+0.7%), with NGOs and government essentially neutral (Figure 3).

Figure 3

Figure 3. Career transitions: Full distribution of sectoral career transitions in (a) Australia and (b) the United States, expressed as percentages of total observed transitions. Corporate serves as the dominant source in both countries, though NGOs are the main destination in Australia, while the U.S. shows a strong corporate-academia pipeline.

4.3.3. Tenure Stability

Median tenures were longest in corporate and NGOs (35 months), compared to 28 months in academia and government. Across sectors, the 75th percentile was consistently 45 months (3.8 years), suggesting a common ceiling for sustained DEI roles. Corporate careers were the most consistent (CV = 0.55), while government was the least stable (CV = 0.69).

4.3.4. Industry Analysis

The U.S. DEI workforce demonstrated strong concentration in Technology & Electronics (36%), followed by Education & Training (19%), Healthcare & Social Services (15%), and Government/Public Services (14%). This industry concentration contrasts with Australia’s NGO/public-sector profile.

4.4. Role-Level Mobility (Australia)

Within the Australian dataset, 71.9% of transitions were DEI to DEI, indicating strong intra-field persistence. A further 13.4% represented entries into DEI from other functions, and 12.4% represented exits. Functional adjacency analysis confirmed HR, OD, and PM as the most common feeder and destination roles, consistent with the positioning of DEI within organisational “people and change” domains.

4.5. Education, Certifications, and Skills

4.5.1. Australia

Long-tenure practitioners were disproportionately drawn from Sociology (63%), Diversity Studies (60%), Psychology (60%), and Business (59%). The most prevalent certifications were Certified Diversity Professional (CDP) (291), Inclusive Leadership (268), and Unconscious Bias Certification (255). Skills clustering highlighted conflict resolution, stakeholder engagement, human resources, training, and facilitation, reflecting the relational and instructional nature of sustained DEI practice.

4.5.2. United States

Credentialing in the U.S. was more formalised and HR-centric. The most common certifications among long-tenure practitioners were Certified Diversity Executive (CDE) (11.6%), Cornell Inclusive Leadership Certificate (11.3%), and Racial Equity Institute Training (11.2%), alongside HR designations such as SHRM-SCP (10.6%) and SHRM-CP (9.9%). This indicates a credential-driven pathway compared with Australia’s reliance on academic backgrounds in the social sciences.

4.6. Key Comparative Findings

1) Different Anchors: Australia’s DEI workforce is NGO/government-driven, while the U.S. is corporate-anchored.

2) Mobility Patterns: Australia feeds into NGOs; the U.S. shows a strong corporate-academia pipeline.

3) Tenure Lengths: Across both countries, DEI careers are typically 2 - 3 years, with a ceiling at ~4 years.

4) Retention vs. Stability: NGOs retain talent better in Australia, while corporate provides consistency in the U.S. despite net losses.

5) Industry Context: The U.S. workforce is tech-heavy, contrasting with Australia’s civic orientation.

6) Education vs. Credentialing: Australia favours social science qualifications, while the U.S. relies on certification-based legitimacy.

4.7. Limitations

The results should be interpreted with caution. First, reliance on LinkedIn data introduces both temporal and selection bias: many profiles were updated recently, creating a skew toward “new starters” and artificially inflating apparent turnover. Second, some sectors—particularly healthcare, social enterprise, and “other” classifications—were under-represented, limiting the generalisability of sector-specific conclusions. Third, coding ambiguities remain, as some DEI roles appeared under alternative titles, making the analysis sensitive to classification decisions.

Because tenure length is intrinsically linked to observed persistence, predictive performance should be interpreted as reflecting strong temporal regularities rather than causal determinants of DEI career sustainability.

Beyond these dataset-specific concerns, broader methodological constraints apply. The use of LinkedIn profiles creates selection bias, privileging digitally visible professionals while under-representing grassroots or community-based DEI work. The analysis was also cross-sectional, limiting our ability to capture the dynamic evolution of DEI careers over time. Furthermore, the exceptionally high predictive scores (ROC-AUC approaching 1.0) warrant cautious interpretation given the strong temporal coupling between tenure and observed persistence. Finally, the study focused exclusively on Australia and the United States; findings may not readily extend to other regions with different institutional, cultural, or labour market contexts.

4.8. Future Work

Future research should address these limitations in several ways. First, expanding data sources beyond LinkedIn—such as professional associations, government labour statistics, or organisational records—would mitigate digital visibility bias and provide a more representative sample. Second, longitudinal tracking of DEI roles would enable analysis of career persistence over time and in response to external shocks (e.g., pandemic-related contractions, legislative changes).

Third, applying the predictive models to additional regions (e.g., Europe, Asia-Pacific, or the Global South) would support cross-cultural comparisons, revealing how institutional, legal, and cultural contexts shape DEI career sustainability. Fourth, integrating qualitative methods such as practitioner interviews, case studies, or organisational ethnographies would enrich the quantitative findings with experiential dimensions of DEI work. Finally, future work could explore advanced modelling approaches (e.g., survival analysis, temporal graph models) to predict not only whether professionals remain in DEI, but also when and under what conditions transitions are likely to occur.

5. Prediction

Predictive Model Performance on DEI Career Outcomes

Table 3

Model	ROC-AUC	Average Precision (AP)	Accuracy
Logistic Regression	1.00	1.00	1.00
Random Forest	1.00	1.00	0.99
XGBoost	0.99	0.98	0.99

Note: Results averaged across train-test splits. All models exhibit extremely high discriminatory power, with ROC-AUC and AP scores approaching unity.

Figure 4

Figure 4. ROC Curves for DEI career predictions: Receiver operating characteristic (ROC) curves for logistic regression, random forest, and XGBoost applied to the Australian and United States datasets. both logistic regression and random forest achieve perfect separation (AUC = 1.00), while XGBoost demonstrates near-perfect performance (AUC = 0.99).

As shown in Figure 4, both Logistic Regression and Random Forest achieve perfect separation with ROC-AUC scores of 1.00, while XGBoost demonstrates near-perfect performance with an AUC of 0.99.

Precision-Recall Curves for DEI Career Predictions

Precision-Recall curves across the three models (see Figure 4) highlight consistent predictive power, with average precision scores ranging from 0.98 to 1.00. These findings indicate that tenure, sectoral transitions, and credentialing provide strong signal for modelling DEI career persistence.

Interpretation of Results

This study examined career trajectories within the Diversity, Equity, and Inclusion (DEI) profession across Australia and the United States, integrating descriptive sectoral analyses with predictive modelling. The findings demonstrate that DEI careers exhibit both structural regularities and significant contextual variation between the two countries.

At the descriptive level, Australia’s DEI workforce was found to cluster around the NGO and government sectors, reflecting the influence of public accountability frameworks such as the Workplace Gender Equality Act. By contrast, the U.S. dataset revealed a corporate-centric ecosystem, with high volumes of exchange between firms and academia, consistent with corporate responses to governance pressures and social justice movements. In both contexts, DEI careers displayed strong intra-field persistence, though entry pathways from HR, Organisational Development, and Project Management remained prominent.

The predictive modelling component extended this descriptive mapping by quantifying determinants of DEI career persistence versus exit. All three classifiers—Logistic Regression, Random Forest, and XGBoost—exhibited exceptionally high discriminatory performance (ROC-AUC = 0.99 - 1.00; AP = 0.98 - 1.00). Although such near-perfect results warrant cautious interpretation, they reflect near-perfect discrimination driven primarily by tenure-based temporal signal rather than causal determinants of DEI career sustainability. Specifically, tenure stability, sectoral context, and professional credentialing emerged as the most salient predictors. For instance, longer current-role tenure and possession of certifications such as Certified Diversity Executive or Inclusive Leadership Training aligned strongly with sustained DEI engagement. Sectoral features also differentiated outcomes, with corporate-to-NGO or corporate-to-government transitions more likely to signal exits, whereas intra-academic or NGO-based moves suggested greater retention potential.

These results advance understanding of DEI careers beyond purely descriptive accounts by demonstrating that career sustainability in this field is not random, but predictable when sectoral positioning, tenure histories, and professional development are taken into account. Importantly, the inclusion of textual features (role descriptors) via TF-IDF confirmed the embeddedness of DEI within people-oriented and change-management roles such as “manager,” “consultant,” “advisor,” and “officer.” This reinforces existing scholarship positioning DEI alongside HR and organisational change functions, while also highlighting the institutionalisation of DEI within leadership and consultancy pathways.

Nonetheless, several limitations temper these findings. First, the reliance on LinkedIn data introduces potential selection bias, privileging professionals with public, digitalised career histories while under-representing grassroots or community-based DEI work. Second, the perfect or near-perfect classification scores raise the possibility of overfitting or residual feature leakage despite careful feature curation. Third, the analysis is cross-sectional, capturing career states at a given moment rather than fully dynamic longitudinal trajectories.

Future research could address these limitations through larger, longitudinal datasets incorporating multiple platforms, enabling the testing of temporal models that capture career flows over time. Additionally, comparative studies across other regions (e.g., Europe, Asia-Pacific, or Global South contexts) would contextualise how national policy frameworks and cultural environments shape DEI career sustainability. Finally, qualitative interviews with practitioners could complement predictive models by capturing lived experiences of burnout, professional recognition, and advancement barriers, providing a fuller picture of DEI career dynamics.

Taken together, the study demonstrates that DEI career outcomes are empirically modellable and structurally contingent on sector, tenure, and credentialing. By bridging descriptive and predictive approaches, this research contributes both methodological innovation and substantive insights to the emerging scholarship on DEI as a professional field.

6. Conclusion

This study provides the first comparative, predictive analysis of DEI career pathways across Australia and the United States, combining descriptive mapping of sectoral transitions with machine learning models of career persistence and exit. The findings reveal both convergence and divergence across national contexts: Australia’s DEI workforce is anchored in NGO and government institutions, whereas the United States exhibits a corporate-dominated ecosystem with significant exchanges between firms and academia. Despite these contextual differences, both countries display strong intra-field persistence and reliance on adjacent entry pathways from HR, Organisational Development, and Project Management.

The predictive modelling component demonstrated that DEI career sustainability is not random but structured around identifiable drivers. Tenure length, sectoral context, and professional credentialing emerged as consistent predictors of persistence, with certifications such as Certified Diversity Executive and Inclusive Leadership aligning with longer tenure. High model performance (ROC-AUC and AP scores approaching 1.0) suggests strong predictive signal, though further work is needed to mitigate potential overfitting and extend analyses beyond LinkedIn data.

By bridging descriptive and predictive approaches, this research contributes to the evolving literature on DEI careers in three ways: it establishes cross-country evidence of sectoral dynamics, identifies concrete predictors of career longevity, and demonstrates the value of predictive analytics in career studies. Future work should broaden the scope to other regions, incorporate longitudinal tracking, and triangulate with qualitative insights to capture the lived experiences behind observed patterns. Overall, the study underscores that DEI careers are both institutionally embedded and analytically predictable, offering a foundation for more sustainable workforce strategies in diversity, equity, and inclusion.

Acknowledgements

The authors gratefully acknowledge the guidance and feedback of Cultural Infusion team. Especially Peter Mousaferiadis, Michael Walmsley, Catherine McCredie, Aida Hakemi, Branquinho, Quincy Hall, My Linh Le, Nabi Zameni and Elizabeth Mousaferiadis whose expertise and mentorship were invaluable in shaping the research design and interpretation of results. Special thanks are extended to peers and colleagues who provided constructive discussions and encouragement throughout the project.

Appendix

Additional Visualisations

Industry level Analysis (AU)

Figure 5

Industry Level Analysis (US)

Figure 6

AU Sector Transition

Figure 7 Figure 8 Figure 9 Figure 10 Figure 11

US Sector Transition

Figure 12 Figure 13 Figure 14 Figure 15 Figure 16

References

1.. Abbott, A. (1988). TheSystemofProfessions. University of Chicago Press. https://doi.org/10.7208/chicago/9780226189666.001.0001 10.7208/chicago/9780226189666.001.0001https://doi.org/10.7208/chicago/9780226189666.001.0001 Abbott, A. 1988 10.7208/chicago/9780226189666.001.0001

2.. Ahmed, S. (2012). OnBeingIncluded. Duke University Press. https://doi.org/10.1515/9780822395324 10.1515/9780822395324https://doi.org/10.1515/9780822395324 Ahmed, S. 2012 10.1515/9780822395324

3.. AHRI (2023). Diversity,EquityandInclusioninAustralian Workplaces:2023Report. https://www.ahri.com.au/wp-content/uploads/DEI-Report-2023.pdf Diversity, E 2023 2023

4.. Al Akasheh, M., Malik, E. F., Hujran, O., & Zaki, N. (2024). A Decade of Research on Machine Learning Techniques for Predicting Employee Turnover: A Systematic Literature Review. ExpertSystemswithApplications,238, Article 121794. https://doi.org/10.1016/j.eswa.2023.121794 10.1016/j.eswa.2023.121794https://doi.org/10.1016/j.eswa.2023.121794 Akasheh, M. Malik, E. Hujran, O. Zaki, N. 2024 121794 10.1016/j.eswa.2023.121794

5.. Bankins, S. (2024). Navigating Career Stages in the Age of Artificial Intelligence: A Systematic Interdisciplinary Review and Agenda for Future Research. JournalofVocationalBehavior,153, Article 104011. https://doi.org/10.1016/j.jvb.2024.104011 10.1016/j.jvb.2024.104011https://doi.org/10.1016/j.jvb.2024.104011 Bankins, S. 2024 104011 10.1016/j.jvb.2024.104011

6.. Barak, M. M. (2022). ManagingDiversity (5th ed.). Sage Publications, Inc. https://www.perlego.com/book/4792519/managing-diversity-toward-a-globally-inclusive-workplace-pdf Barak, M. Publications, I 2022

7.. Colvin, C. (2024). DEI Job Postings Are Down. Why Do These Roles Lack Longevity? HRDive. https://www.hrdive.com/news/dei-job-postings-are-down-why-do-these-roles-lack-longevity/724241/ Colvin, C. 2024

8.. Culture Plus Consulting (2025). TheMaturationofDiversity,EquityandInclusioninAustralia:FromAdvocacytoStrategy. https://cultureplusconsulting.com/the-maturation-of-the-dei-profession-in-australia-from-advocacy-to-strategy/ Diversity, E 2025

9.. Diversity Council Australia (2023). CareerTrajectoriesinDiversity,Equity & Inclusion. https://www.dca.org.au/news/blog/career-trajectories-in-dei Diversity, E 2023

10.. Dobbin, F., & Kalev, A. (2016). Why Diversity Programs Fail: And What Works Better. HarvardBusinessReview,94, 52-60. https://hbr.org/2016/07/why-diversity-programs-fail Dobbin, F. Kalev, A. 2016

11.. Dobbin, F., & Kalev, A. (2022). GettingtoDiversity:WhatWorksandWhat Doesn’t. Harvard University Press. Dobbin, F. Kalev, A. 2022

12.. Hou, Z., Lu, S., Lee, N., & Moieni, R. (2025). VirtualDiversity Assessment:EnhancingInclusivenessofOrganisational Websites. Diversity Atlas. https://atlas.culturalinfusion.com/virtual-diversity-are-you-showing-your-diversity-commitment-online/ Hou, Z. Lu, S. Lee, N. Moieni, R. 2025

13.. Hunt, V., Prince, S., Yee, L., & Dixon-Fyle, S. (2018). DeliveringthroughDiversity. McKinsey & Company. https://www.mckinsey.com/business-functions/organization/our-insights/delivering-through-diversity Hunt, V. Prince, S. Yee, L. Dixon-Fyle, S. 2018

14.. Kalev, A., & Dobbin, F. (2023). Getting to Diversity: How Democratizing Career Systems Can Open Opportunity. ThePractice. https://dobbin.scholars.harvard.edu/publication/getting-diversity-how-democratizing-career-systems-can-open-opportunity-0 Kalev, A. Dobbin, F. 2023

15.. Lee, N., Moieni, R., & Mousaferiadis, P. (2023). Diversity and Inclusion Career Path Analysis. OpenJournalofSocialSciences,11, 114-141. https://doi.org/10.4236/jss.2023.119009 10.4236/jss.2023.119009https://doi.org/10.4236/jss.2023.119009 Lee, N. Moieni, R. Mousaferiadis, P. 2023 10.4236/jss.2023.119009

16.. Minor, K. (2023). Three Years after George Floyd’s Murder: Where Is DEI Now, and What Have Companies Learned? SeniorExecutive. https://seniorexecutive.com/three-years-after-george-floyds-murder-where-is-dei-now-and-what-have-companies-learned/ Minor, K. 2023

17.. Ng, E. S., & Sears, G. J. (2020). Walking the Talk on Diversity: CEO Beliefs, Moral Values, and the Implementation of Workplace Diversity Practices. JournalofBusinessEthics,164, 437-450. https://doi.org/10.1007/s10551-018-4051-7 10.1007/s10551-018-4051-7https://doi.org/10.1007/s10551-018-4051-7 Ng, E. Sears, G. Beliefs, M 2020 10.1007/s10551-018-4051-7

18.. Roberson, Q. M. (2019). Diversity in the Workplace: A Review, Synthesis, and Future Research Agenda. AnnualReviewofOrganizationalPsychologyandOrganizationalBehavior,6, 69-88. https://doi.org/10.1146/annurev-orgpsych-012218-015243 10.1146/annurev-orgpsych-012218-015243https://doi.org/10.1146/annurev-orgpsych-012218-015243 Roberson, Q. Review, S 2019 10.1146/annurev-orgpsych-012218-015243

19.. Suddaby, R., & Viale, T. (2011). Professionals and Field-Level Change: Institutional Work and the Professional Project. CurrentSociology,59, 423-442. https://doi.org/10.1177/0011392111402586 10.1177/0011392111402586https://doi.org/10.1177/0011392111402586 Suddaby, R. Viale, T. 2011 10.1177/0011392111402586

20.. Tornone, K. (2025). 3 Charts that Show What Has Happened to DEI Roles-and DEI Pros. HRDive. https://www.hrdive.com/news/3-charts-to-explain-dei-roles-dei-pros-changes/760763/ Tornone, K. 2025

21.. UnivDatos (2024). GlobalDiversityandInclusion Market Report2024-2032. UnivDatos Market Insights. https://univdatos.com/reports/diversity-and-inclusion-di-market 2024

22.. University of Melbourne (2022). EngagingDiverse Students:DiversityandInclusion Audit Report. https://melbourne-cshe.unimelb.edu.au/__data/assets/pdf_file/0004/4143721/diversity-and-inclusion-audit-report-June-2022.pdf 2022

23.. Wickham, M. D., Fiedler, T., & Donnelly, T. I. (2025). Addressing the Diversity Paradox: A Resource‐Based View of Dei Best Practices. SustainableDevelopment,33, 6260-6275. https://doi.org/10.1002/sd.3467 10.1002/sd.3467https://doi.org/10.1002/sd.3467 Wickham, M. Fiedler, T. Donnelly, T. 2025 10.1002/sd.3467

24.. Workplace Gender Equality Agency (2025). Reporting:WhatYour Organization NeedstoDo. https://www.wgea.gov.au/ 2025

25.. World Economic Forum (2024). Diversity,EquityandInclusion Lighthouses2024. https://www3.weforum.org/docs/WEF_Diversity_Equity_and_Inclusion_Lighthouses_2024.pdf Diversity, E 2024