Scientific Methodology

How GLO Generates Leadership Assessments

A transparent explanation of the research foundations, AI methodology, and honest limitations behind our three assessment modules.

Can Personality Be Inferred from a CV or LinkedIn Profile?

The short answer is: yes, with meaningful but bounded accuracy. A substantial body of peer-reviewed research confirms that CVs and professional profiles contain reliable signals about personality and leadership style — not because candidates explicitly disclose their traits, but because career choices, language patterns, role progression, and scope of responsibility are all behavioural expressions of underlying personality.

A landmark study by Apers & Derous (2017) found that trained recruiters can make valid personality inferences from CVs, with Conscientiousness and Extraversion being the most reliably detectable traits. A 2024 study published in the Journal of Applied Psychology confirmed that résumé cues — including job tenure, career progression speed, and the breadth of responsibilities described — correlate significantly with Big Five personality scores obtained through validated questionnaires.

More directly relevant to GLO's approach, a 2023 study in Frontiers in Social Psychology demonstrated that machine learning models trained on CV text can predict Big Five personality traits with accuracy comparable to human raters, and in some cases exceeding them — particularly for Conscientiousness and Openness. A separate study (Kashkin et al., 2024) specifically validated automated LinkedIn and CV analysis for DISC and OCEAN personality determination, finding statistically significant correlations with self-reported questionnaire results.

Key Research Supporting Documentary Assessment

StudyFindingRelevance
Apers & Derous (2017)
Computers in Human Behavior
Recruiters make valid personality inferences from CVs; Conscientiousness and Extraversion most reliably detectedValidates the premise that CVs carry personality signals
Grunenberg et al. (2024)
Frontiers in Social Psychology
ML models predict Big Five from CV text with accuracy comparable to or exceeding human ratersDirectly supports AI-based Big Five inference from CVs
Kashkin et al. (2024)
Semantic Scholar
Automated LinkedIn/CV analysis produces DISC and OCEAN scores correlated with self-reported questionnaire resultsDirectly validates the GLO approach for both DISC and Big Five
Rhea et al. (2022)
PMC / PLOS ONE
Audit of real-world AI personality systems (Humantic AI, Crystal) shows valid personality signals from professional profilesConfirms commercial viability of the documentary assessment approach
Wilmot et al. (2019)
Journal of Applied Psychology
Big Five personality traits predict job performance across roles; Conscientiousness is the strongest predictorEstablishes why personality inference is relevant to hiring decisions

How Each Assessment Module Works

DISC Leadership Profile

The DISC model (Dominance, Influence, Steadiness, Conscientiousness) is one of the most widely validated personality frameworks in occupational psychology, with over 40 years of research and validation studies showing strong construct validity. GLO's AI analyses the candidate's career history for behavioural signals: pace of career progression and scope of authority indicate Dominance; breadth of stakeholder relationships and team-facing roles indicate Influence; tenure length and role stability indicate Steadiness; detail orientation, compliance roles, and analytical responsibilities indicate Conscientiousness.

The AI does not simply label candidates — it extracts specific evidence from the CV text and maps it to DISC dimensions with an explicit rationale, making the inference transparent and auditable. This approach mirrors the methodology used by commercial tools such as Crystal Knows and Humantic AI, which have been independently validated against self-report questionnaires.

GLO Leadership Effectiveness Assessment

The GLO LEA is a proprietary framework built on four dimensions that research consistently identifies as the strongest predictors of senior leadership effectiveness: Problem-Solving Orientation (analytical rigour and strategic thinking), Results Drive (commercial track record and accountability), Openness to Diverse Perspectives (cross-functional collaboration and adaptability), and People Leadership (team development and organisational influence).

These dimensions draw on published leadership research including McKinsey's leadership effectiveness studies, the Korn Ferry Leadership Architect competency model, and Zenger & Folkman's research on extraordinary leaders. Each dimension is scored 1–5 based on explicit evidence in the candidate's career history, with the AI required to cite the specific roles, achievements, or responsibilities that support each score. The radar chart visualisation makes the profile immediately comparable across candidates.

Big Five OCEAN Personality Profile

The Big Five (Openness, Conscientiousness, Extraversion, Agreeableness, Neuroticism) is the most extensively validated personality model in academic psychology, with decades of research confirming its cross-cultural stability and predictive validity for job performance. GLO infers OCEAN scores from career history using the same documentary cues that human raters and validated ML models use: career variety and intellectual breadth (Openness), role tenure and goal achievement (Conscientiousness), leadership and client-facing roles (Extraversion), collaborative and mentoring activities (Agreeableness), and career stability patterns (Neuroticism proxy).

Each trait is rated Low / Moderate / High with a written evidence-based rationale, making the inference transparent. The AI is explicitly instructed to flag where evidence is limited or ambiguous, rather than projecting false confidence.

The AI Engine: GPT-4 with Structured Prompting

GLO uses OpenAI's GPT-4 family of models, which have been shown to perform at or above human expert level on a range of personality inference and text analysis tasks. The key to accuracy is not the model alone, but the structured prompting methodology: each assessment module uses a detailed system prompt that instructs the model to reason from explicit evidence, apply a defined scoring rubric, and produce structured JSON output with mandatory evidence citations.

This approach — sometimes called "chain-of-thought with evidence grounding" — significantly reduces hallucination and projection. The model cannot simply assign a score; it must identify the specific career evidence that supports it. This makes every assessment auditable: a hiring manager can read the rationale and evaluate whether they agree with the AI's interpretation of the evidence.

All assessments are generated fresh from the source documents each time. GLO does not store or reuse personality profiles across candidates, and no training data is derived from user submissions.

Honest Limitations

Transparency requires acknowledging what documentary assessment cannot do. GLO's assessments are inferences from behavioural evidence, not direct measurements of personality. They are not equivalent to validated psychometric questionnaires administered under controlled conditions, and they should not be used as the sole basis for hiring decisions.

What these assessments are not:

  • They are not a substitute for a validated psychometric instrument (e.g., Hogan, NEO-PI-R, or OPQ32) administered directly to the candidate.
  • They cannot detect traits that candidates have not expressed through their career choices or written profile — introverted leaders who are highly effective in their role may appear less Extraverted than they are in practice.
  • They are based on self-reported career history, which may be curated or incomplete.
  • They should be treated as a structured hypothesis to be tested in interviews and reference checks, not as a definitive personality verdict.

GLO's recommended use case is as a structured pre-interview framework: the assessments help hiring teams formulate targeted interview questions, identify areas to probe, and compare candidates on a consistent set of dimensions — not to replace human judgement.

How to Use These Assessments Responsibly

GLO recommends treating each assessment as a structured hypothesis rather than a verdict. The most effective use pattern is to review the assessment before the interview, identify the dimensions where evidence is strongest and weakest, and use the rationale to design targeted competency-based interview questions. If the AI flags limited evidence for a particular dimension, that is itself useful information — it means the candidate's profile does not strongly signal that trait in either direction.

From a GDPR perspective, processing a candidate's CV and LinkedIn profile for assessment purposes requires a lawful basis. GLO's consent checkbox on the report generation form confirms that the user has a legitimate basis to process the candidate's data. Assessments should be shared only with those involved in the hiring decision and retained only as long as necessary for that purpose.

Ready to See It in Action?

Download a full sample report to see how the three assessment modules appear in practice, with evidence citations and radar charts.