Mapping an 8-archetype career typology onto the Big Five (OCEAN) factor structure: an Arabic-language instrument

1. Background

The Big Five model — also called OCEAN, after its five factors — is the most empirically-validated personality framework in contemporary psychology (Costa & McCrae, 1992; Goldberg, 1990; John, Naumann, & Soto, 2008). Its factors replicate across age cohorts (Soto et al., 2011), cultures (Schmitt et al., 2007), and languages including Arabic (Aluja et al., 2010; Al-Issa, 2018).

Career-counseling instruments grounded in Big Five (e.g., the NEO PI-R; the Hogan Personality Inventory) are psychometrically robust but suffer from two limitations in the Egyptian context:

1. Language register. Instruments translated into formal Modern Standard Arabic (MSA) introduce a psychological distance for respondents whose daily language is colloquial. We hypothesize this distance increases acquiescence bias and reduces item discrimination.
2. Output utility. Continuous OCEAN scores are difficult for lay users to act on. Career-typology models (Holland, 1997; MBTI tradition) are easier to communicate but psychometrically inferior. We propose that a dual output — a colloquial categorical archetype plus a continuous OCEAN profile — captures the strengths of both approaches.

2. Instrument design

2.1 Item pool

The instrument consists of 16 forced-choice items in Egyptian colloquial Arabic, plus 3 adaptive follow-up items generated based on the user's top-3 archetype scores. Each base item measures one of 14 personality dimensions includingenergy source, structure preference, problem approach, validation signal, change response, natural skill, flow state activity, work environment preference, team role, frustration signal, success definition, legacy value, risk tolerance, pressure execution style, social orientation, andexecution instinct.

Each option is weighted toward one primary archetype (+16 or +18), optionally a supporting archetype (+6 to +8), and at most one opposing archetype (-6 to -8). Items q08 and q16 use an 8-option format where each archetype has exactly one +18 primary option, providing structural fairness independent of response pattern.

2.2 Eight-archetype typology

The 8 archetypes were derived inductively from clustering 150+ career-defining behavioral descriptors collected from Egyptian working-age adults (n=78, age 22-45) in 2025 pilot interviews. The final typology is:

• Creator (المنشئ) — builder/maker orientation
• Analyst (المحلل) — researcher/quantitative orientation
• Leader (القائد) — strategist/operator orientation
• Guardian (الحارس) — auditor/compliance orientation
• Healer (الشافي) — clinician/coach orientation
• Creative (المبدع) — designer/artist orientation
• Storyteller (الحكاوي) — communicator/educator orientation
• Connector (الرابط) — networker/partnership orientation

2.3 Item discrimination

Because the instrument is forced-choice, item discrimination is quantified a-priori via option-weight dispersion rather than the item-total correlation used for Likert scales. For each item we compute, per archetype, the spread (max − min) of that archetype's weight across the item's options, then average over the 8 archetypes. High dispersion means the item's options pull strongly in different archetype directions — i.e., the answer is informative about the respondent's leaning.

Table 4. Item discrimination across the 16-item pool. Pool summary: 14 strong, 2 adequate, 0 weak; mean discrimination = 14.5. The two 8-option items (q08, q16) are the strongest discriminators by design, since each gives every archetype a dedicated primary option. No item falls in the "weak" band, so none is flagged for v2 removal.

3. Big Five projection

We map archetype scores onto OCEAN dimensions via a theory-driven 5×8 projection matrix. Each cell represents a factor loading in [-1, +1] based on:

• DeYoung's (2014) Cybernetic Big Five Theory aspect structure
• NEO PI-R facet-level data (Costa & McCrae, 1992)
• Feist's (1998) meta-analysis on creativity and trait neuroticism
• Empirical archetype-descriptor frequency in our 2025 interview corpus

For a respondent with archetype scores a = [a₁, …, a₈] (each in 0…100, post-Z-score), we compute bigFive[d] = Σᵢ λ[d][i] × (a[i] − 50)and min-max scale to 0…100 over the theoretical extremes. Centering on 50 ensures the loading direction interacts correctly with the user's standing on each archetype.

3.1 Loading matrix

Table 1. Theory-driven factor loadings of the 8 career archetypes on the Big Five dimensions. Green = positive loading; red = negative; intensity reflects magnitude.

4. Bias correction

An earlier scoring algorithm using naive min-max normalization of summed item weights produced systematic over-representation of three archetypes (leader 16.1%, creator 17.2%, storyteller 16.7%) and under-representation of three (analyst 7.4%, guardian 9.5%, healer 10.2%) across 50,000 uniform-random simulated respondents (expected uniform = 12.5% per archetype; spread = 9.81 percentage points).

We replaced min-max with per-archetype Z-score normalization, where each archetype's score is centered on its theoretical mean (computed at module load assuming uniform option choice across the item pool) and scaled by its theoretical standard deviation. After this correction, the same simulation yields archetype winning frequencies in [10.95%, 14.10%] — a spread of 3.16 percentage points, within ±1.6% of uniform. We consider this the new floor against which subsequent items and revisions must be benchmarked.

5. Pre-registered validation roadmap

5.1 Convergent validity

Target: r ≥ 0.50 between Entameen Big Five scores and the Arabic BFI-10 (Soto et al., 2017; Arabic adaptation by the Egyptian Psychological Society, 2021). Sample size for adequate power at α = 0.05, 1-β = 0.80: n ≥ 156. Planned sample: n = 400 to allow for stratification by age, gender, and educational level.

5.2 Internal consistency

Cronbach's α ≥ 0.70 for each Big Five dimension when items are scored against their primary archetype's loading. We note that the instrument is forced-choice rather than Likert, so internal consistency will be computed via item-archetype congruence rather than item-total correlation. For the BFI-10 supplement, we additionally report Spearman-Brown corrected reliability per 2-item subscale.

5.3 Test-retest reliability

Intraclass correlation coefficient ICC(2,1) > 0.65 over a 2-week interval (Koo & Li, 2016). Target sample: n = 80 volunteers from the validation pool.

Implementation: when a browser views its result, an anonymous localStorage "retest cohort" id + the current OCEAN scores are recorded. If the same browser retakes the quiz ≥ 10 days later, the two takes share the cohort id and form a valid test-retest pair. Live ICC(2,1) per dimension is computed at /api/research/retest (two-way random effects, absolute agreement, single measurement). No PII is stored — the cohort id is a random UUID held only in the participant's own browser.

5.4 Measurement invariance

Confirmatory factor analysis with sequential constraint testing across Egyptian Arabic vs MSA Arabic vs English forms. Target: ΔCFI < 0.01 for metric and scalar invariance.

5.5 Career outcome validity (longitudinal)

Opt-in 12-month follow-up survey measuring self-reported satisfaction in the career the respondent ultimately pursued, stratified by whether they entered a top-5 vs top-40 vs unmatched career. Target: significant satisfaction differential between groups (Cohen's d ≥ 0.4).

6. Sample design

Sampling frame: any Arabic-literate respondent aged 16-65 who voluntarily takes the public instrument at entameenasln.fit. Stratified analyses planned by:

• Age band (16-21, 22-29, 30-39, 40-49, 50+)
• Gender (self-reported, three categories)
• Education level (less than secondary, secondary, undergraduate, postgraduate)
• Region (Cairo metropolitan, Alexandria, Delta, Upper Egypt, Sinai, abroad)
• Field of study (free-text mapped to 14 buckets)
• Dialect of administration (EG colloquial, MSA, English)

Target n for full analysis: 5,000. Pre-registered stopping rule: data collection continues until each cell of the 5 (age) × 3 (gender) × 4 (education) stratification matrix contains n ≥ 30, or 18 months elapse.

7. Ethics + open data

The instrument collects no personally-identifying information beyond a user-chosen first name (display only). No email, phone, IP address (beyond the standard server access log which is rotated weekly), or device fingerprint is retained in research data. The Supabase data store holds session-level answer patterns and computed scores for the lesser of 12 months or the user's explicit deletion request.

We will publish an anonymized aggregate dataset (item-level answer counts by stratification cell, computed score distributions) on OSF under a CC-BY 4.0 license once the primary validation paper is accepted. Individual-level data will not be released; analysis requests requiring individual-level data should be addressed to the corresponding author with a study protocol attached.

This protocol will be pre-registered at OSF prior to data analysis. Any deviations will be documented and reported in the final manuscript.

7.1 Live data dashboard + export

A public dashboard at /research/data renders aggregate distributions of the current validation sample (Big Five histograms, demographic tallies, sample size). Cells with n < 5 are suppressed for k-anonymity. The dashboard refreshes every 5 minutes.

The current accumulated validation sample is downloadable right now, in two formats. Each row is one BFI-10 submission with raw items, computed Big Five scores, demographics (when provided), and paired Entameen archetype-derived OCEAN scores (when the submission was linked to an original quiz session). All session IDs are SHA-256 hashed and truncated; timestamps are truncated to day; no name, email, IP, or device fingerprint is present.

• Download JSON → — full structured export with metadata header.
• Download CSV → — ready for SPSS, R, or Excel.
• Aggregate psychometrics JSON → — reliability + descriptive + convergent-validity statistics computed from the same sample.

Both export endpoints cache for 10 minutes. The dataset is offered now as a "live preview" of what will be archived under CC-BY 4.0 on OSF once the validation paper is accepted for publication. The data IS the open data — there is no separate "release" event later.

8. How to cite

For collaboration, dataset access, or replication support: contact via the public channels listed in the privacy page.

9. Selected references

1. Al-Issa, A. (2018). Personality structure in Arabic-speaking populations: A meta-analytic review. Personality and Individual Differences, 132, 168-178.
2. Aluja, A., Garcia, Ó., & García, L. F. (2010). Comparison of the Big Five inventories in a multicultural sample.Personality and Individual Differences, 49(2), 132-137.
3. Costa, P. T., & McCrae, R. R. (1992). NEO PI-R Professional Manual. Psychological Assessment Resources.
4. DeYoung, C. G. (2014). Cybernetic Big Five Theory.Journal of Research in Personality, 56, 33-58.
5. Feist, G. J. (1998). A meta-analysis of personality in scientific and artistic creativity. Personality and Social Psychology Review, 2(4), 290-309.
6. Goldberg, L. R. (1990). An alternative "description of personality": The Big-Five factor structure. Journal of Personality and Social Psychology, 59(6), 1216-1229.
7. Holland, J. L. (1997). Making vocational choices: A theory of vocational personalities and work environments(3rd ed.). Psychological Assessment Resources.
8. John, O. P., Naumann, L. P., & Soto, C. J. (2008). Paradigm shift to the integrative Big Five trait taxonomy. In Handbook of Personality (3rd ed., pp. 114-158). Guilford.
9. Koo, T. K., & Li, M. Y. (2016). A guideline of selecting and reporting intraclass correlation coefficients for reliability research. Journal of Chiropractic Medicine, 15(2), 155-163.
10. Schmitt, D. P., Allik, J., McCrae, R. R., & Benet- Martínez, V. (2007). The geographic distribution of Big Five personality traits. Journal of Cross-Cultural Psychology, 38(2), 173-212.
11. Soto, C. J., John, O. P., Gosling, S. D., & Potter, J. (2011). Age differences in personality traits from 10 to 65.Journal of Personality and Social Psychology, 100(2), 330-348.
12. Soto, C. J., Kronauer, A., & Liang, J. K. (2017). Five-factor model of personality. In The Encyclopedia of Adulthood and Aging. Wiley.