How Jyotirmay measures personality traits (Methods)
Jyotirmay's personality model is designed as a practical, lightweight profiling system that helps personalize daily guidance by translating questionnaire responses into clear trait scores using a standard five-factor structure. This page explains the method in straightforward terms: the trait framework we use, the two inventories we offer, how items are written, how responses are scored and converted into understandable outputs, what information is stored, and the limits of interpretation. The goal is transparency about how personalization works, while making clear that this is a non-clinical trait profile for guidance and self-reflection, not a diagnostic or treatment tool.
Abstract
This document describes Jyotirmay’s personality measurement framework used for personalization of daily guidance. The system estimates stable trait dispositions using a Five-Factor Model (Big Five) structure from brief self-report inventories. The output is a practical trait profile for tailoring non-clinical coaching guidance. It is not intended for diagnosis, treatment planning, risk stratification, or determination of psychiatric disorder. 1. Objective and intended use
Jyotirmay uses trait estimation to:
personalize the form and friction level of behavioral suggestions, select coaching micro-skills more likely to be adopted, reduce one-size-fits-all advice.
Trait estimation is not used to:
infer psychiatric diagnoses, substitute for clinician evaluation, decide medication or therapy plans.
2. Construct model: Five-Factor Model
The system operationalizes personality using five broad domains:
Extraversion Agreeableness
Conscientiousness
Emotional Stability (the inverse pole of neuroticism)
Openness / Intellect (terminology varies; the construct is treated explicitly)
The Five-Factor structure is treated as a measurement model (latent trait approximation) rather than an ontological claim about fixed “types.” Domain scores are continuous.
3. Instruments and administration
3.1 Inventory forms
Two forms are supported:
Short form (≈10 items): rapid initial estimate, low user burden. Long form (≈60 items): higher resolution, improved personalization stability. The system may present the short form as an entry point and recommend the long form when personalization quality is emphasized.
3.2 Response format
Items use a bounded agreement scale (e.g., 5-point Likert). The scale is treated as ordinal with a practical approximation to interval scoring for domain aggregation. 3.3 Item-writing principles
Items are designed to be:
behaviorally anchored (observable tendencies rather than abstract labels), low reading complexity,
balanced across domains and facets,
inclusive of reverse-keyed items to reduce acquiescence bias.
Where multiple languages are supported, the translation process should include:
semantic equivalence checks, avoidance of culturally loaded idioms, and, when possible, back-translation review.
4. Scoring pipeline
4.1 Pre-processing Map responses to numeric values.
Reverse-score designated reverse-keyed items using:
reverse = (max + min) − response
(e.g., for 1..5, reverse = 6 − response)
4.2 Domain aggregation
For each domain:
compute the mean (or sum) across its items,
compute a completeness flag (minimum answered items threshold).
4.3 Scaling for user-facing representation
To present scores consistently, the system transforms raw domain means into a 0–100 scale (monotonic transform), primarily for readability.
Crucially:
The 0–100 scale is not inherently a percentile unless a normative calibration sample is explicitly defined. If norms are later introduced, the page should clearly distinguish raw score, standard score, and percentile.
4.4 Uncertainty and stability indicators (recommended)
For scholarly transparency, the system can compute simple indicators such as:
response consistency flags (straightlining detection), item nonresponse rate,
form length flag (short vs long),
and (if repeated measures exist) test-retest drift metrics. These indicators can be used to qualify the confidence in personalization without implying clinical risk assessment.
5. Psychometric considerations
5.1 Reliability
Short inventories trade reliability for speed; long forms generally improve stability. Internal consistency metrics (e.g., ω/α) can be tracked, but interpretation must respect the small number of items per domain in very brief forms. 5.2 Validity
The framework targets:
content validity (coverage of domain meaning),
construct validity (expected domain interrelations),
predictive utility for tailoring coaching behaviors (the product’s primary aim). Clinical validity is not a design target and should not be inferred.
5.3 Cross-language measurement invariance (recommended)
If multilingual inventories are used for scoring comparability, the evaluation plan should include:
differential item functioning checks where feasible,
domain-level invariance testing across languages,
and monitoring of systematic score shifts attributable to translation. 6. Data storage and privacy boundaries (method-relevant)
For reproducibility and user continuity, Jyotirmay stores:
the user’s trait domain scores and any derived summaries,
the personality inventory responses (as part of the user’s record), and timestamps of administration.
Jyotirmay does not require:
the user’s name or phone number for trait computation. A methodologically clean approach is to store records under a pseudonymous user identifier and to maintain strict separation between identifiers and psychometric data where identifiers exist in other parts of the system.
7. Interpretation limits and clinical boundary statement
Trait profiles are probabilistic summaries of self-report tendencies, not diagnoses. Scores can shift with life context, self-insight changes, and measurement noise. Outputs must not be used to make clinical determinations or to replace psychiatric care.
References
Widiger, T. A. (2019). “The Five Factor Model of personality structure: an update.” (Open access, PMC):
https://pmc.ncbi.nlm.nih.gov/articles/PMC6732674/ (PMC)
McCrae, R. R. (1992). “An introduction to the five-factor model and its applications.” (PubMed record):
https://pubmed.ncbi.nlm.nih.gov/1635039/ (PubMed)
Gosling Lab, University of Texas at Austin — Ten-Item Personality Measure (TIPI) overview and usage context:
https://gosling.psy.utexas.edu/scales-weve-developed/ten-item-personality-measure-tipi/ (Gosling)
Thørrisen, M. M., et al. (2023). “The Ten-Item Personality Inventory (TIPI): a scoping review…” (Open access, PMC):
https://pmc.ncbi.nlm.nih.gov/articles/PMC10330951/ (PMC)
International Personality Item Pool (IPIP) — official site and item bank context:
https://ipip.ori.org/ (IPIP)
