Aggregate
Construct seven descriptive composites by summing the 29 available TCI subscales. Scoring conventions require independent instrument-level verification.
FIELD ATLAS 01 / 07AFFINITREE · 2026AN EXPLORATION IN SEVEN COLORS
BEGIN WITH A SINGLE PORTRAIT
A portrait in
seven dimensions.
What if our differences could be mapped rather than ranked? Affinitree began with seven measurements of temperament and character, then explored the space between one profile and another.
An illustrated seven-axis portrait · no individual research records shown
00 / FROM SNAPSHOT TO STRUCTUREONE PORTRAIT → A LIVING IDEA
Where it started
The first map held still.
Affinitree started with seven measurements, then drew the relationships between those measurements and other profiles. Each response formed a still image, although a life keeps moving.
Affinitree is the origin, not the finished vision. The idea develops into Rhizome ↗, where an evolving network can represent changing relationships and appropriately authorized interactions. This site reports only what the original static data can support.
01 / THE COLOR LANGUAGEA PALETTE WITH PURPOSE
Color is not decoration
Every color carries a trait.
Affinitree never used color as a generic category label. Each hue belongs to one of seven measured dimensions. Together, those colors create a portrait that is unique to its particular combination of values.
Select a dimension to inspect its place in the original taxonomy.
SEVEN-DIMENSION MODEL
SELECTED DIMENSION
Novelty Seeking
Temperament dimension
The seven colors match Affinitree's established trait assignments. A profile is a combination of dimensions, never just one color.
02 / TRY THE MATHEMATICSAN INTERACTIVE INSTRUMENT · SYNTHETIC PROFILES
The shape lab
Change one thread.
Change the whole picture.
Two hypothetical profiles, seven adjustable dimensions. Move the colored sliders and watch the radial portrait change. The distance calculation updates immediately, showing which differences contribute most.
Reference · fixed Variation · editable
A PERSONALITY IS A COMBINATION
Draw a different pattern.
The reference portrait stays fixed. Adjust the second portrait to see how one dimension changes the overall distance.
EUCLIDEAN DISTANCE / NORMALIZED EXAMPLE0.00
This is a geometric difference, not a compatibility score or psychological diagnosis.
CONTRIBUTION TO SQUARED DISTANCE
These scores are synthetic teaching examples on a common 0–1 scale. Nothing is submitted, stored, or associated with a visitor. The research correlations later in the atlas use a separate baseline dataset.
03 / WHAT THE DATA SAYDESCRIPTIVE · EXPLORATORY
Observed relationships
Patterns emerge. Certainty doesn't.
Seven composite scores were reconstructed from 29 subscales in 50 baseline records. The resulting correlations show which traits tended to move together in this cohort, without assigning causes or psychological diagnoses.
50Baseline profiles
29Measured subscales
7Composite dimensions
1,225Profile comparisons
EXPLORATORY DISCOVERIES
9/ 21
Trait correlations below the 5% Benjamini–Hochberg false-discovery-rate threshold.
STRONGEST OBSERVED RELATIONSHIP
PERSISTENCE×SELF-TRANSCENDENCE
r = 0.630
Higher Persistence was associated with higher Self-Transcendence within the baseline sample. The result is descriptive, not a population estimate.
Row and column borders retain each trait's color. Signs and line styles encode association direction.
Pearson correlations for n = 50 baseline profiles. The 21 unique pairs were adjusted together using Benjamini–Hochberg. No respondent-level values are displayed.
Five strongest associations
Ordered by absolute Pearson correlation
04 / THE PROJECTION7 DIMENSIONS → 2 DIMENSIONS
The cost of a simpler picture
What survives when seven become two?
Multidimensional scaling tries to keep the distances between profiles consistent while placing them on a flat surface. It preserves much of the broad structure, but changes some nearest-neighbor relationships.
0.90Spearman rank agreement between original and projected distances
0.203Normalized Stress-1; smaller is better
54.4%Five-nearest-neighbor relations retained
Medians of 1,225 distinct profile-pair distances, grouped into equal-count bins. The diagonal represents perfect agreement. Pairwise comparisons are not statistically independent.
READING THE MAPUseful from afar.
Useful from afar.
Imperfect up close.
The projected map preserved the rank ordering of pairwise distances well (ρ ≈ 0.90), but only 54.4% of five-nearest-neighbor relationships remained unchanged.
Positions are a guide to similarity under one distance model, not permanent identities. A nearby point is not necessarily a close match in every dimension.
CHECK THE DISTORTION BEFORE TRUSTING THE MAP
05 / THE GROUPING QUESTIONASSIGNMENT ≠ DISCOVERY
A useful negative result
Four groups are not four types.
A clustering algorithm will make groups when asked. A silhouette diagnostic tells us whether those groups are well separated. Here, the evidence for discrete clusters is weak.
FOUR-GROUP SILHOUETTE0.160
Weak separation in the seven normalized dimensions. The four named roles are an interface metaphor, not a discovered psychological taxonomy.
Ward hierarchical clustering on normalized composite traits. Higher silhouette scores indicate stronger separation; no external labels were available.
06 / THE NEXT QUESTIONFROM OBSERVATION TO CHANGE
The conceptual lineage
The first map was only a beginning.
Affinitree asked how profiles relate at one moment. Rhizome extends the question toward how relationships, choices, and permissioned interactions might change a person's position over time.
01 / AFFINITREEOne survey, one position
02 / TRANSITIONTime, context, changing signals
03 / RHIZOMEA living, consent-aware network
The evolution described here is a design direction, not a longitudinal finding or a claim that Affinitree records ongoing behavior.
07 / RESEARCH METHODREPRODUCE · QUESTION · REASSESS
Behind the visuals
Show the work. Keep its limits.
The statistics are reproduced from the committed baseline workbook by a standalone Python analysis. The presentation does not ingest questionnaire submissions or publish respondent-level records.
Correlate
Calculate 21 Pearson correlations and control the exploratory false discovery rate with Benjamini–Hochberg adjustment.
Embed
Min-max scale seven composites; calculate Euclidean pairwise distances; fit two-dimensional metric MDS with a fixed random seed.
Evaluate
Assess rank agreement, normalized stress, nearest-neighbor retention, and Ward-clustering silhouettes rather than interpreting shapes at face value.
REPRODUCIBLE RESEARCH
Run the analysis locally
The original workbook stays in data/. The published report uses aggregate outputs only.
python scripts/research_analysis.pyInterpretation boundaries
These results describe 50 baseline records, not a representative population or a validated clinical study. The subscale-to-composite aggregation has not been independently checked against the TCI scoring manual. The statistical tests assume independent profiles. No causal, diagnostic, or population-level claim follows from the correlations. Clustering is descriptive, and the animated network scenes are illustrative rather than observed longitudinal data.