Same joined v2 arrangement. Now each deeper construct is its own little axis-medallion, its outcomes plotted as axis-points at a distance from the part's centre relative to each score, exactly like the HEXACO core's six factors. Each part is sized relative to the core by its cross-variance with HEXACO (how strongly it covaries: values largest, attachment smallest). Parts still connect to every related HEXACO factor (weighted) and link to each other by membranes.
Try LowConnect (very low Emotionality) to see the minimum-size floor keep the connection part legible.
Move any score and watch the core and the four axis-medallions reshape live, beside the image. The deeper tests drive their own axes too.
Working review prototype. Each deeper construct is its OWN axis-medallion: its outcomes (attachment = anxiety/avoidance, drive = BAS/BIS, needs = autonomy/competence/relatedness, values = the four value quadrants) plotted as axis-points at a distance from the part centre by score, like HEXACO's six. Part SIZE = its cross-variance with HEXACO (sum of |r|: values largest, attachment smallest). Parts connect to every related HEXACO factor (weighted) and link to each other by membranes. Deeper sub-scores are sample values; the quiz would supply real ones. Correlations designed, pending validation.