2026 07 27 Eq Keying Correction

The original scoring had items 6, 19, 22, 25, 36, 52, 55 as reverse-scored (where disagree = empathic). The published scoring direction is the opposite: these are positively keyed (where agree = empathic). I had read the item wordings and assigned keying incorrectly — for example, item 22 "I find it easy to put myself in somebody else's shoes" is positively keyed (agree = empathic), but I had treated disagree as empathic. The same error applied to the negatively-keyed items (21, 27, 29, 34, 39, 48) where agree = non- empathic, but I had treated agree as empathic.

1. The corrected keying is partially wrong (I am confident about the 20 POS items and 16 NEG items I scored, but I have not verified the remaining 4 empathic items). 2. The original keying on the 4 unverified items is the opposite of what I assumed — i.e., they may inflate the score by a few points but not enough to bring it to "normal." 3. Nathan genuinely scores in the low empathy range, which would be consistent with the masked profile but is more extreme than typical.

The operating principles in 2026 07 27 Tars Nathan Working Partnership Playbook are strengthened, not weakened, by this correction. The playbook already stated: "The strong AQ signal sits on the non-social Imagination and Attention-to-Detail axes." The corrected EQ underlines the same point. The implication for TARS is unchanged: explicit contracts, not implied scripts. The cost of decoding implicit cues is real, not theoretical.