# Ten Points Up: methods and claim sheet

Publication date: 2026-11-04  
Evidence label: reproducible methods note using constructed scores and a targeted primary-source review

## Provenance

All five score pairs in `examples.csv` are constructed teaching examples. They are not observations of people, Altamens accounts, devices, games or analytics. No random sample or customer data was used.

## Calculations

For starting score `s` and ending score `e`:

- Point change: `e - s`
- Ordinary percent change: `100 * (e - s) / s`, defined only when `s != 0`
- Log-percent change: `100 * ln(e / s)`, defined only when both values are positive

The log-percent calculation follows Cole's sympercent presentation. It is direction-symmetric: swapping start and end changes only the sign. That property does not establish that an arbitrary score is a valid ratio scale.

## Reproduction

Python 3 standard library only:

```text
python3 calculate.py --self-test
python3 calculate.py 40 50
python3 calculate.py 100 50
```

Expected self-test output: `self-test passed`.

## Claim boundaries

- The arithmetic demonstrates denominator dependence, direction asymmetry and undefined zero-baseline percentages.
- It does not estimate learning, practice, reliability, measurement error or meaningful change.
- Methods findings from controlled trials, physiological measurements and assay simulations bound the statistical interpretation. They are not evidence that Altamens changes cognition or everyday outcomes.
- Ordinary percentage, point change and log change answer different questions. No formula is universally preferred without a defensible scale and estimand.

## Review status

Literature search, drafting, calculations, source checks and publication preparation used an OpenAI automated assistant. The calculator and displayed outputs received automated consistency checks. There was no independent human methods review and no journal peer review.