How to read a personal comparison

Three views answer different questions

A game result tells you what happened in one recorded run. A descriptive comparison places that result beside earlier comparable runs. A personal association model asks whether a recorded condition and an outcome varied together across a longer history. These views should not be read as interchangeable evidence.

In Results → Compounds, a score can appear from the first matching observation. It does not need a statistical finding to be worth keeping. When a descriptive reference is available, it uses earlier eligible runs with matching task, scoring, presentation, device and input context. An unavailable reference means the app could not form that comparison. It does not mean the latest result was average.

Count comparable days, not retries

Nootrack keeps task protocols separate. A new scoring rule, a different task presentation or a change in input method can create a different comparison group. A phone touch result is not automatically interchangeable with a desktop keyboard result, even when the game has the same familiar name.

The personal model uses the first eligible observation per day within an exact comparable protocol. Replaying repeatedly on one afternoon does not manufacture a month of evidence. Its support checks also require enough days in each comparison group, a sufficiently long calendar span and overlapping periods. A long history can still be unsuitable when all observations with one condition occurred in a different era.

Read the estimate with its uncertainty

Check the unit before interpreting the direction. A shorter response time and a larger count of correct answers point in different numerical directions. A sleep or heart measurement is not automatically a grade where more is better.

The estimate summarizes the association the model could fit. The interval shows uncertainty under that model and its stated assumptions. If the interval spans both directions, the data do not clearly distinguish them. A narrow interval does not remove errors in logging, missing context or selection of which days were recorded.

Read the evidence status and usable count beside the figure. Nootrack adjusts for several recorded context variables and checks multiple candidate associations together. It can withhold an estimate when the comparison is unsupported. A blank or uncertain result is a legitimate outcome, not a reason to keep changing filters until a preferred story appears.

Keep other explanations in view

The model includes recorded practice history, calendar trend, clock time and other logged compounds, with sleep adjustment when qualifying data are available. These controls work with the information actually recorded. They cannot reconstruct an unlogged intake, establish adherence or turn an observational notebook into a randomized experiment.

Timing matters too. The main association summarizes the observed mix of intake timing within its logging window. That window is not a claim that a substance has cleared from the body. Separate timing and dose sensitivity views have additional eligibility rules and do not identify a best dose or an optimal time to take it.

A short reading checklist

Before saving a conclusion, open the evidence details and ask the questions below. Keep the answer descriptive: “these recorded days differed” is more faithful to this notebook than a treatment claim.

• Which exact task and metric does this result describe?
• How many eligible days support each side of the comparison?
• Did both conditions occur across a shared period?
• What does the interval still allow, and which context was missing?
• Were other compounds, a new device or an unusual routine present?