What makes an observation a measurement?
Calibration, a defined scale, uncertainty, model fit, invariance, and a traceable record are not ancillary details. They are what make a quantitative result interpretable.
A scale must be established before it can be used.
Latent measurement uses calibrated items as reference points for locating artificial respondents on a common scale.
In Rasch measurement, respondent location and item difficulty are expressed in the same metric. Their difference determines the modeled probability of a successful response.
Respondents and items share one latent scale.
Rasch measurement provides a deliberately constrained model: one latent dimension, one item-difficulty parameter, and one respondent location. The constraint is scientifically useful because it makes invariance and model fit testable rather than assumed.
The scale is not directly observed. It is inferred from systematic structure in the response data.
Measurements should not depend on accidental particulars of the instrument.
For suitable items measuring the same construct, comparisons among respondents should remain stable. For suitable respondents, comparisons among items should remain stable. Departures from that expectation are scientifically informative and belong in the validity argument for the instrument.
A point estimate is incomplete without its precision.
Every estimated respondent location and item difficulty carries uncertainty. Reporting that uncertainty is part of the measurement, not an optional statistical decoration.
The record must preserve how the result was obtained.
A defensible measurement record identifies the instrument, its version, the respondent configuration, measurement occasion, collection conditions, analysis method, software state, and uncertainty attached to the result.
That record is what allows later results to be compared, reproduced, challenged, or linked to a common scale.