Study designs and interpretation
Set Study design according to how the observations, point sets and groups were acquired. The consensus is the mean of the aligned coordinates. Its interpretation depends on which traces represent independent observations.
repeatability
This is the default. Use it for genuinely independent scans of the same feature. Assign a distinct Independent observation ID and Original point-set ID to each scan. One trace produces a profile, with sample spread and SEM unavailable. With at least two traces, inspect the mean, median, sample standard deviation and conditional SEM.
For two scans of the edge of the same bench, SD describes the observed point-by-point variability. SEM is SD divided by the square root of the number of traces declared independent, in the corresponding X or Y unit. A shared calibration can introduce dependence. This SEM is conditional on the independence model, and the service does not propagate the shared component.
calibration_sensitivity
Use this design for the same observation and original points reprocessed using different calibration realizations. Keep the same Independent observation ID and Original point-set ID and give the calibrations distinct identities. The report should show Declared independent observations: 1 even when several aligned traces contribute. The spread describes how the result varies across the calibrations you considered. The SEM for independent replicates is unavailable.
hierarchical
Use it if you have independent real groups, for example two tools, and repeats acquired by group. Each acquisition has its own observation and its own set of points. Control requires at least two groups and more tracks than groups. Two instruments with two independent scans each are a simple design to check.
At each aligned coordinate, the one-factor model estimates within-source spread from deviations from group means. The between-source component uses the differences between group means, correcting for within-group spread and group sizes. Total spread is the square root of the sum of the squares of both components. A negative between-group estimate is reported as zero and flagged as truncated. If the component cannot be identified, its results are missing. The SEM used for repeatability is unavailable for this design.
Mean, spread, SEM and covariances
- Mean and median
- They are calculated, separately for X and Y, between the tracks aligned to the same coordinate. The Consensus chart represents the average. With yx all X-aligned are equal to the grid. A zero X dispersion is therefore normal.
- Sample SD
- Standard sample deviation between tracks, response units and at least two tracks. The Mean − sample SD and Mean + sample SD lines show the average Y minus or more an SD. They describe vertical dispersion in Y, even in the graph of a path. To read the X dispersion, consult the Dataset profile.
- Conditional SEM
- Read it only for the design and assumptions described in repeatability. The average ± SEM limits are descriptive quantities derived from the model. Confidence and prediction intervals are reported as unavailable.
- XY covariance
- Sample covariance between X and Y of the traces at the declared correspondence. Its unit is the product of the two axis units, or the square of the common path unit. Shared calibration covariance is a separate quantity and is reported as unavailable.
- Localization or calibration uncertainty
- Read the above qualification. Unspecified indicates that it is not specified. A known zero uncertainty is a numerical value distinct from a missing field. Digitizer preparation results report the release status of uncertainty for the entire Dataset, without applying automatic weights to the tracks.