Interpreting Your Results

Start by reviewing the scatter plot. The data points should cluster closely around the regression line, and the regression line should closely follow the perfect agreement line (y = x). A regression slope near 1.0 and an intercept near 0 (for Design A3) indicate proportional measurement across the range.

Visual patterns to look for:

• Tightly clustered points along the regression line — consistent linearity throughout the measurement range.
• Curvature at the extremes — possible saturation or hook effect near the upper or lower AMR limits. Check whether extreme levels pass the residual criterion.
• Outlying individual points — may represent transcription errors, specimen handling issues, or true imprecision at that level. Review raw replicates before excluding.

The residual plot is the primary tool for assessing linearity level by level. Residuals should be randomly distributed around zero without systematic patterns.

• All residuals within the SEa window — the study passes linearity at all levels. No action is required.
• One or two isolated levels outside the SEa window — investigate those specific levels for sample preparation errors, carryover, or outliers. A single failing level does not necessarily invalidate the study if a documented technical explanation exists.
• Systematic U-shape or S-shape pattern — suggests the measurement procedure has genuine nonlinear response, not random error. Consult the manufacturer or consider a narrower reportable range.
• Residuals that increase with concentration — proportional bias that worsens at high concentrations. The percentage-based SEa criterion at high concentrations will capture this.

Review the status column for each concentration level. All levels must display PASS for the study to demonstrate linearity across the full measurement range. A FAIL at any level warrants investigation before the linearity claim is accepted.

The distribution of data table confirms that samples are spread across the measurement range. Uneven distribution — for example, all samples clustered in the middle of the range — reduces confidence that the extremes of the AMR have been adequately evaluated.

When accuracy analysis is enabled, review the bias between measured means and assigned values at each level. A systematic positive or negative bias across all levels indicates a calibration issue rather than a linearity problem. Bias that grows with concentration suggests proportional error; bias that is constant regardless of concentration suggests a fixed offset.

The AMR verification result confirms whether the study has demonstrated acceptable linearity and accuracy throughout the manufacturer's claimed measurement range. Review the proximity check to confirm that the lowest and highest concentration levels in the study were sufficiently close to the claimed AMR bounds to constitute a meaningful verification of the full range.