Editverse Free Calculator
Bland-Altman Plot Calculator
Assess agreement between two measurement methods — mean bias and 95% limits of agreement, plotted against the average — with the exact R command behind the result. A full interpretation guide follows below. Validated against R 4.3.3.
In Simple Words
A Bland-Altman plot shows how well two methods of measuring the same thing agree. It plots the difference between each paired measurement against their average, then draws the mean difference (the bias) and the limits of agreement — the band within which most differences fall.
The key insight is that a high correlation between two methods does not mean they agree. Two methods can correlate almost perfectly while one reads consistently higher than the other. Bland-Altman answers the question that matters clinically: how far apart can the two methods be for an individual patient?
When to Use Bland-Altman Analysis
Use Bland-Altman whenever you want to know whether a new measurement method can replace an established one, or whether two devices, raters, or assays give interchangeable results. It is the standard for method-comparison and agreement studies in laboratory and clinical research.
Decide in advance what limits of agreement would be clinically acceptable, then judge the observed limits against that threshold — the statistics do not define acceptability for you. For agreement on a continuous scale expressed as a single coefficient, see Lin’s concordance correlation; never substitute a correlation coefficient for an agreement analysis. Unsure which agreement method fits? Our Test Selector can help.
Expected: 2 columns: Method 1 values | Method 2 values
Drop a CSV here or click to choose. Parsed in your browser: nothing is uploaded.
Expected: 2 columns: Method 1 values | Method 2 values
m1 <- c({m1})
m2 <- c({m2})
d <- m1 - m2
c(bias = mean(d), lower = mean(d)-1.96*sd(d), upper = mean(d)+1.96*sd(d))User guide
When to use. Assess agreement between two measurement methods via mean bias and limits of agreement.
Assumptions. Differences are approximately normal and constant across the measurement range.
Common mistakes. Never use the correlation coefficient to assess method agreement.
Worked example
Click Load example data above, then Run Analysis to see a fully worked result with figure, formula and interpretation for this test.
References
- Bland JM, Altman DG. Statistical methods for assessing agreement. Lancet. 1986;1:307-310.
Worked Examples
Example 1 — A small, acceptable bias
Two devices measure the same six samples. The mean difference is about −0.6 units with limits of agreement of roughly −1.4 to +0.2 units. If differences of up to about 1.5 units are clinically unimportant, the devices agree well enough to be used interchangeably.
Example 2 — High correlation, poor agreement
Two methods correlate at r = 0.99 yet one reads consistently 5 units higher. The correlation looks superb, but the Bland-Altman plot reveals a 5-unit bias — the methods track each other perfectly while disagreeing substantially, which correlation alone would hide.
Assumptions to Check
Paired measurements of the same quantity. Each subject is measured once by each method on the same scale, so the difference is meaningful.
Approximately normal, constant differences. The standard limits assume the differences are roughly normal and their spread does not change across the measurement range; if spread grows with magnitude, a transformation or proportional limits are needed.
A pre-defined acceptability threshold. The limits of agreement only become a verdict once compared with a clinically acceptable difference set in advance.
Common Mistakes
Using correlation to judge agreement
A high correlation coefficient says the methods move together, not that they agree. Bland-Altman, not Pearson’s r, is the correct tool.
Ignoring proportional bias
If the difference grows with the size of the measurement, fixed limits of agreement are misleading. Check the plot for a trend and model it if present.
Reading the limits without a clinical threshold
Limits of agreement are only useful against a pre-set acceptable difference. Without that benchmark, “wide” or “narrow” is undefined.
Too few pairs
With very few measurements, the limits of agreement are estimated imprecisely. A confidence interval on the limits, and an adequate sample, are important.
Frequently Asked Questions
Why not just use correlation?
Correlation measures whether two methods move together, not whether they give the same value; two methods can correlate perfectly yet disagree by a constant or proportional amount.
What are the limits of agreement?
They are the mean difference plus and minus about 1.96 standard deviations of the differences, the band within which roughly 95% of differences between the methods fall.
What is proportional bias?
It is when the difference between methods changes with the magnitude of the measurement, shown as a slope in the Bland-Altman plot and requiring proportional rather than fixed limits.
Does the result match R and MedCalc?
Yes. The calculator uses the standard mean-difference and limits-of-agreement formulas and shows the R command so you can reproduce bias and limits exactly.
Related Calculators
References
- Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1(8476):307–310. PMID 2868172.
- Mantha S, Roizen MF, Fleisher LA, Thisted R, Foss J. Comparing methods of clinical measurement: reporting standards for Bland and Altman analysis. Anesth Analg. 2000;90(3):593–602. PMID 10702443.
- Bland JM, Altman DG. Measuring agreement in method comparison studies. Stat Methods Med Res. 1999;8(2):135–160.
Need This Analysis Done for Your Manuscript?
Our PhD biostatisticians run method-comparison and agreement studies — Bland-Altman bias and limits with confidence intervals, plus regression-based comparison — with reproducible R output.