A Bland-Altman plot is a graphical measurement often used to compare experiment results in analytical chemistry or medical diagnostics. Also called a "Tukey mean-difference plot," it can help analyze the outcome of methods in chemistry as well as in clinical research. Two measurement techniques are typically compared by plotting values for their differences, in relation to calculations of the average of both. A standard x- and y-axis graph is generally used. If it is an accepted standard that another is being compared to, then the differences may be graphed against just a single method.
When new medical diagnostic methods are developed, they are often compared to ones in use. The results and outcome of each can differ, but the reason why is not always clear. A Bland-Altman plot can be used to compare the measurements of test equipment, such as that used to measure respiration. It is often used, therefore, to compare methods that are intended to measure the same thing. These methods can correlate in their results, but the nature of the sample is sometimes the reason for this.
A Bland-Altman plot typically helps researchers know whether the correlation between two methods actually means the results are the same. The plot is usually made by taking the number of samples in each method being compared, and converting them into data points. Each sample is shown as a point that is the mean of the two measurements, which is usually represented by the horizontal x-axis. The difference between each is typically indicated on the y-axis; each coordinate is found by using a mathematical formula.
Average differences can also be reference using horizontal lines on the graph. These typically represent the extent of mathematical agreement between the two methods. The mean difference is usually subtracted from by 1.96 times the standard deviation, another mathematical calculation of the sample results.
Relationships between the differences in measurements can be visualized by a Bland-Altman plot; the magnitude of them is often made apparent as well. Outlying plots can help to assess the consistency of the measurements too. Numbers that are within plus or minus the standard deviation of 1.96 are usually not emphasized; the two methods compared on the Bland-Altman plot, therefore, are typically believed to have similar outcomes in this case. Repeated outcomes of a method can also be identified, often with standard deviation calculations, to compare precision with the size of what is being measured.
