Using SPSS to produce Bland Altman plots
Using the data on the previous page we illustrate, step-by-step, how to use SPSS syntax and its graphing editor to produce Bland-Altman plots. Note: you will need to adjust the file names and variable names and their graph labels in the below accordingly for using on data on your own PC.
We firstly make sure we have the data in a saved SPSS spreadsheet. We only need wright1 and mini1 for our calculations.
DATA LIST LIST/ subject wright1 wright2 mini1 mini2 (5 F8.0). BEGIN DATA 1 494 490 512 525 2 395 397 430 415 3 516 512 520 508 4 434 401 428 444 5 476 470 500 500 6 557 611 600 625 7 413 415 364 460 8 442 431 380 390 9 650 638 658 642 10 433 429 445 432 11 417 420 432 420 12 656 633 626 605 13 267 275 260 227 14 478 492 477 467 15 178 165 259 268 16 423 372 350 370 17 427 421 451 443 END DATA. SAVE OUTFILE='C:\Documents and Settings\peterw\Desktop\My Documents\My Documents2\BLAND AND ALTMAN PLOT\dat.sav' /KEEP=WRIGHT1 MINI1.
We now run some syntax that will produce a difference by average scatterplot plus the mean and its 95% confidence limits ready for su to add on shortly. Remember you will need to change the file names, variable names and the graph labels.
DATA LIST LIST/ subject wright1 wright2 mini1 mini2 (5 F8.0). BEGIN DATA 1 494 490 512 525 2 395 397 430 415 3 516 512 520 508 4 434 401 428 444 5 476 470 500 500 6 557 611 600 625 7 413 415 364 460 8 442 431 380 390 9 650 638 658 642 10 433 429 445 432 11 417 420 432 420 12 656 633 626 605 13 267 275 260 227 14 478 492 477 467 15 178 165 259 268 16 423 372 350 370 17 427 421 451 443 END DATA. SAVE OUTFILE='C:\Documents and Settings\peterw\Desktop\My Documents\My Documents2\BLAND AND ALTMAN PLOT\dat.sav' /KEEP=WRIGHT1 MINI1. GET FILE='C:\Documents and Settings\peterw\Desktop\My Documents\My Documents2\BLAND AND ALTMAN PLOT\dat.sav'. COMPUTE difference = WRIGHT1 - MINI1 . compute average = (WRIGHT1 + MINI1)/2. EXECUTE . OMS /SELECT TABLES /IF SUBTYPES=['Descriptive Statistics'] /DESTINATION FORMAT=SAV OUTFILE='C:\Documents and Settings\peterw\Desktop\My Documents\My Documents2\BLAND AND ALTMAN PLOT\des.sav'. DESCRIPTIVES VARIABLES= DIFFERENCE /STATISTICS=MEAN STDDEV. OMSEND. SAVE OUTFILE='C:\Documents and Settings\peterw\Desktop\My Documents\My Documents2\BLAND AND ALTMAN PLOT\dat.sav' /KEEP=WRIGHT1 MINI1 DIFFERENCE AVERAGE. GET FILE='C:\Documents and Settings\peterw\Desktop\My Documents\My Documents2\BLAND AND ALTMAN PLOT\des.sav'. RENAME VARIABLES (MEAN STD.DEVIATION=M SD). COMPUTE L95=M-1.96*SD. COMPUTE U95=M+1.96*SD. IF ($CASENUM LE 1) ID=$CASENUM. SORT CASES BY ID. SAVE OUTFILE='C:\Documents and Settings\peterw\Desktop\My Documents\My Documents2\BLAND AND ALTMAN PLOT\des.sav' /KEEP=ID M N L95 U95. EXE. GET FILE='C:\Documents and Settings\peterw\Desktop\My Documents\My Documents2\BLAND AND ALTMAN PLOT\dat.sav'. COMPUTE ID=1. EXE. SORT CASES BY ID. EXE. MATCH FILES FILE=* /TABLE='C:\Documents and Settings\peterw\Desktop\My Documents\My Documents2\BLAND AND ALTMAN PLOT\des.sav' /BY ID. EXECUTE. VAR LABEL difference 'Wright peak flow meter #1 - Mini Wright peak flow meter #1' /average '(Wright peak flow meter #1 + Mini Wright peak flow meter #1)/2'. GRAPH /SCATTERPLOT=average WITH difference /MISSING=LISTWISE .
Running the above in a syntax window should give you a data file containing the original data plus three columns called M (the mean difference), L95 and U95 representing the lower and upper endpoints of tis 95% confidence interval together with a scatterplot of the difference by average in the output file. To finish things off we need to edit the graph manually. This may be done as follows:
Firstly make sure that the range of the Y-axis contains the lower (L95) and upper (U95) confidence limits for the average inter-test difference (M). If the Y axis needs adjusting click on the graph and go to Edit>Select Y axis and then enter the range of the y-axis (and any increments) required.
We are now ready to add the three lines to the plot for the mean difference(M in the outputted data file) and the limits of its 95% confidence interval (called L95 and U95 in the data file). Go to options>Add a reference line to Y- axis. Put Y equal to M by typing Y=-2.11 in the box. Click Apply and Close. A line at Y=-2.11 should appear on the plot.
Repeat the previous step twice more for the limits of the confidence interval putting Y equal to the values -78.10 and, then, 73.86 representing L95 and U95 respectively clicking Apply and Close each time.
You should now be finished and have a scatterplot of the inter-test difference (y-axis) by the average test score (x-axis) with three lines for the mean and its 95% confidence limits looking something like this.