Statistics Glossary → Johnson-Neyman technique (separate slopes)
Johnson-Neyman technique (separate slopes)concept
A method that finds the range of values of a continuous moderator (covariate) over which an effect or group difference is statistically significant.
When group interacts with a continuous variable (the slopes differ between groups), the group difference is not a single number; it differs at every value of the covariate. The 'pick-a-point' approach tests the difference at only a few values (e.g. mean ± 1 SD). The Johnson-Neyman technique instead computes the boundary values at which the confidence interval for the difference just excludes zero, defining a 'region of significance' along the covariate axis.
The method was developed by Johnson and Neyman for ANCOVA when slopes are not homogeneous, and today it is also widely used in moderation analysis. The region of significance can extend beyond the observed range of the data; that part should not be interpreted. Because the regions rest on testing many points at once, the boundaries are estimates rather than sharp thresholds.
An everyday example
Two teaching methods work differently depending on students' prior-knowledge score: for students with little prior knowledge the new method is far better, while for those with a lot the two methods are nearly equal. The Johnson-Neyman technique might show, for example, that the advantage of the new method is significant below a prior-knowledge score of 62 and not above it.
In MerQur
The 'Extra outputs: Separate slopes (J-N)' checkbox of ANCOVA: it gives the slope in each group and, with two groups, computes the Johnson-Neyman region of significance.