Yt = c + φ Yt -1 + εt , where εt is white noise, (2)
with parameter φ=.6, then stickiness = .6. This means that 60% of any shock in Yt is carried over to the next period. Similarly, if the univariate model is AR(2) with parameters φ1 = 0.6 and φ2 = 0.15, then stickiness = .75. A priori, we expect consumer attitudinal metrics to be stationary, i.e. the sum of the AR parameters is less than 1 because of memory decay effects that are well-documented in psychology (Baddeley, Eysenck and Anderson 2009). Stickiness also relates to the well-known measure of half-life of a marketing impact.