Interpretation lagged probit model

[Mkkl]

Hi there!

I have a question concerning the interpretation of the results of my probit model. I have estimated a probit model with only a constant and 1 lagged regressor for mutiple lags of this regressor. So simply said Y = a + B*X[i-k] for several k,s and my dependent variable is binary. Now when I chose a preferred lag h based on a pseudo R-squared and I plot the binary variable together with NCDF(a + B*X[i-h]) how should I interpret the graph? NCDF is the standard normal cdf and the graph is plotted without any lag. Is it correct to interpret it as that NCDF(a + B*X[i-h]) is the probability that Y will occur in h periods?

Kind regards,

Michael