Suppose we consider the first two observations based on a logistic regression model, where the first variable in observation 1 takes a value of x1=6 and observation 2 has x1-4. Suppose we realized we made an error for these two observations and the first observation was actually x1=7 (instead of 6) and the second observation actually had x1=5 (instead of 4). Then the predicted probability from the logistic regression model would increase the same amount for each observation after we correct these variables.