To investigate whether the multiplication of frequency score and
severity score added more details of impact than using only
frequency or severity, three multiple regression models were
applied in prediction of two major variables; DMFT and number of
sextants with deep periodontal pockets. The three approaches
were, all subsets regression, the stepwise method to select the best
predictor among the three kinds of scores, and the hierarchical
regression. For all three approaches, the OIDP score was a better
predictor than either frequency or severity score separately. All
subset regression model with OIDP showed the lowest mean
square residuals. In the stepwise method, only OIDP was included
in the equation to predict both dependent variables. However, the
improvement, by multiplying frequency and severity score, was
not statistically significant (P>0.01)