Hansen, Kaplan, Zamir and Zwick (STOC 2019) introduced a systematic way to use “bias” for predicting an assignment to a Boolean variable in the process of PPSZ and showed that their biased PPSZ algorithm achieves a relatively large success probability improvement of PPSZ for Unique 3SAT. We propose an additional way to use “bias” and show by numerical analysis that the improvement gets increased further.