Abstract
We show that an asymmetric Bernoulli random variable is symmetry resistant in the sense that any independent random variable, which when added to it produces a symmetric sum, must have a variance at least as much as itself. The main instrument is to use Skorokhod embedding to transfer the discrete problem to the realm of stochastic calculus.
Kagan, Abram; Mallows, Colin L.; Shepp, Larry A.; Vanderbei, Robert J.; Vardi, Yehuda. Symmetrization of binary random variables. Bernoulli 5 (1999), no. 6, 1013--1020. MR1735782 (2001c:60020)
Karatzas, Ioannis; Shreve, Steven E. Brownian motion and stochastic calculus.Second edition.Graduate Texts in Mathematics, 113. Springer-Verlag, New York, 1991. xxiv+470 pp. ISBN: 0-387-97655-8 MR1121940 (92h:60127)