We analyze a random transport model of a scalar quantity on a discrete
space-time. By changing a parameter which is a portion of the quantit
y transported at a time, we observe a continuous change of steady-stat
e distribution of fluctuations from Gaussian to a power-law when the m
ean value of the scalar quantity is not zero. In the symmetric case wi
th zero mean, the steady-state converges either to a trivial no fluctu
ation state or to a Lorentzian fluctuation state with diverging varian
ce independent of the parameter. We discuss a possible origin of the i
ntermittent behaviors of fully-developed fluid turbulence as an applic
ation.