Stochastically forced discrete dynamical systems are considered. Using first approximation systems, we study dynamics of deviations of stochastic solutions from deterministic equilibria. Necessary and sufficient conditions of the existence of stable stationary solutions of equations for mean-square deviations are derived. Stationary values of these mean-square deviations are used for the estimations of the dispersion of random states nearby stable equilibria and analysis of noise-induced transitions. Constructive application of the suggested technique to the analysis of various stochastic regimes in Ricker population model with Allee effect is demonstrated.