A modified CD*I model for the origin of structure in the universe, based on an inflation model with two iuteracting scalar fields, is analyzed to make predictions for the statistical properties of the density and velocity fields aud the microwave background anisotropy. The initial gauge-invariant potential (which is defined as (, = £p/(p + p) + 3_, where ? is the curvature perturbation amplitude and p is the pressure, is tl_e stt,n of a Gaussian feld oi and the square of a Gaussian field _2. We assume a Har,'ison-Zel'dovicla scale-invariant power spectrum for _bl, and a log-normal 'peak' power spectru,u for o,.. The location and the width of the peak are described by parameters/,'_ aud a. respectively. Comparing our theoretical calculations for two dimensional density angular correlation function to the results of the APM su,'vey, we set certain pa,-a,neter regime for k_, a and the amplitude ratio of Gaussian and non-Gaussian fields. The model is motivated to some extent by inflation models with two interacting scalar fields, but is mainly interesting as an example of a model whose statistical properties change with scale. On small scales, it is almost identical to a standard scale-invariant Gaussian CD,N{model. On scales near the location of the peak of the non-Gaussian field, the distributions have long tails in high positive values of the density and velocity fields. Thus, it is easier to get large-scale streaming velocities than the standard CDM model. \Ve calculate the quadrupole amplitude of fluctuations of the cosmic microwave background radiation and the ,'ms va,'iation of the temperature field smoothed with a 10°F\VHM Gaussian. and find a reasonable agreement with the new COBE results. Biasing should be important in our model. \Ve discuss one of the possible biasing schemes very b,'iefly, and further discussions on this will be in our future work. Subject. headin9._" large scale structure-statistics-cosmic background radiation