We review the data analysis algorithm used for the analysis of LIGO S2 data in coincident with externally identified gamma-ray burst events, such as GRB030329. The method uses cross-correlation of two interferometer data streams albeit with several additional layers of pre- and post-processing. The pair of data streams is transformed into a three dimensional random field by scanning the relevant free parameters that define a cross-correlation. The random field is progressively collapsed to a single statistic by using criteria such as the clustering of high significance pixels. An extensive study of the performance of the algorithm in both stationary, gaussian as well as realistic models of non-stationary noise is presented. Analytical results related to various aspects of cross-correlation are compared with the numerical results.