The break-junction technique is widely used to measure electronic properties of nanoscale junctions including metal point-contacts and single-molecule junctions. In these measurements, conductance is measured as a function of electrode displacement yielding data that is analyzed by constructing conductance histograms to determine the most frequently observed conductance values in the nanoscale junctions. However much of the rich physics in these measurements is lost in this simple analysis technique. Conductance histograms cannot be used to study the statistical relation of distinct junction configurations, to distinguish structurally different configurations that have similar conductance values, or to obtain information on the relation between conductance and junction elongation. Here, we give a detailed introduction to a novel statistical analysis method based on the two-dimensional cross-correlation histogram (2DCH) analysis of conductance traces and show that this method provides new information about the relation of different junction configurations that occur during the formation and evolution of metal and single-molecule junctions. We first illustrate the different types of correlation effects by using simulated conductance traces. We then apply this analysis method to several different experimental examples. We show from break-junction measurements of different metal point-contacts that in aluminum, the first conductance histogram peak corresponds to two different junction structures. In tantalum, we identify the frequent absence of adhesive instability. We show that conductance plateaus shift in a correlated manner in iron and vanadium junctions. Finally, we highlight the applicability of the correlation analysis to single-molecule platinum-CO-platinum and gold-4,4'-bipyridine-gold junctions.