Testing our hypotheses on large datasets