Bootstrapping implicitly assumes that the data are independent of one another. Therefore, if there are correlations in the data (non-independent data), bootstrapping would be ineffective. Bootstrapping involves sampling with replacement from the original dataset, which is actually a fundamental aspect of the method, not a situation where it would be ineffective. The presence or absence of outliers doesn't directly affect bootstrapping's effectiveness. Re-sampling from regression residuals is a valid application of bootstrapping in regression analysis.