Burns and Spangler (2000) report a study of the effect of homework compliance on the outcome of CT for depression in which they used structural equation modeling techniques to determine whether completion of homework produced improvements in depression or if it was that improvement in depression resulted in increased homework compliance. They found that individuals who did more homework improved much more than individuals who did little homework and that depression severity did not seem to influence homework compliance.

Schmidt and Woolaway-Bickel (2000) argue that we should look at the quality of homework completion, not just at the quantity of homework completed. Their study of the relationship between homework compliance and outcome of CBT for panic disorder produced partial support for this idea.

Finally, and importantly, Kazantzis (2000) examined the statistical power of studies which examined the effects of homework on treatment outcome. He found that many studies have relatively low power levels and may fail to detect the effects of homework on outcome due to low statistical power even when homework has a real effect. In the future it may be important for researchers to increase the power of their studies by increasing sample sizes, improving measures, or using more sensitive designs.