Sample size importantly limits the usefulness of instrumental variable methods, depending on instrument strength and level of confounding

IV methods are of most value in large studies if considerable unmeasured confounding is likely and a strong and plausible instrument is available.