Indiscriminate (i.e., carless, random, insufficient effort) responses, commonly believed to weaken effect sizes and produce Type II errors, can inflate effect sizes and potentially produce Type I errors where a supposedly significant result is actually artifactual. We demonstrate how indiscriminate responses can produce spuriously high correlations in depression and hopelessness data in a nonclinical population (i.e., undergraduates), how this inflation occurs, where this misrepresentation is likely to happen, and how to guard against it. Although previous researchers have succeeded in showing this effect with samples of entirely simulated data, this study is the first to our knowledge to show that indiscriminate responding causes Type I errors in observed data.