Thousands of studies of SARS-CoV-2 seroprevalence have been published since the beginning of the pandemic. Researchers have reported a range of methods used to estimate seroprevalence from study data. Because diagnostic tests are imperfect, false negatives and false positives can be expected, as typically described by a test's sensitivity and specificity. A number of methods exist in the statistical literature to correctly estimate disease prevalence in the presence of test misclassification, but these methods seem to be less known and not routinely used in the epidemiology literature. Using straightforward calculations, we estimated the amount of bias introduced when reporting the proportion of positive test results instead of using sensitivity and specificity to estimate disease prevalence. We also derive a range of disease prevalence, given sensitivity and specificity, where this bias is within desired tolerances. Expected bias is often more than is desired in practice. We apply these results a recent school-based cohort study of COVID-19 in children in Switzerland. We conclude with some tips on software packages which may be used to compute less biased seroprevalence estimates.