SCAN, AN EXPLORATORY PROGRAM FOR PRELIMINARY-ANALYSIS OF BIOASSAY AND IMMUNOASSAY DATA

Bioassay and immunoassay response data conform in theory to a specified model. At its simplest, it is expected that the measured 'response', e. g. the inactivation of bacteria for a bioassay or radioactive count emission for a radioimmunoassay, is 'caused' by the administered treatment, e. g. antibiotic injected into bacterial culture or the amount of antigen added to a test tube. We describe the program SCAN which has been designed for use in conjunction with analysis of variance of immunoassay or bioassay data and which may also be used with more general one-way analysis of variance. Data within treatment groups are examined for outliers and normality of distribution. Overall statistics for normality and homogeneity of variance are also computed. Contributions of individual groups to these statistics are tabulated to permit the detection of aberrant groups. Means and variances are available in a form which facilitates detection of any response error relationship.