This generic science education study demonstrates the contrast of findings obtained through analyzing all the raw data as opposed to removing various combinations of identified potential outliers provided through the application of six diagnostic procedures. Outliers are defined as extreme data points with large residuals. It is argued that an outlier check should be performed on all quantitative data analyzed in science education research to "sound a warning" concerning the accuracy of the collected raw data before standard parametric procedures are applied. Five of the diagnostic procedures are regression diagnostics: the Leverage Point, the Studentized Residual, DFITS, Cooke's Distance, and DFBETA. These values are obtained through the use of Statistical Analysis Software (SAS) by including the influences option in the model step in Proc Reg. A hand calculation of critical values is then done to obtain cut-off values to identify potential outliers. The other diagnostic procedure demonstrated is a visual inspection of the scatter plot of the raw data to identify suspicious data points. Cooke's Distance, being the most conservative procedure, is recommended as the preferred diagnostic procedure to highlight suspicious points in science education research. The outlier diagnostic procedure, the SAS procedure to obtain calculated diagnostic values, and a calculation of diagnostic cut-off values are appended. (KR)