ABSTRACT

To measure and evaluate criterion validity for norm-referenced measurements, one may employ several different methods: comparison of group means, correlations, and error analysis. Before performing a Pearson Product-Moment Correlation, the first step in evaluating the validity of measurements through correlation is to examine the nature of the data. Whenever one prepares to find a correlation, it is important to look at a visual representation of the data first, by constructing a scatter plot. This chart can be created within the spreadsheet. This chapter looks at the overall shape of the chart. The closer the data are to a thin, straight line, the stronger the correlation. If the data were round like a ball, then the correlation would be very weak. If the data were U-shaped (right-side-up or inverted), or if they made a nice curve, then we may have a strong relationship, just not a linear relationship.