Technical note: the initial stages of statistical data analysis.

Abstract

OBJECTIVE To provide an overview of several important data-related considerations in the design stage of a research project and to review the levels of measurement and their relationship to the statistical technique chosen for the data analysis. BACKGROUND When planning a study, the researcher must clearly define the research problem and narrow it down to specific, testable questions. The next steps are to identify the variables in the study, decide how to group and treat subjects, and determine how to measure, and the underlying level of measurement of, the dependent variables. Then the appropriate statistical technique can be selected for data analysis. DESCRIPTION The four levels of measurement in increasing complexity are nominal, ordinal, interval, and ratio. Nominal data are categorical or "count" data, and the numbers are treated as labels. Ordinal data can be ranked in a meaningful order by magnitude. Interval data possess the characteristics of ordinal data and also have equal distances between levels. Ratio data have a natural zero point. Nominal and ordinal data are analyzed with nonparametric statistical techniques and interval and ratio data with parametric statistical techniques. ADVANTAGES Understanding the four levels of measurement and when it is appropriate to use each is important in determining which statistical technique to use when analyzing data.