ANALYZING EXISTING STATISTICS
    An Illustration of the Analysis of Existing Statistics: Durkheimís Study of Suicide
    Units of Analysis
    Problems of Validity
    Problems of Reliability
    Sources of Existing Statistics

    In this section, you will come to recognize the enormous mass of data already complied and published in the form of statistical reports, all of which can be used for further analysis.  Much of the data we'll consider are from various government agencies, but you'll also see that many nonprofit organizations and businesses also provide data for analysis.  In addition to the libraries-full of printed reports, the web has now vastly expanded this resource.

    The use of existing statistics requires a revisit from our old friend, the unit of analysis.  It's important, you'll see, to determine what the unit of analysis is in a statistical table.  Otherwise, you will have difficulty using those data in your own analyses.  And sometimes, the units of analysis in the available data are not those you would prefer.  This was the case in Emile Durkheim's classic study of Suicide.  Unable to study the characteristics of individual people who committed suicide, Durkheim was forced to drawn inferences from the suicide rates and other characteristics of nations and regions of nations.

    Existing statistics also force us to re-examine the measurement-quality criteria of validity and reliability.  In terms of validty, you may want to study the extent of marijuana use in a society, but the only data available to you are the numbers arrested for marijuana use: presenting a validity problem.  And we'll also see that officially collected and reported data are not always trustworthy: a reliability problem.