input field 58-70;

            The data for all the DATAS 2000 for SAS programs is read in using the same Data One format.  Do not delete negative numbers or zeros before reading in the data, the programs will automatically do this for you (when needed).

Output Tables:

Digitnum.sas produces all the output shown in the Summary of the Digital Analysis Tests for Data Profile, Basic Digits Tests, & Number Duplication.  Each table in the output is appropriately headed.

            Digitnum.sas also prints the Mean Absolute Deviation for the first digits, second digits, and first-two digits.  Chapter 8 of Digital Analysis Using Benford’s Law: Tests and Statistics for Auditors (www.acl.com) discusses the Mean Absolute Deviation and Chapter 10 of the book provides guidelines as to an acceptable MAD for each of the main digital tests in disbursement audits.

            All the DATAS 2000 for SAS programs show as one of the first output tables the first 25 observations in Data One.  This is a check for the user to check that the field (or fields) to be analyzed are the fields that were intended to be analyzed and that no error has been made in the formatting or pointing steps in Data One.

            In addition to the output described in the Summary, Digitnum.sas also produces a table of the “10 and over” numbers by descending total value.  In September, 1996, at a retailing client, the author spotted a duplication of the number $2 million on page 72 of the number frequency table.  When multiplied by the number of occurrences, this turned into a significant dollar amount.  To avoid big dollar amounts from being buried in the printout, the frequencies are sorted by total value.  For example, if $2 million appears 32 times then the total value is $64 million.  An analysis done in September, 1997 had high duplications of odd high dollar amounts (e.g., 342 times $251,345.66) and this led to the early conclusion that the data set contained errors.

Basic Digit Graphs:

            Digitnum.sas produces first digit, second digit, and first-two digit graphs.  The graphs produced by SAS/Base are not as neat as graphs produced by (say) Excel.  SAS prints either an “a” or an “e” in the plot area for actual and expected proportions.  If these two proportions are close then only an “a” or and “e” is printed and one of the symbols is hidden.  It is recommended that users use the Digitnum.xls programs to produce neat output for management reports.

            To produce neat graphs in Excel after executing Digitnum.sas on the PC the user needs to run the Digitnum.xls program in Excel 97 or Excel 2000.

1.      Load Excel 97 or Excel 2000.

2.      Load the Digitnum.xls program.

3.      Open the program as Read Only and Enable the macros if prompted for this.

4.      Run the macro Digital_Tables_Graphs_Create using Tools/Macro/Macros/Run.  This macro will produce neat graphs in the same format as is shown in the Summary.

If the DATAS2000_SAS/Output directory contains a digitout.csv file from a prior run of Digitnum.sas then SAS appends the current output to the existing digitout.csv file.  This creates a problem because the Excel macro reads the counts at the start of the file.

Users need to delete any digitout.csv file from the Output directory before any run of Digitnum.sas to ensure that digitout.csv only contains the results from the most recent execution of Digitnum.sas.  If Digitnum.xls executes and it “senses” that the digitout.csv file contains the results of more than one execution of Digitnum.sas a warning message is displayed on the main sheet in cells B29 and B30.  The warning is shown in bold print.