By default, the column width to represent the requested statistics is 8 characters. The adjacent columns are separated by one character. The number of decimals used for e.g. means depends on the maximum value of that variable. If sums are requested then the number of decimals used is determined by the product of the maximum value and the number of observations. For numerical grouping variables the same 8 char. column width is used. For string grouping variables used as row variables the length of the string variable (if greater than 8) defines the column width used. If the grouping structure of a string variable is searched by the program then the found actual maximum length is used. This default behaviour may be overridden by giving the printing formats of object variables and/or grouping variables, e.g. VARIABLES=Knowdge(######.#),Applicat(1234.12345) GROUPING=Sex(#####),Grade(12345) The number of decimals given in the format concerns means, sums, deviations, standard errors of means, minimum and maximum values. The column width defined by the format concerns frequencies and percentages too. Depending on the defined column width percentages are printed with 2, 1 or no decimals (if the column width is at least 6 then two decimals are given). This default behaviour of printing percentages may be altered by specifying the number of decimals used by PCTDEC. The given number (0-5) must agree with the column width in use. In general column width can vary from 4 to 32. Further information: M = More information on the MTAB operation