49.0 FREQ Command The FREQ command produces series frequency tables and simple cross tabulation tables. For complex cross tabulations it is suggested that a branch be made to SAS(r). The FREQ command handles missing data but does not handle character data at this time. The main objective of the cross tabulation option of the FREQ command (called by the TABLE sentence) is to allow analysis of residual outliers. Cross tabulation of the residual or the absolute value of the residual against right hand side or potential right hand side variables can be used in the identification stage of equation estimation. Form of FREQ command. B34SEXEC FREQ options parameters$ VAR $ TABLE options parameters$ B34SEEND$ Options of FREQ sentence. VERTHIST - Graphically display results in a vertical histogram. HORIZHIST - Horizontally display data in a horizontal histogram. NOFREQ - Suppress frequency analysis. Used if TABLE is present and only cross tabulations are desired. Pamameters on FREQ sentence. IBEGIN=n1 Sets first observation to use. Default = 1. IEND=n2 Sets last observation to use. Default = last observation in dataset. TYPE = key Sets type of frequency analysis done. Options are: EQUIDISTANT - Sets equidistant ranges. USERGROUP - User sets groups from RANGE sentence. EVALUES - Uses exact values. For n observations this option could product a max of n frequencies. This is default. NRANGE = Sets number of ranges for TYPE=EQUIDISTANT. Default = 10. RANGE(R1, R2, R3) User sets ranges. These must be increasing. If this option is used, then these ranges are used for all series. If this is not desired then, use the VAR sentence and make multiple calls to the FREQ command. A max of 200 values can be supplied. Notes: Unless the default TYPE=EVALUES is in effect, ranges are set for variables. If this is not desired, then it is recommended that the FREQ command be called for each variable or the TYPE=EQUIDISTANT option. VAR sentence. The var sentence is used to specify the variables to analyse. If this sentence is not present, all non character variables will be analysed. TABLE sentence. The TABLE sentence sets up variables to analyse using cross tabulations. TABLE sentence options EXACT - Calculate Exact probabilities for Contingency table. This calculation can take a great deal of time. For references 1. Cantor, Alan "A computer algorithm for testing significance in M by K contingency tables," Psycometrika 44(1) 1979 75-83 2. Freeman, G. H. and J. H. Halton "Note on an exact treatment of contingency, goodness of fit and other problems of significance", Biometrika 38, 1951, 141-149. 3. Hancock, T. W. "Algorithm 434," Comm. ACM 18(2) 1975, 117-119. 4. March, D. L. "Algorithm 434: Exact probabilities for R x C contigency tables," omm. ACM. 15(11) 1972, 991-992. TABLE sentence parameters. TYPEX = key Sets type of cross tabs done for x. Options are: EQUIDISTANT - Sets equidistant ranges. USERGROUP - User sets groups from RANGE sentence. EVALUES - Uses exact values. For n observations this option could product a max of n frequencies. This is default. NRANGEX = Sets number of ranges for TYPEX=EQUIDISTANT. Default = 10. RANGEX(R1, R2, R3) User sets ranges. These must be increasing. A max of 200 values can be supplied. TYPEY = key Sets type of cross tabs done for Y. Options are: EQUIDISTANT - Sets equidistant ranges. USERGROUP - User sets groups from RANGE sentence. EVALUES - Uses exact values. For n observations this option could product a max of n frequencies. This is default. NRANGEY = Sets number of ranges for TYPEY=EQUIDISTANT. Default = 10. RANGEY(R1, R2, R3) User sets ranges. These must be increasing. A max of 200 values can be supplied. Example # 1. User wants to obtain frequency distrubutions for all non character variables. This option may produce a great deal of output since there are potentially N catagories. b34sexec freq$ b34srun$ Example # 2. Same as example # 1 except want to look at x and y only. b34sexec freq$ var x y$ b34seend$ Example # 3. Same as example # 1 except want to look at x and y only. Data will be analysed into 5 groups. b34sexec freq type=usergroup range(.1 .2 .8 .9 .10)$ var x y$ b34seend$ Example # 4. Same as example # 1 except want to look at x and y only. Data will be analysed 10 equidistant groups. b34sexec freq type=equidistant nrange=10$ var x y$ b34seend$ Example # 5. User wants a cross tabulation between x and y. No frequency plots are given b34sexec freq nofreq$ table(x,y)$ b34seend$