Small Script Showing Effect of Merge and Match-Merge
====================================================

data Fil1 ;
   input X Y A $;
     datalines;
        1  3  ab
        2  2  ab
        1  4  bc
        2  7  ac
        2  8  dd
        3  5  af
        4  6  fw
        7  7  hj
           ;  
       run;

data Fil2 ;
   input V W A $;
     datalines;
        1  3  ab
        1  4  bb
        2  7  ac
        2  8  dg
        3  5  ab
        4  6  fw
           ;  
       run;

proc sort data=Fil1 out=Fil1B;
    by A;
proc sort data=Fil2 out=Fil2B;
    by A;

options nocenter;
data mrg1 ;
   merge Fil1 Fil2;
proc print data=mrg1; run; 

Obs    X    Y    A     V    W
 1     1    3    ab    1    3
 2     2    2    bb    1    4
 3     1    4    ac    2    7
 4     2    7    dg    2    8
 5     2    8    ab    3    5
 6     3    5    fw    4    6
 7     4    6    fw    .    .
 8     7    7    hj    .    .


SO WE CAN SEE THAT ORDINARY MERGING (without `MATCH'ing)
just brings together the two datasets with the same 
indexing for rows and adopts the LAST given elements
for any doubly-defined columns.

data mrg2 ;
   merge Fil1B Fil2B;
   by A ;
proc print data=mrg2; run;

Obs    X    Y    A     V    W
  1    1    3    ab    1    3
  2    2    2    ab    3    5
  3    2    7    ac    2    7
  4    3    5    af    .    .
  5    .    .    bb    1    4
  6    1    4    bc    .    .
  7    2    8    dd    .    .
  8    .    .    dg    2    8
  9    4    6    fw    4    6
 10    7    7    hj    .    .


BUT MATCH-MERGE BRINGS TOGETHER ALL OBSERVATIONS 
FROM BOTH SORTED DATASETS (in all possible ways, 
when the same observations occur more than once 
in a single dataset) using as many of the total 
set of columns as are properly defined for each
occurring value of the BY-variable(s), and 
filling out the undefined values as "missing".