我的猜测也是%if 之后默认加了%eval()运算，而不是%sysevalf()运算，因为那个比较结果和%eval()的方式一致
但是有没有官方的对于
“%if语句会自动将%eval函数用在%if后的条件上”
粗粗查了下，没有找到

帮助里面我也没找到，我是从《SAS macro programming made easy》的165页看到的：
The macro functions that automatically invoke %EVAL around the expressions supplied to them are
      => %SUBSTR and %QSUBSTR
      => %SCAN and %QSCAN
The macro language statements that automatically invoke %EVAL around the expressions supplied to them are
      => %DO
      => %DO %UNTIL
      => %DO %WHILE
      => %IF

经hopewell大大的指点
我在SAS HELP中找到了星星点点的原文

Macro Expressions

How the Macro Processor Evaluates Logical Expressions

......
A special case of a character operand is an operand that looks numeric but contains a period character. If you use an operand with a period character in an expression, both operands are compared as character values. This can lead to unexpected results.
......
Because the %IF-THEN statement in the COMPNUM macro uses integer evaluation, it does not convert the operands with decimal points to numeric values. The operands are compared as character strings using the host sort sequence, which is the comparison of characters with smallest-to-largest values.


原文以
%macro compnum(first,second);   
%if &first>&second %then %put &first is greater than &second;   
%else %if &first=&second %then %put &first equals &second;   
%else %put &first is less than &second;
%mend compnum;
调用
%compnum(10,2.0)
出结果
10 is less than 2.0
解释这个现象，应该和楼主的意思是一致的
附上hopewell大大的指点，sas support上的：
All parts of the macro language that evaluate expressions (for example, %IF and %DO statements) call %EVAL to evaluate the condition. For a complete discussion of how macro expressions are evaluated, see Macro Expressions.

http://support.sas.com/documentation/cdl/en/mcrolref/61885/HTML/default/viewer.htm#a000208971.htm

谢谢各位的帮助！我明白他的机制了！