To understand the difference between results returned by the two different WHERE qualifiers, ask yourself what does the [subjurisdiction] field contain?
Using the where qualifier
means that only those records that contain the two characters M and B in sequence, and only those two characters, will be selected. Anything else, including white space characters of any kind (spaces, newlines, unicode characters that are invisible, etc), that precedes or follows the MB sequence will not be equal to 'MB'.
You don't say what your [subjurisdiction] field contains, but we can infer what it contains based on your description of the data type as character(25) and then consulting the PostgreSQL documentation, for example, at https://www.postgresql.org/docs/9.1/datatype-character.html
That tells us... "Values of type character are physically padded with spaces to the specified width n, and are stored and displayed that way."
So, if you are looking at a character(25) field that seems to contain the characters MB, it actually does not contain only 2 characters, it contains 25 characters, the characters M and B followed by 23 space characters.
If you want to match that, you have to write:
WHERE [subjurisdiction]='MB '
...where you've put 23 space characters between the B and the concluding single ' quote. Spaces are real characters, just as real as, say, an underscore _ character. You don't expect 'MB' to be equal to 'M_B' or 'MB_', so you shouldn't expect 'MB' to be equal to 'M B' or to 'MB ' (space character between the B and the ').
The solution is to take advantage of Manifold's handy and ever-popular string manipulation SQL functions, such as StringTrim:
WHERE StringTrim([subjurisdiction], ' ') ='MB'
We use the above to trim any space characters (the space between the ' ' in the function arguments) before making the comparison to the string literal 'MB'.
I've tested the first query in pgAdmin, QGIS, and Access
Meaningless, since Access doesn't add space characters to data to pad them, QGIS has no database (so you were really using some other DBMS), and the PostgreSQL documentation says it explicitly ignores space characters used for padding.
That latter bit, by the way, is an uncharacteristically sloppy approach to data for Postgres, since either a literal is what you mean it to be or it isn't, with sharp minds not buying any of this "oh sometimes I mean it when I use a space character and sometimes I don't".