Linear weights.
Think of offense as an equation.
1B + 2B + 3B + HR + BB + IBB + HBP + SB + Sac + SacFly + OUTS + CS + HIDP = Runs
That's a 13 variable equation, and thirteen equations would give you enough data to solve a system of equations. There are 30 equations every year, and you can take the averages over the years, which means they've kind of got a good idea of the run value of each event. In fact, I've seen some forms where they break down the outs into K, FB, GB, etc (the number of total equations per year, 30, one for each team, allows this.....ie, they can go up to 30 variables).
BBRef keeps linear weights performance under the "Advanced Batting" table, found by clicking the "Batting" tab on the player page. The number they list is called BtRuns, and it is relative to league average (for all players, not position depedent).
[table]Player|2008|2009|2010
Crawford|-9.0|13.7|26.4
Werth|12.5|25.4|38.9[/table]
The edge clearly goes to Werth, however, Crawford is pretty consistenly good for about a +15 in UZR, and Werth, like many average fielders, tends to go through some ups and downs.
Overall, it's probably a wash. I'd give the edge to Werth on two things, handedness in his new home park, and the fact that he can play a more premium defensive position (RF vs. LF)