Check the logical reads with profiler, it's a good chance that it is the same for all three examples.

Check the execution plan and see the differences (if any).

The methods you use are all alike, there is almost no difference in performance.

You should not measure just one execution, but at least 3 executions per each method and take the average elapsed time and logical reads.

Logical reads is more reliable parameter (than elapsed time) you should look at to see if one method is better than other.

Other methods of optimization will probably give you a better gain in performance. For example, a covering index.