From what I've been taught by several people throughout med school and residency, it depends on the indication. As was said above, warfarin makes a person transiently hypercoagulable by selectively depleting proteins C and S first, which are actually anticoagulant proteins that act on factor V. They have relatively short half-lives, thus when you give warfarin and inhibit vitamin K dependent clotting factor synthesis, the proteins with the shortest half-lives are the ones that run out first.
If you're just anticoagulating for prophylactic purposes, such as with atrial fibrillation, it's generally thought to be acceptable to not bridge with heparin or a LMWH to a therapeutic INR. The reason for this is that the theoretical clot risk in afib is exceedingly low -- only a few percent per year -- so the chances of you provoking a clot in the 3-5 days that someone is on their way to becoming therapeutic on warfarin is negligible. As such, it is felt that this small risk does not justify the cost, time, and inconvenience of using a bridge, and warfarin is just started as an outpatient.
On the other hand, if you're taking care of a patient who has a DVT or PE, the last thing you want to do is make them further hypercoagulable, even if just for a day or two. You want them therapeutically anticoagulated immediately, and want them to remain so until they no longer require treatment. This is why the bridge is thought to be important. Whether or not it actually pans out in terms of data though...that I'm not sure. I'd be curious to see what you come across in your PubMed searches. Last I checked, there were even some theoretical concerns that prompted hematologically oriented folks to recommend continuing the bridge for 48hrs after the INR is therapeutic even (which I rarely see done on the wards).