Some curious information on this thread. Strangely, PPC-Coach's statement is more in line with reflecting contempt for the system than conveying advice, unless he simply means that broad matches have a wider ad triggering potential. I don't understand the point of separating match types, and the statement above is incorrect. Quality score is calculated for each keyword before every impression. The score is based on many factors. If the campaign has been running for a while, the most important factors to be considered are the keyword's maximum CPC, its relevance to the landing page content, its relevance to the ad being displayed, the ad and keyword's CTR history, the ad and keyword's relevance to the search query, and the keyword's overall CTR history on the Adwords system. If the broad match version of a particular keyword performs poorly with a certain ad, it doesn't mean that the same ad will rank low when triggered by the phrase or exact match versions of the same keyword or any other keyword in the ad group. This is simply because ads are not assigned a score, only keywords are. And since every keyword is assigned an individual score, the poor performance of one cannot impact the score of another. Anyway . . . That's how I understand it. If this is incorrect, please clarify.
All I can say is, since I've taken the Broad & Phrase matched keywords out of my ad groups and put them into their own .. my min click has gone down and my Ad Position seems to have improved slightly. Also - it's making life much more easier when doing split testing, as an ad that works for [widgets] might not work as well for "widgets"/widgets.
See... for all those who didn't believe in separating match types My point has just been proven by a third party person =)