On the Mises Blog, D. W. McKenzie discusses whether or not personal crime prevention should be considered "socially wasteful".
Unsurprisingly, he finds that it is not.
One thing not counted for in the argument that protecting your yard in a visible way just makes other people more likely to be attacked is that making more cars or more houses harder to break into makes it somewhat costlier for a thief, who must spend more time determining which cars or houses he has the best chance of getting into.
This might seem a small price to pay for the criminal; if a criminal can't get into one car, there is usually another near by. However, this assumes that all cars would be equally attractive targets if not for, e.g., the club. Assuming that crime is not only directed at easy targets but at the most lucrative ones, then it is likely that car thieves or home robbers already spend some time checking out their targets; this means that being forced to forego robbing a particular car (home) may be somewhat costly in terms of a criminal's time and thus increase the expected opportunity costs from robbery (e.g., more time and energy spent scoping out cars/home, more time and energy devoted to determininng how to spot an easy car/home), decreasing crime in general. This is particularly true if the most valuable cars (homes) are the ones most likely to have security devices; this means that thieves might have to settle for much lower payoffs from stealing, and would thus make stealing a less attractive option.
That is all.