An election experiment

In California, certain laws go directly to the voters for scrutiny. This year we will be deciding propositions 1 through 12. To help inform voters, the Secretary of State has distributed a handy 143-page packet, printed on newsprint, one per household. Each ballot measure includes a summary and description of the law written with some attempt at neutrality, followed by arguments in favor and against the proposition, with shorter rebuttals to those arguments. Depending on the issue, the arguments tend to use a lot of all capital letters. I find this practice annoying and difficult to read, the print equivalent of screaming.

So here is the experiment I would like to see. Gather some suitably random sample of a voting population. Present them with some booklets with some ballot measures, perhaps taking some care to include a variety of issues. Distribute randomly three different versions of the booklet, noting who gets what. One version has arguments and rebuttals in ordinary prose. One version has arguments on one side of the issue written in ordinary prose, while the opponents scream, as in this excerpt from this year's Prop 4:

IF OUR DAUGHTERS COULDN'T COME TO US, for whatever reason, THE MOST IMPORTANT THING IS KEEPING THEM SAFE. New laws cannot force our teens to talk to us, but they may force them into back alleys...or worse.

The third set of booklets has the other side writing all the same claims in normal sentence case while the proponents holler:

Yet a young girl can get an abortion WITHOUT A FAMILY MEMBER BEING NOTIFIED—and this could endanger her safety, even her life.

The experiment should even switch back and forth between whether the gratuitous capital letters were pushing for the liberal or conservative viewpoint.

Then, see which version of the ballot pamphlet gets more votes one way or the other. Does all that extra emphasis really help to convey the urgency of the point, or does it merely suggest that the argument wasn't convincing enough and so had to be stated more forcefully?

Of course, this experiment may have little bearing on real election results. Even if both sides could be persuaded (or compelled) to mind their manners, a whole host of influence outside the booklet is also at work. Still, it would be instructive to see how one controllable factor influences the results.