I think that if I were in Nazi Germany, it would be acceptable to forcibly show people videos of concentration camps to save Jews. However, if I were in Nazi Germany, it would also be acceptable to cheat, to destroy property, or even to shoot people in order to save Jews.
To a rational vegetarian who thinks that eating animals is as bad as killing Jews, it would be moral to do anything in order to prevent the eating of animals. However, such a variety of vegetarian is a menace to society from the point of view of other people who don't share in his ideology, and would (according to those other people) need to be stopped--and rationality doesn't really matter at this point; whether he is stopped would only be a question of who is more powerful.
Now, suppose that a vegetarian thinks that eating animals is only 1/1000 as bad as killing Jews. If it's okay to do otherwise evil things to save a Jew in Nazi Germany, then this vegetarian ought to be willing to do them to create 1001 vegetarians. So vegetarians should be shooting people right and left (or doing other inherently immoral things) if they think it advances vegetarianism enough.
Of course, it's not possible for a vegetarian to be absolutely certain about vegetarianism, and once he accounts for his own uncertainty about vegetarianism in the calculation, the possibility that he is wrong and will have committed murder makes it no longer palatable to kill people to create vegetarians even if you're creating lots and lots of vegetarians.
(And if he then reasons "okay, I shouldn't kill to create 1001 vegetarians, but my uncertainty about vegetarianism is so low that it's okay to kill to create a million vegetarians, that still doesn't work because he is also uncertain about what his uncertainty is.)
Doing an (otherwise) mildly immoral thing to create many vegetarians is similar, except that of course the acceptable bound on his uncertainty about vegetarianism is higher--maybe high enough that he actually could justify it. But then this brings in other considerations--if he can mildly hurt other people for vegetarianism, then others will also be able to mildly hurt him for their own pet causes, for instance, and things may be overall better off if nobody is allowed to hurt others for their causes.
Being uncertain about one's uncertainty doesn't mean you shouldn't act according to your moral beliefs, the uncertainty can go both ways. You still have to assign probabilities in the end.
Assuming you attribute the same importance to reducing suffering, no matter in what type of being it occurs, it still wouldn't make sense to "shoot people left and right". People who care a lot about animals are already being branded as "extremists", you'd have more impact by going about it with a more thought-out strategy like e.g. movement-building ...
I notice that a large fraction of Effective Altruism people are vegetarian. This makes sense: in general Effective Altruists take moral issues seriously, even when that means changing your lifestyle. I'm not sure it's a good balance, though.
One way to think about this is to convert it into money. How much would I need to be paid to give up eating meat? All animal products? How much money would I need to spend on myself to be about as happy as I would be with less money but continuing to eat animals? I'd probably be willing to go vegetarian for about $500/year, vegan for maybe $2000/year.
It turns out you can probably pay to convince other people to be vegetarian much more cheaply than that. I estimate the cost of a vegetarian-year at $4.29 to $536 while Brian Tomasik estimates $11 with better methodology (which I looked at). This is by placing ads on facebook for a site where people can watch an animial cruelty video and ideally become vegetarian or vegan.
If you would get more than $11/year worth of enjoyment out of continuing to eat meat, why not give $11/year to convince someone else to not eat meat for you? Or give $50/year and be on the safe side?
(While you're giving money, you should probably give it to the organization that you think will do the most good with it, which I think is probably one of GiveWell's top charities. The nice thing about money as opposed to actions is that it's easy to redirect.)
I also posted this on my blog.