When I ask the user a question with two options of which I want the user to choose a specific one, should I formulate the sentence so that the desired option it the latter or the other way around?
Are there researches about at which position the options is more likely to be chosen?
In psychology there is a lot of research into whether primacy (first presented) or recency (most recently presented) most affects the choices that people will make. To cut through a lot of theory, in most cases primacy dominates - especially when the choices are presented very close to each other in time.
So if you want someone to go for a particular option over another (all things being equal of course), it is best to present the option you want them to take first.
If it's critical, I would suggest doing some A/B testing yourself to see whether it holds for your particular case.
Most articles on this topic require access to overpriced academic journals, but here are two good papers that are freely available:
Generally speaking, human beings remember the beginning and endings of sequences and sentences, and muddle the points in the middle. Journalistic style writing uses this concept pretty regularly, usually sticking with brief sentences wherein facts are at the beginning and ending of sentences. For example: The fire, which lasted 12 hours, destroyed 17 buildings. However, if we wanted to emphasize how long the fire burned, we would write: The fire destroyed 17 buildings and burned for 12 hours.
Without the middle, we still get the pertinent information (what the author thinks is pertinent anyway) of both sentences.
Having said that, I have not heard of (nor read) any research or studies being performed on the word order of a sentence altering/inspiring a desired response. Phraseology and word choice maybe, but not order. Leading a user to a conclusion (Apple's "buy it" button on their product detail pages) by making certain elements in a sequence conspicuous is also relatively normal. One might argue that if you would prefer a user to pick one option over another to make that the only option.
Facebook, for example, is pretty much a signup form with very brief blurbs about what Facebook is (depending on which home page you get today). Twitter.com does the same thing. Google. And so on.
What you're looking to do is introduce response bias in your questioning. It'll get your results rejected from a scientific journal, and it's a very common tactic in partisan polling.
Now, there is a lot of research - specifically if you're looking to push a specific price point.
For pricing, there are two supported ideas.
The first is that potential purchasers with no information will have a higher average price, but a lower maximum price. By providing three options (instead of two), you're essentially pushing for the middle one with users that have no comparison (or with a unique product that has no direct alternative).
The second idea is that we will frame our decision based on the nearest alternative. Again with three options, we'll generally choose the "better" of the options if we can "chunk" them mentally into 2-similar options vs. 1 different option. In other words, if you're offering 3 options: Option 1: Cheese (Mozzarella) Option 2: Cheese (Cheddar) Option 3: Apples ... then users will select their preferred cheese (2 options, selecting the "better" one), no matter how they feel about fruit (unless they're anti-cheese).
Now, that won't help if you're set on two options - unless you're willing to play with the sentence leading up to the options (which is what you'd like to do).
For non-pricing examples, you'll present both options, but give additional depth to one of them. With the Cheese / Fruit example, you'd say something like:
"Please select from the following two options, Fruit or Cheese. Remember: Your choice is for generalities, not specifics. Fruit should be evaluated as a whole, and cheese should not be evaluated as if it was a competition between cheddar and mozzarella."
There's no obvious bias involved - you're explaining the positive option with fruit, and the negative option with cheese. Involving additional choices though, for one of the options (even in what appears to be a negative option), you're actually providing more cognitive wiggle room. You're increasing the cognitive space allocated to cheese, and improving that selection.
The easiest way to explain this to others is to say "Do you like options, or no?". Everyone likes options (just not too many - and that's a different discussion) - so increasing the options available, even when it's not a true option, should get your results the way you want them.
