Often, before or after collating possible variables from the literature, you may notice a gap or think of some obvious variable that no one has used, but which you feel may have an effect.
Dr. Ian Kennedy, thanks for the brainstorming tips. I think such self-brainstorming session would not only help to avoid heavy influence of literature, but also would guide while looking for relevant literature regarding new variables as well. Sometimes, literature could be scant on real gap(s) where intuition and brainstorming could greatly help. At the end of the day, literature support is necessary, but only with good judgment and caution.