Putting blinds on a study makes its results subject to less bias and outside influence. The more blinds there are the more representative of normal the data will become, and thus the more believable. With each addition of a blind, less people become liabilities. In single blind, the researchers are liabilities only. In double blind, the statisticians/data analysts are the only liabilities. With each person you take off the liability list, it neutralizes the effect of their bias on the data. It also stops them from influencing which patient receives what treatment or giving preferential treatment to a patient. It even mitigates damage from accidental slip-ups, like telling a patient they are on the placebo. Studies in which the researchers have little emotional attachment to the subjects do not require multiple blinds as much because there is less motivation to tamper with the protocol. A general rule of thumb is that more blinds = less data fouling.