I think it is more: Complication allows the researchers' biases to slip in more easily, since among other things any sort of cross-check is nearly impossible, which leads to softer results, especially when being evaluated by someone with different biases.