In a psychological experiment, all participants were asked to rate the trustworthiness of a series of neutral facial expressions shown on computer (e.g., 1 for the lowest trustworthiness, 7 for the highest). There are 60 faces to be rated.
Under each expression, there’s a button. If the participant choose to click it, a priming word will pop out. For example, “Happy”, “Sad”, “Hate”, “Energetic”… Positive and negative priming words were randomly given.
The participant can ignore the button, and rate the trustworthiness based on his/her intuition. Thus, whether choose to click and unveil the priming word purely depends on the participant’s autonomy.
150 participants attended the experiment. Some of them chose to click and unveil some priming words (e.g., 10 out of 60 trials, or 20 out of 60 trials), however, some just ignored all the priming words (i.e., judge the trustworthiness purely based on own intuition).
Now, we want to explore whether unveiling positive and/or negative priming word would affect the perception of facial trustworthiness. The participants can be group with gender(M/F), or age group…
But my question is how to group the “behaviours” of the participants. As mentioned, some participants may only unveil 1-2 priming words, and some may even ignore all the priming words. While some may unveil nearly half of them. Last but not least, there’re positive and negative priming words. How do I choose variables? Should I hypothesize that “the more positive words unveil, the higher the trustworthiness will be perceived – and vice versa”?
How to formulate research question and How to choose the statisitical method ?