What you want to (can :) measure - the Dependent Variable.
The dependent variable is a measured human behaviour. The independent variables in our case are the following:
We count the number of mistakes that the user makes while playing the song. This can be easily be found, by counting the number of crosses that are shown in the android app. The Android app continuously monitors the notes that the user is playing. The only wrong notes that the app doesn't catch is when the user plays an open note (a note played with no fingers on the fret)
Another variable that we can measure is the time that the user takes to finish a song. We understand that the user's previous experience with playing a guitar and familiarity with the song can play a role on this varible as well.
This is another variable similar to the number of mistakes. May be we can use accuracy to measure the correctness of the notes played by the user over all the set of songs he is asked to play.
We use the android application as a feed back to the user. We will be doing two sets of experiments. One with the user given the feed back. So that the user can get to know whether he played the right or not. Also, he can know what is the next note to be played.
Another set of experiments involve the user being asked to play songs that he is familiar/unfamiliar with. This can help us understand the influence of this independent variable
Familiarity with the song can play a role on the effect of this device. If the user is familiar with the song, then he can easily know whether the note he/she is going play is a higher/lower note. And accordingly he can prepare himself to play the note.
The lighting on the room is important in this experiment. A little dark room can help the user to identify the color of the LED but a minimum lighting is required for him to figure out the fret and string that he wants to play.