ARTICLE AD BOX
This is a project in PHP but i am happy with any suggestions, regardless of language. The coding is less the issue than a missing piece in logic i am struggling with.
Users answer a series of 7-point Likert scale questions (from strongly against to strongly support with center neutral).
Each question is assigned a specific point on a two-axis cartesian grid. The grid is 20x20 (10x10 in four quadrants around a central axis at (0,0)) but the size is unimportant.
The desired result is plotting a single point (x,y) of best fit, which should be a relatively easy concept. If the scale (s) was from 0-3, that would be relatively easy. Just a weighted average of all the answers where each point was treated as (sx,sy) and an average was taken based on total sum(s). That would give a point of best fit allowing for 0.
Where the scale is -3 to +3, a different problem arises and that scaling approach doesn't work because there's no opposite (x,y) to assign the negative values to. Someone who disagrees with point (5,5) is just as likely to be (10,10) or (10,0) as (0,0) or (0,10) so picking a theoretical opposite point across the axis wouldn't fix anything.
All i have been able to propose to this point, without any success in implementing it, is to treat each point as a gravitational element where it either attracts or repels with the scale as the force of attraction/repulsion, attempting to find the equilibrium point of best fit for the system.
Anyone who has implemented such an analysis of Likert scale problems, any pointers on how to approach it?
