Evaluations from reader services can be ruined by subjectivity creeping into the results. Ratings that range broadly, for example, 3-9 or 4-9, don't yield a valid result. The science of statistics deals with how to arrange scores, numbers in a way to find the truth. It handles skewed numbers in an understandable way. The science can handle a series of running numbers or all numbers reported at one time. I had statistics in college and know it can be very helpful. A good mathematical model would compute averages, and "typical" results to get at the truth. Such a system could easily handle reader subjectivity. It might compute averages, or eliminate skewed results. For example, the model could state so many scores fall outside a typical range, and are either eliminated totally or given a low weight in a computed final score.
I suggest any reader ratings service would be wise to consult with a statistician to develop a mathematicaL model to analyze ratings, so a more valid score or sum rating is produced.
I suggest any reader ratings service would be wise to consult with a statistician to develop a mathematicaL model to analyze ratings, so a more valid score or sum rating is produced.
Comment