Better Use of Scales as Measuring Instruments in Mental Disorders

Objective: Linearity implies high correlation but the converse is not true. For meaningful application of correlation and other descriptive and inferential analysis, checking of linearity, and assumptions of correlation including normality are needed The paper describes method of converting ordinal scores from Likert/ Rating scale to continuous, monotonic, equi-distant scores with fixed zero point and following normal distribution. Method: The method involves selection of weights for different response categories of different items so that weighted item score forms an arithmetic progression. Normalization of such score followed by further weights to items to ensure equal item-total correlation justifies addition of such converted scores. Results: Converted scores satisfying many desired properties can assess progress/deterioration of a patient over time and facilitate comparison, ranking, classification and assessing effectiveness of treatment plan. It also helps in computing reliability of Likert or Rating scale avoiding assumptions of Cronbach’s alpha. Conclusions: Converted scores will help the researchers and practitioners to find improved content validity and meaningfully undertake correlation and other analysis under parametric set up for deriving useful and valid conclusions about the sample, population and test parameters