Rg. Downey et Cv. King, MISSING DATA IN LIKERT RATINGS - A COMPARISON OF REPLACEMENT METHODS, The Journal of general psychology, 125(2), 1998, pp. 175-191
The effects of using two methods (item mean and person mean) for repla
cing missing data in Likert scales were studied. The results showed th
at both methods were good representations of the original data when bo
th the number of respondents with missing data and the number of items
missing were 20% or less. As the numbers of missing items and of resp
ondents with missing data increased for the person mean substitution m
ethod, a spurious increase in the inter-item correlations (and, theref
ore, reliability) for the scale was produced. The item mean substituti
on reduced the reliability estimates of the scale. These results sugge
st caution in the use of the person mean substitution method as the nu
mbers of missing items and respondents increase.