Answers are grounded in this paper and cite their sources.
One hundred and twenty eight subjects responded to one of four differently labeled 5-point Likert-type response scales. Although no significant differences in test means and in reliability estimates were found among response scales, the scales did differ on measures of variability. Subjects using scales with more nearly absolute endpoints tended to concentrate their responses on the middle scale steps; those responding to scales with less nearly absolute endpoints tended to use the scale steps in a more balanced manner. Greater item variance was associated with scales having less rather than more nearly absolute endpoints.
No takes yet. Share an insight, caveat, or question.
Wyatt et al. (1987) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: