Lozano, L., García-Cueto, E., & Muñiz, J. (2008). Effect of the Number of Response Categories on the Reliability and Validity of Rating Scales. Methodology.
Notes:
p.6: Our results permit the conclusion that, considering criteria of reliability and validity, the minimum number of response categories for items with Likert-type format should be al least of four. As regards the ideal number, the data indicate that from seven categories onwards the gains are scarce from a psychometric point of view, suggesting the use of between four and seven. — Highlighted Dec 13, 2015
p.6: it should be ensured that the number of response options will be such that it does not exceed the discriminative capacity of the subjects. If too many alternatives are offered and the subject has problems discriminating between them, there is greater likelihood of the introduction of newmeasurement errors. — Highlighted Dec 13, 2015