As I recall, Likert scales are agree-disagree (well, strongly disagree
to strongly agree) scales (usually in either 5 or 7 gradations). The
items relate to the extent of agreement with a statement per se
(theoretically anyway) rather than the underlying phenomenon. Any
way, it would look like this:
This product satisfies my needs. Strongly disagree disagree
slightly disagree neither agree nor disagree slightly agree
agree strongly agree
If you were to use this for academic research, you would ask
respondents to indicate their level of agreement with several other
statements as well (This product does not meet my needs, I am
satisfied with this product, I would buy this product again, I would
recommend this product to a friend, ...). Then you would *****s the
reliability (maybe with Cronbach's alpha) and factor structure (using
factor analysis). I'm not sure about other software, but it's fairly
easy on SPSS (even if you only have the base system).
Then, you use the valid items to create a summated scale (add the
scores, using reverse-coding for items like I"This product does not
meet my needs." Most social scientists would use parametric
procedures (mean, standard deviation, t-test) for the summated scale
even if they wouldn't use them for a single item (after all, the level
of measurement is ordinal, not interval.).
|
