I hope this help; I am unsure what you are asking.  To assess a floor or ceiling effect, I would use standard descriptive methods such as box plots, numerical skew, etc. Those are certainly available in SPSS.

Tobit analysis is designed to estimate a regression model with data censored by floor or ceiling effects that might be useful to you. I found this presentation helpful: http://www.stat.columbia.edu/~madigan/G6101/notes/logisticTobit.pdf

If the floor or ceiling effects cause your data to become dichotomous (or can easily be collapsed into two categories without much loss of information) and you want to predict that variable, then logistic regression would be appropriate.

Tetrachoric correlations are estimated Pearson correlations estimated from dichotomous data. If you dichotomized your data by coding your most extreme response versus the others, then a matrix of tetrachoric correlations would produce better outcomes for covariance models, including factor analysis, structural equations modeling, and regression.

A completely different traditional approach would be to apply a transformation like square root, log, or inverse to try to make the data more normal.

You could also use item response theory if you have "scales" of Likert items, even if they have floor/ceiling effects. In most instances, scores have a less severe floor or ceiling effect than individual items.

Hi,

Your question: how to evaluate ceiling effect

In cognitive psychology, the measurement of the time to respond to a given stimulus is often of interest. In these measurements, a ceiling may be the lowest possible number ie., the fewest milliseconds to a response, rather than the highest value, as is the usual interpretation of "ceiling".

In response time studies, it may appear that a ceiling had occurred in the measurements due to an apparent clustering around some minimum amount of time.  Such as the 250 ms needed by many people to press a key. However, this clustering could actually represent a natural physiological limit of response time, rather than an artifact of the stopwatch sensitivity (which of course would be a ceiling effect).

Further statistical study, and scientific judgment, can resolve whether or not the observations are due to a ceiling or are the truth of the matter.

References:

1.  Kaufman, Alan S. (2009). IQ Testing 101. New York: Springer Publishing. pp. 151–153. ISBN 978-0-8261-0629-2.

2.  Po, Alain Li Wan (1998). Dictionary of Evidence-based Medicine. Radcliffe Publishing. p. 20. ISBN 978-1-85775-305-9.

3.  Vogt, W. Paul (2005). Dictionary of Statistics & Methodology: A Nontechnical Guide for the Social Sciences (Third ed.). SAGE. p. 40 (entry "ceiling effect"). ISBN 978-0-7619-8855-7.

First, you can look whether your data are skewed – depending on the skew (positive vs. negative), you can then use different kinds of transformations to correct for skewness (e.g. log transformation). If this does not work, you may want to try nonparametric tests (e.g., bootstrapping). 

Thank you all for your helpful comments!

The data are positively skewed and unfortunately this may be due to the topic of questionnaire, which is about ethics.  Responses are given on likert type scale from 1-6. If I correct the skeweness, wouldn't  I lose data? I dont know if I can dichotomize my data, but wouldn't  I lose info again?

Does IRT provide specific analysis for ceiling/ floor effects? (As far as I recall It provides different type of information about my items, but I dont know whether is helpfull in this case).