-------------------------------------------------------------------------------
help for delta                                             Jean-Benoit Hardouin
-------------------------------------------------------------------------------

Syntax

        delta varlist [if exp] [in range] [, ci(#) nodots minscore(#)
              maxscore(#)]

Description

    delta computes the generalized delta index of scale discrimination
        developed by Hankins (2007) based on the original work of Ferguson
        (1949). This index measures the scale's ability to distinguish
        between individuals.  A value of 1 indicates that the test has
        maximal discrimination (all possible scores occur with the same
        frequency) and a value of 0 means that the test has minimal
        discrimination (all the respondents have the same score).  A value of
        0.9 results from a set of scores that is normally distributed. A
        value of 1 is observed if the scores follow a uniform distribution.
        Individuals with a missing score are omitted.

Options

    If varlist is composed of only one variable, the delta module considers
        that this variable is the score of the individuals.

    ci(#) estimates the confidence interval by boostrap. # is the number of
        replications to be performed. By default, no confidence interval is
        calculated.

    nodots avoids displaying a dot for each replication (only with ci).

    minscore(#) defines the minimal value of the score. By default, this
        value is fixed to 0.

    maxscore(#) defines the maximal value of the score. By default, the
        maximal observed score is used.

Output

    r(delta): Observed value of the delta index.

Examples

    . delta itemA*

    . delta itemA*, ci(500) dots

    . delta score, scoremax(8)

 References

    Ferguson G. A. (1949) On the theory of test discrimination.
        Psychometrika, 14: 61-68.

    Hankins M. (2007) Questionnaire discrimination: (re)- introducting
        coefficient delta. BMC Medical Research Methodology, 7: 19.

Author

    Jean-Benoit Hardouin, PhD, assistant professor
    Team of Biostatistics, Clinical Research and Subjective Measures in
        Health Sciences
    University of Nantes - Faculty of Pharmaceutical Sciences
    1, rue Gaston Veil - BP 53508
    44035 Nantes Cedex 1 - FRANCE
    Email:  jean-benoit.hardouin@univ-nantes.fr
    Websites AnaQol and FreeIRT

Also see

    Online: help for alpha and loevH if installed.