IRT Modeling Lab

2. Dichotomization (optional)

You may wish to use the data with dichotomous IRT models. In order to do that the data must be scored such that there are only two options for each item.

In the data set provided, each item has 5 options. Therefore, you will need to reduce the number of options to 2.

Deciding how to collapse multiple options into only 2 options.

There are different strategies to collapse options. Most of the time, a simple split in the middle will do. However, if you have an odd number of options as in the example below, the conceptual approach is usually taken to decide what to do with the middle variable.

Example

  • Are you Agreeable?
    Disagree Strongly   Disagree   Neither   Agree   Agree Strongly  

    In this case, we view 'Neither' as a non-endorsement of the criteria. Therefore we split the data into Non-endorsed(Disagree Strongly, Disagree, Neither) and Endorsed (Agree, Agree Strongly). It is good practice to make this split uniformly for the entire dataset.

    However, if you have items which do not have an apparent point to split the data.

    Example

  • I was denied training because of my gender
    Never   Once a month   2 to 3 times a month   Once a week   Every day  

    In this situation, it may make sense to select never as a Non-endorsed option and everything else as Endorsed.

    In our example, we will be splitting the data into Non-endorsed(Disagree Strongly, Disagree, Neither) and Endorsed (Agree, Agree Strongly). Below is the SPSS syntax we used to recode multiple option items into dichotomous items.

    SPSS Syntax:
      *Dichotomize items.
        * Get reversed scored file.

    GET
        FILE='C:\IRTtutorial\raw_reversed.SAV'.
    EXECUTE .

        * Dichotomize.
    recode
    c1 c2 c3 c4 c5 c6 c7 c8 c9 c10 a1 a2 a3 a4 a5 a6 a7 a8 a9 a10
    (1=0) (2=0) (3=0) (4=1) (5=1).
    Execute.

    SAVE OUTFILE='C:\IRTtutorial\raw_dichot.SAV'.
    Execute.
         		 

    If you ran the syntax above, SPSS would have created a new file raw_dichot which has the items scored dichotomously. You may download the file in order to compare your file with ours.

    Continue to next step
    		•