Table 11.1 Item misfitting responses

(controlled by FITI=, LINELENGTH=, MNSQ=, OUTFIT=, T7OPTIONS=)

 

These tables show the items for which the t standardized outfit (or infit, if OUTFIT=N) statistic is greater than the misfit criterion (FITI=). FITI=-10 with MNSQ=No  displays all items. Items are listed in descending order of misfit. The response codes are listed in their sequence order in your data file. The residuals are standardized response score residuals, which have a model led expectation of 0, and a variance of 1. Negative residuals indicate that the observed response was less correct (or, for rating (or partial credit) scales, lower down the rating scale) than expected. The printed standardized residual is truncated, not rounded, so that its actual value is at least as extreme as that shown. Standardized residuals between -1 and 1 are not printed. For exact details, see XFILE=. "M" indicates a missing response.

 

"X" indicates that the person obtained an extreme score."X" in Tables 7.1 and 11.1 indicates that the observation is part of an extreme score, and so does not have a standardized residual.  All residuals for extreme scores are conceptually zero. Persons with "X" in Item Table 11.1 (because they have extreme scores) do not appear in Table 7.1, and vice-versa, because those extreme persons and items are not part of the misfit process. Basically "X" says "this observation is excluded from the misfit computation".

 

For Table 11, the diagnosis of misfitting items, items with a t standardized fit greater than FITI= are reported. Selection is based on the OUTFIT statistic, unless you set OUTFIT=N in which case the INFIT statistic is used. T7OPTIONS= selects the detail lines to display:

 

TABLE OF POORLY FITTING ITEMS     (PERSONS IN ENTRY ORDER)

NUMBER   NAME    POSITION        MEASURE   INFIT (ZSTD) OUTFIT   MISFIT OVER 2.0

 

 23 WATCH A RAT                   2.00     5.8   A    8.1

     RESPONSE:    1:  0 2 1 1 1  2 2 0 2 0   0 1 0 1 1  0 1 0 0 0   0 1 1 0 0

   Z-RESIDUAL:          X     2  3     3       2

 

     RESPONSE:   26:  1 2 0 2 1  M 0 0 1 1   1 0 1 0 0  1 0 0 0 0   0 0 0 2 1

   Z-RESIDUAL:          3   6 2                        -2      -2         4

 

                                                 v This is on the fit plots

  5 FIND BOTTLES AND CANS         2.21     5.2   B    6.5

     RESPONSE:    1:  1 2 0 0 1  2 0 0 1 0   0 2 0 0 1  1 0 0 0 0   0 1 1 0 1

   Z-RESIDUAL:          X     2  4             6         -2

 

Example: Table 11.1 with T7OPTIONS= OERZ  in an analysis of Example0.txt:

 

TABLE OF POORLY FITTING ACT   (KID IN ENTRY ORDER)

NUMBER - NAME -- POSITION ------ MEASURE - INFIT (MNSQ) OUTFIT

 

     23  Watch a rat                2.18     2.4   A    4.1

  OBSERVED:      1:  0    2    1    1    1     2    2    0    2    0    <- T7OPTIONS= O

  EXPECTED:          0.4  X    0.5  0.3  0.1   0.2  1.2  0.5  0.2  0.3  <- T7OPTIONS= E

  RESIDUAL:         -0.4       0.5  0.7  0.9   1.8  0.8 -0.5  1.8 -0.3  <- T7OPTIONS= R

Z-RESIDUAL:                              2     4              4         <- T7OPTIONS= Z

 

The OBSERVED value is the scored response, before recounting due to STKEEP=No, if any.

The EXPECTED value is the OBSERVED value - RESIDUAL value.