TCCFILE= specifies the file to hold the coordinates for the Test Characteristic Curve and the Test Information Function. See Test Characteristic Curve and the Test Information Function in the Graph Window, also Table 20. Also the accumulated probabilities across rating-scale categories for all items.
For customized range and increments, use TCCHIGH=, TCCLOW=, TCCINCR=. Decimal places: UDECIM=
; TCC + TIF CURVES MEASURE SCORE INFO SEM 0 1 2 -8.00 .05 .05 4.31 24.95 .05 .00 -7.98 .06 .05 4.27 24.94 .05 .00 -7.97 .06 .06 4.24 24.94 .06 .00 -7.95 .06 .06 4.20 24.94 .06 .00 -7.94 .06 .06 4.17 24.94 .06 .00 -7.92 .06 .06 4.14 24.94 .06 .00 -7.90 .06 .06 4.11 24.94 .06 .00 |
The contents of the TCCFILE= |
MEASURE |
The person ability measure (theta in logits or user-scaled USCALE= units) |
SCORE |
Test Characteristic Curve, TCC: the expected score on the test for the MEASURE |
INFO |
Test Information Function, TIF: the statistical information in the score |
SEM |
Standard Error of Measurement for the MEASURE when estimated from the SCORE = 1 / √ INFO |
0 |
Cumulative probability of lowest category (with lowest category numbered 0) |
... |
Cumulative probability of intermediate categories |
Highest |
Cumulative probability of highest category |
NORMED, FREQUENCY, CUM.FREQ, PERCENTILE |
See SCOREFILE= |
Example 1: www.rasch.org/memo1.htm Figure 2 shows the TCC for two person groups on the same plot. How can I do this?
These needs two analyses, but only one control file and data file. Put the group code in column 1 of the person label. Then, here is the procedure with Example0.txt - since item 19 is extreme for one group, it is omitted with IDELETE=19 from both groups
analysis 1: at the Extra Specification prompt, type the person group code for the first group:
PSELECT=F
do the analysis, then
Output Files menu:
Ouput TCCFILE= to Excel temporary file
analysis 2: at the Extra Specification prompt, type the person group code for the second group:
PSELECT=M
do the analysis, then
Output Files menu:
Ouput TCCFILE= to Excel
Excel plot:
copy the first two columns of both Excel files into one worksheet.
scatterplot the MEASURE columns against the matching SCORE columns
Here is what I see:
Example 2: The Scale Characteristic Curves in Dogan (2018) "An Application of the Partial Credit IRT Model in Identifying Benchmarks for Polytomous Rating Scale Instruments", PARE, 23:7.
Output the TCCFILE= to Excel and plot the cumulative probability curves against the measures. Here they are for Example0.txt using the Rating Scale Model:
