Simulated file specifications

Winsteps uses two methods  to simulate data:

1) probabilistically-generated data based on anchored parameter estimates

2) re-sampling-with-replacement from the current dataset


Winsteps uses the estimated (or anchored) person, item and Andrich Thresholds or person-response-string re-sampling-with-replacement to simulate a data file equivalent to the raw data. This can be used to investigate the stability of measures, distribution of fit statistics and amount of statistical bias. Each time SIFILE= is run, or selected from the Output Files pull-down menu, a simulated data file produced. Do simulated analyses with several simulated datasets to verify their overall pattern.




The parts of the dialog box outside the red rectangle are described in Output file specifications.

The file format matches the input data file if both are in fixed-field format. When SIFILE= is written with CSV=Y, comma-separated or CSV=T, tab-separated, the item responses precede the person label.

Simulated data files:

invoked with SIFILE=

Number of files:

SINUMBER=, specifies the number of simulated files to produce. If SINUMBER= is greater than 1, then the data file name is automatically incremented, and so is the SISEED= pre-set seed value

Seed number (0 for random):

SISEED=, controls whether the pseudo-random number generator is seeded with the system clock (0 or 1), or with a user-chosen value, (2 and above)

Simulate: use measures or use the data

SIMEASURE=, chooses whether the simulated data is generated from the estimated measures (use measure), or by re-sampling from the observed data (use the data).

If you wish to over-ride the estimated measures, then use IAFILE=, PAFILE= and SAFILE=

Re-sample persons: No or Yes: Persons

SIRESAMPLE=, controls whether re-sampling occurs (sampling with or without replacement), and, if it does, how many person records to include in the simulated data file

Complete data: Yes or No - allow missing data

SICOMPLETE=, Yes for complete data. No for missing data patterns to be repeated in the simulated data file

Extreme scores: Yes or No - avoid extreme scores

SIEXTREME=, Yes to allow the simulated data to include extreme (zero, minimum  possible or perfect, maximum possible) scores. No to avoid generating extreme scores (when possible)


Winsteps simulates data two ways:

i) using the parameter values (persons, items, Andrich thresholds) from the current analysis. This way generates simulated data according to the probabilistic distributions defined by the Rasch model and the generating Rasch parameters. This is for investigations relating to exact Rasch conditions.

ii) by resampling (with replacement) the data (observations, responses) in the current analysis. This way generates data according to the empirical distribution of the generating data and is for investigations relating to datasets like this one.


Example 0 . KCT.txt simulated with CSV=N fixed field format (re-sampling response strings):


Dorothy F 111111111100000000       -.2594          13

Elsie   F 111101111100000000      -1.3696          14

Thomas  M 111111111010000000       -.2594          31

Rick    M 111111111010000000       -.2594          27


KCT.txt simulated with comma-separated, CSV=Y, HLINES=Y, QUOTED=Y format  (re-sampling person measures):


"1-4","2-3","1-2-4","1-3-4","2-1-4", ... ,"KID","Measure","Entry"

1,1,1,1,1,1,0,1,1,1,0,0,0,0,0,0,0,0,"Rick    M",-.2594,27

1,1,1,1,1,1,1,1,1,0,1,0,0,0,0,0,0,0,"Helen   F",-.2594,16

1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,"Rod     M",1.9380,28

1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,"William M",.9229,34


Example 1. It is desired to investigate the stability of Rasch measures.

(1) Estimate measures from your control and data files (e.g., SF.txt)

(2) Choose SIFILE= from the Output Files menu.

(3) Choose to output a permanent file:



(4) Simulated data filename: SFSIMUL.TXT

(5) Rerun Winsteps with your Winsteps control file and DATA=SFSIMUL.TXT on the "Extra Specifications" line.

(6) Compare person, item and Andrich Thresholds.


Example 2. To estimate the measure standard errors in a linked equating design.

1. Do a concurrent calibration with Winsteps

2. Simulate data files SIFILE= from the Output Files menu.

Specify "complete data" SICOMPLETE= as "No" to maintain the same data pattern.

Save 10 simulated sets, SINUMBER=, as S.txt S2.txt .....

3. Rerun your Winsteps analysis 10 times

Specify in Extra Specifications "DATA=S.txt PFILE=P1.txt CSV=TAB" etc.

This will produce 10 PFILE=s. Export them in Excel format.

4. Use Excel to compute the standard deviation of the measures for each person based on the 10 person measures

5. These are the model standard errors for the equating design for the measures.

6. Inflate these values by 20%, say, to allow for systematic equating errors, misfit, etc.


Example 3. If you do need estimation-bias-correction (STBIAS=) that is as accurate as possible with your data set, you will need to discover the amount of bias in the estimates and correct for it:

1. In your control file, STBIAS=No and USCALE=1

2. Obtain the Winsteps estimates for your data

3. Simulate many datasets using those estimates. (SIFILE= on the Winsteps Output Files menu).

4. Obtain the Winsteps estimates from the simulated data sets

5. Regress the simulated estimates on your initial estimates. These will give a slope near 1.0.

6. Obtain the Winsteps estimates for your data with USCALE = 1/slope. The set of estimates in 6 is effectively unbiased.


Example 4. You need to simulate data from generating values. You can use Winsteps to simulate a dataset.


1. Obtain the generating item difficulties, person abilities and threshold values. If you need a normal distribution of person abilities, you can generate this with Excel.


a. From your standard analysis, output IFILE=if.txt, SFILE=sf.txt


b. Use Excel or similar to simulate a normal distribution of person abilities with the mean and S.D. that you want.

In Excel:

Cell A1 = Mean

Cell B1 = S.D.

Cell A2 = =ROW()-1

Cell B2 = =NORMSINV(RAND())*$B$1 +$A$1

then copy A2, B2 for as many rows as you want the sample size.


c. Copy Columns A and B into a text file, pf.txt. Delete row 1.


For a random uniform distribution of item difficulties,  use the Excel formula:

=(item difficulty range)*(RAND() - 0.5) + (mean item difficulty)


d. In your Winsteps control file:



PAFILE = pf.txt

SIFILE= simulated.txt


2. Construct a Winsteps control file including the generating values in IAFILE=, PAFILE=, SAFILE=


3. Make a rectangular dataset with a row of valid responses (can be the same one) as wide as the number of items

and with a column of valid responses (can be the same one) as long as the number of persons.


For instance, number of persons = 2000, number of items =100, then an artificial dichotomous Winsteps control file and dataset can be:







1-2000 1-100 1



1 2.3

2 1.5




1 3.1

2 -2.8




0 0

1 0





and nothing else.


4. Run Winsteps. Ignore the results of this analysis. Choose SIFILE= option from the output files menu. Click on "complete data" to simulate the entire data matrix.


Example 5. Bootstrap estimation of the confidence interval for a reliability.

Bootstrapping is based on generating new datasets using "sampling with replacement" from the current dataset. The new datasets can be generated in Winsteps using:

Simulate: use the data

Re-sample persons: yes, with same number of rows as the original data.


Compute the reliability of each new dataset. See "Performing multiple simulations in Batch mode".


After transformation with Fisher's z, the distribution of the linearized reliabilities (mean and S.D.) are the linearized expected value and linearized S.E. of the observed linearized reliability for the original data set. Transform the linearized (mean-S.D.), mean, and (mean+S.D.) back to the original reliability metric by reversing the Fisher 's z transformation.


Example 6. Multiple simulations in Batch mode. See BATCH=

These can construct bootstrap confidence intervals for DIF estimates, etc.

Set up 100 simulations in a batch file, and extract the relevant numbers from the 100 output DIF tables.

PowerGREP is great software for extracting values from files. For instance:
To pick out lines 10-35 in the output files (after line 9, for 26 lines):

Action type: Search

File sectioning: Search and collect sections

Section search: \A([^\r\n]*+\r\n){9}(([^\r\n]*+\r\n){0,26})

Section collect: \2

Search: the search string: .* for everything



Performing multiple simulations in Batch mode

1. Use NotePad to create a text file called "Simulate.bat"

2. In this file:


REM - produce the generating values: this example uses example0.txt:

START /WAIT c:\Winsteps\Winsteps BATCH=YES  example0.txt example0.out.txt PFILE=pf.txt IFILE=if.txt SFILE=sf.txt


REM - initialize the loop counter

set /a test=1



REM - simulate a dataset - use anchor values to speed up processing (or use SINUMBER= to avoid this step)

START /WAIT c:\Winsteps\Winsteps  BATCH=YES example0.txt example0%test%.out.txt PAFILE=pf.txt IAFILE=if.txt SAFILE=sf.txt SIFILE=SIFILE%test%.txt SISEED=0


REM - estimate from the simulated dataset

START /WAIT c:\Winsteps\Winsteps BATCH=YES example0.txt data=SIFILE%test%.txt SIFILE%test%.out.txt pfile=pf%test%.txt ifile=if%test%.txt sfile=sf%test%.txt

REM - do 100 times

set /a test=%test%+1

if not "%test%"=="101" goto loop



3. Save "Simulate.bat", then double-click on it to launch it.

4. The simulate files and their estimates are numbered 1 to 100.


5. The files of estimates can be combined and sorted using MS-DOS commands, e.g.,

Copy if*.txt combinedif.txt

Sort /+(sort column) <combinedif.txt  >sortedif.txt



6. Individual lines from the output files can be written to one file using MS-DOS batch commands. For instance, using an MS-DOS batch routine (.bat or .cmd), the same text line can be extracted from many text files and output into a new text file. The new text file can be be pasted into Excel. Save these MS-DOS commands as extract.bat in the folder that has the files of statistics. Double click on extract.bat to execute it.


rem replace 2 with the number of lines to skip before the line you want

@echo off

setlocal EnableDelayedExpansion

if exist result.csv del result.csv

for %%f in (*.txt) do (

 echo %%f

 set i=a

 for /F "skip=2 delims=" %%l in (%%f) do (

   if "!i!" == "a" echo %%f, %%l >> result.csv

   set i=b



notepad result.csv



Help for Winsteps Rasch Measurement Software: Author: John Michael Linacre

The Languages of Love: draw a map of yours!

For more information, contact or use the Contact Form

Facets Rasch measurement software. Buy for $149. & site licenses. Freeware student/evaluation download
Winsteps Rasch measurement software. Buy for $149. & site licenses. Freeware student/evaluation download

State-of-the-art : single-user and site licenses : free student/evaluation versions : download immediately : instructional PDFs : user forum : assistance by email : bugs fixed fast : free update eligibility : backwards compatible : money back if not satisfied
Rasch, Winsteps, Facets online Tutorials


Forum Rasch Measurement Forum to discuss any Rasch-related topic

Click here to add your email address to the Winsteps and Facets email list for notifications.

Click here to ask a question or make a suggestion about Winsteps and Facets software.

Rasch Publications
Rasch Measurement Transactions (free, online) Rasch Measurement research papers (free, online) Probabilistic Models for Some Intelligence and Attainment Tests, Georg Rasch Applying the Rasch Model 3rd. Ed., Bond & Fox Best Test Design, Wright & Stone
Rating Scale Analysis, Wright & Masters Introduction to Rasch Measurement, E. Smith & R. Smith Introduction to Many-Facet Rasch Measurement, Thomas Eckes Invariant Measurement with Raters and Rating Scales: Rasch Models for Rater-Mediated Assessments, George Engelhard, Jr. & Stefanie Wind Statistical Analyses for Language Testers, Rita Green
Rasch Models: Foundations, Recent Developments, and Applications, Fischer & Molenaar Journal of Applied Measurement Rasch models for measurement, David Andrich Constructing Measures, Mark Wilson Rasch Analysis in the Human Sciences, Boone, Stave, Yale
in Spanish: Análisis de Rasch para todos, Agustín Tristán Mediciones, Posicionamientos y Diagnósticos Competitivos, Juan Ramón Oreja Rodríguez
Winsteps Tutorials Facets Tutorials Rasch Discussion Groups



Coming Winsteps & Facets Events
May 22 - 24, 2018, Tues.-Thur. EALTA 2018 pre-conference workshop (Introduction to Rasch measurement using WINSTEPS and FACETS, Thomas Eckes & Frank Weiss-Motz),
May 25 - June 22, 2018, Fri.-Fri. On-line workshop: Practical Rasch Measurement - Core Topics (E. Smith, Winsteps),
June 27 - 29, 2018, Wed.-Fri. Measurement at the Crossroads: History, philosophy and sociology of measurement, Paris, France.,
June 29 - July 27, 2018, Fri.-Fri. On-line workshop: Practical Rasch Measurement - Further Topics (E. Smith, Winsteps),
July 25 - July 27, 2018, Wed.-Fri. Pacific-Rim Objective Measurement Symposium (PROMS), (Preconference workshops July 23-24, 2018) Fudan University, Shanghai, China "Applying Rasch Measurement in Language Assessment and across the Human Sciences"
Aug. 10 - Sept. 7, 2018, Fri.-Fri. On-line workshop: Many-Facet Rasch Measurement (E. Smith, Facets),
Oct. 12 - Nov. 9, 2018, Fri.-Fri. On-line workshop: Practical Rasch Measurement - Core Topics (E. Smith, Winsteps),



Our current URL is

Winsteps® is a registered trademark

John "Mike" L.'s Wellness Report: I'm 72, take no medications and, March 2018, my doctor is annoyed with me - I'm too healthy!
According to Wikipedia, the human body requires about 30 minerals, maybe more. There are 60 naturally-occurring minerals in the liquid Mineral Supplement which I take daily.