The hardware and bandwidth for this mirror is donated by METANET, the Webhosting and Full Service-Cloud Provider.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]metanet.ch.
The equateIRT package computes:
Load the package equateIRT and the data
Estimate a two parameter logistic model for 5 data sets with the R package mirt
library("mirt")
m1 <- mirt(data2pl[[1]], SE = TRUE)
m2 <- mirt(data2pl[[2]], SE = TRUE)
m3 <- mirt(data2pl[[3]], SE = TRUE)
m4 <- mirt(data2pl[[4]], SE = TRUE)
m5 <- mirt(data2pl[[5]], SE = TRUE)
Create an object of class modIRT
(since package versione
2.5.0 it is possible to skip the import.mirt function)
mlist<- list(m1, m2, m3, m4, m5)
test <- paste("test", 1:5, sep = "")
mod2pl <- modIRT(est.mods = mlist, names = test, display = FALSE)
coef(mod2pl$test1)[1:5]
## Dffclt.I1 Dffclt.I2 Dffclt.I3 Dffclt.I4 Dffclt.I5
## 0.058224616 0.028019255 0.073234265 0.415936521 -0.006686047
The linkage plan
## [,1] [,2] [,3] [,4] [,5]
## [1,] 20 10 0 0 10
## [2,] 10 20 10 0 0
## [3,] 0 10 20 10 0
## [4,] 0 0 10 20 10
## [5,] 10 0 0 10 20
A graphic of the linkage plan with package sna
library(sna)
par(mar=c(0, 0, 0, 0))
set.seed(6)
gplot(lplan, displaylabels = TRUE, vertex.sides = 4, vertex.cex = 5, vertex.rot =45, usearrows = FALSE, label.pos = 5, label.cex = 1, vertex.col = 0)
Estimation of direct equating coefficients between Forms 1 and 2 using the mean-mean method.
NOTE: Item parameters are converted to the scale of Form 2.
## Direct equating coefficients
## Method: mean-mean
## Link: test1.test2
## Link: test1.test2
## Method: mean-mean
## Equating coefficients:
## Estimate StdErr
## A 1.21061 0.028998
## B -0.14242 0.028106
Estimation of all direct equating coefficients between forms with common items using the mean-mean method
## Direct equating coefficients
## Method: mean-mean
## Links:
## test1.test2
## test1.test5
## test2.test1
## test2.test3
## test3.test2
## test3.test4
## test4.test3
## test4.test5
## test5.test1
## test5.test4
Direct equating coefficients for Forms 1 and 2
## Link: test1.test2
## Method: mean-mean
## Equating coefficients:
## Estimate StdErr
## A 1.21061 0.028998
## B -0.14242 0.028106
Estimation of all chain equating coefficients of length 4
## Chain equating coefficients
## Method: mean-mean
## Paths:
## test4.test5.test1.test2
## test3.test2.test1.test5
## test5.test1.test2.test3
## test4.test3.test2.test1
## test5.test4.test3.test2
## test1.test2.test3.test4
## test1.test5.test4.test3
## test2.test3.test4.test5
## test2.test1.test5.test4
## test3.test4.test5.test1
Chain equating coefficients for path {1, 2, 3, 4}
## Path: test1.test2.test3.test4
## Method: mean-mean
## Equating coefficients:
## Estimate StdErr
## A 1.25371 0.046538
## B -0.49825 0.038497
Estimation of all chain equating coefficients of length 4 from Form 1 to Form 4
## Chain equating coefficients
## Method: mean-mean
## Paths:
## test1.test2.test3.test4
## Path: test1.test2.test3.test4
## Method: mean-mean
## Equating coefficients:
## Estimate StdErr
## A 1.25371 0.046538
## B -0.49825 0.038497
Estimation of chain equating coefficient for path {1, 5, 4}
pth <- paste("test", c(1,5,4), sep = "")
chainec154 <- chainec(direclist = direclist2pl, pths = pth)
summary(chainec154)
## Path: test1.test5.test4
## Method: mean-mean
## Equating coefficients:
## Estimate StdErr
## A 1.15964 0.033424
## B -0.40028 0.033072
NOTE: Item parameters are converted to the scale of Form 4.
When two forms are linked through more than one path, the equating coefficients can be averaged using the bisector method. Estimation of bisector equating coefficients:
ecall <- c(cec14, chainec154)
fec <- bisectorec(ecall = ecall, weighted = TRUE, unweighted = TRUE)
fec
## Bisector and weighted bisector equating coefficients
## Method: mean-mean
##
## Link: test1.test4
## Paths:
## test1.test2.test3.test4
## test1.test5.test4
## Link: test1.test4
## Method: mean-mean
## Equating coefficients:
## Path Estimate StdErr
## A test1.test2.test3.test4 1.25371 0.046538
## A test1.test5.test4 1.15964 0.033424
## A bisector 1.20559 0.030635
## A weighted bisector 1.18974 0.029330
## B test1.test2.test3.test4 -0.49825 0.038497
## B test1.test5.test4 -0.40028 0.033072
## B bisector -0.44814 0.029917
## B weighted bisector -0.43163 0.029704
Extract the equating coefficients
## link path A B
## 1 test1.test4 test1.test2.test3.test4 1.253712 -0.4982537
## 2 test1.test4 test1.test5.test4 1.159638 -0.4002830
## 3 test1.test4 bisector 1.205588 -0.4481367
## 4 test1.test4 weighted bisector 1.189738 -0.4316305
Extract item parameters of two forms being equated in the original scale and item parameters of the first form converted to the scale of the second form.
## Item test1 test4 test1.as.test4
## 1 Dffclt.I1 0.058224616 NA -0.36235841
## 2 Dffclt.I10 0.654862838 NA 0.34748486
## 3 Dffclt.I2 0.028019255 NA -0.39829488
## 4 Dffclt.I21 NA -0.18966258 NA
## 5 Dffclt.I22 NA -0.57151991 NA
## 6 Dffclt.I23 NA -0.97068963 NA
## 7 Dffclt.I24 NA 0.28221916 NA
## 8 Dffclt.I25 NA 0.02656655 NA
## 9 Dffclt.I26 NA -0.12088864 NA
## 10 Dffclt.I27 NA 0.47589388 NA
## 11 Dffclt.I28 NA -0.01063699 NA
## 12 Dffclt.I29 NA -0.93598686 NA
## 13 Dffclt.I3 0.073234265 NA -0.34450086
## 14 Dffclt.I30 NA 0.54314189 NA
## 15 Dffclt.I31 0.522747833 NA 0.19030260
## 16 Dffclt.I32 0.831399476 NA 0.55751724
## 17 Dffclt.I33 0.575285842 NA 0.25280907
## 18 Dffclt.I34 -0.342019714 NA -0.83854437
## 19 Dffclt.I35 0.428922604 NA 0.07867514
## 20 Dffclt.I36 0.852091712 NA 0.58213558
## 21 Dffclt.I37 0.456475784 NA 0.11145621
## 22 Dffclt.I38 -0.137534728 NA -0.59526078
## 23 Dffclt.I39 0.519128781 NA 0.18599687
## 24 Dffclt.I4 0.415936521 NA 0.06322510
## 25 Dffclt.I40 0.841940468 NA 0.57005826
## 26 Dffclt.I41 NA -1.12849656 NA
## 27 Dffclt.I42 NA -0.77000076 NA
## 28 Dffclt.I43 NA 1.08701579 NA
## 29 Dffclt.I44 NA 0.88830215 NA
## 30 Dffclt.I45 NA 0.22917144 NA
## 31 Dffclt.I46 NA -0.78046634 NA
## 32 Dffclt.I47 NA 0.13591938 NA
## 33 Dffclt.I48 NA -0.21787300 NA
## 34 Dffclt.I49 NA -0.05849482 NA
## 35 Dffclt.I5 -0.006686047 NA -0.43958511
## 36 Dffclt.I50 NA -0.06296350 NA
## 37 Dffclt.I6 -0.773019644 NA -1.35132145
## 38 Dffclt.I7 0.155039212 NA -0.24717439
## 39 Dffclt.I8 0.337353040 NA -0.03026867
## 40 Dffclt.I9 -0.182437547 NA -0.64868338
## 41 Dscrmn.I1 1.076092086 NA 0.90447806
## 42 Dscrmn.I10 1.341200299 NA 1.12730710
## 43 Dscrmn.I2 1.123453456 NA 0.94428629
## 44 Dscrmn.I21 NA 0.96336363 NA
## 45 Dscrmn.I22 NA 1.01240417 NA
## 46 Dscrmn.I23 NA 1.05633835 NA
## 47 Dscrmn.I24 NA 0.87356846 NA
## 48 Dscrmn.I25 NA 1.06908275 NA
## 49 Dscrmn.I26 NA 1.10250771 NA
## 50 Dscrmn.I27 NA 1.04526491 NA
## 51 Dscrmn.I28 NA 0.93408343 NA
## 52 Dscrmn.I29 NA 0.93004199 NA
## 53 Dscrmn.I3 1.091380106 NA 0.91732796
## 54 Dscrmn.I30 NA 1.11220053 NA
## 55 Dscrmn.I31 1.483887762 NA 1.24723892
## 56 Dscrmn.I32 1.300806488 NA 1.09335525
## 57 Dscrmn.I33 1.353845457 NA 1.13793562
## 58 Dscrmn.I34 1.381906844 NA 1.16152181
## 59 Dscrmn.I35 1.272436496 NA 1.06950967
## 60 Dscrmn.I36 1.008690352 NA 0.84782549
## 61 Dscrmn.I37 1.288728736 NA 1.08320364
## 62 Dscrmn.I38 1.473055586 NA 1.23813424
## 63 Dscrmn.I39 1.342341968 NA 1.12826670
## 64 Dscrmn.I4 1.281925510 NA 1.07748539
## 65 Dscrmn.I40 1.457139147 NA 1.22475614
## 66 Dscrmn.I41 NA 1.03344365 NA
## 67 Dscrmn.I42 NA 0.96422272 NA
## 68 Dscrmn.I43 NA 0.99182420 NA
## 69 Dscrmn.I44 NA 0.95219518 NA
## 70 Dscrmn.I45 NA 1.13447404 NA
## 71 Dscrmn.I46 NA 1.18533478 NA
## 72 Dscrmn.I47 NA 1.04662444 NA
## 73 Dscrmn.I48 NA 0.94374714 NA
## 74 Dscrmn.I49 NA 0.80777460 NA
## 75 Dscrmn.I5 1.007880221 NA 0.84714455
## 76 Dscrmn.I50 NA 1.05217297 NA
## 77 Dscrmn.I6 0.949042894 NA 0.79769054
## 78 Dscrmn.I7 1.057666073 NA 0.88899061
## 79 Dscrmn.I8 1.201282017 NA 1.00970284
## 80 Dscrmn.I9 0.976907845 NA 0.82111162
Equated scores with the true score equating method
## The following scores are not attainable: 0
## theta test4 test1.as.test4 StdErr
## 1 -3.2018628 1 1.015953 5.003132e-02
## 2 -2.4342926 2 1.955948 6.878631e-02
## 3 -1.9548627 3 2.893737 7.874430e-02
## 4 -1.5906884 4 3.841937 8.359716e-02
## 5 -1.2874493 5 4.805130 8.512002e-02
## 6 -1.0206725 6 5.784930 8.446458e-02
## 7 -0.7769609 7 6.781477 8.251528e-02
## 8 -0.5479067 8 7.794013 8.003024e-02
## 9 -0.3275850 9 8.821124 7.769419e-02
## 10 -0.1113201 10 9.860811 7.609765e-02
## 11 0.1050414 11 10.910495 7.564081e-02
## 12 0.3256710 12 11.966967 7.639765e-02
## 13 0.5553007 13 13.026319 7.802189e-02
## 14 0.7999529 14 14.083854 7.975624e-02
## 15 1.0681829 15 15.133999 8.052101e-02
## 16 1.3736059 16 16.170197 7.900210e-02
## 17 1.7410355 17 17.184780 7.366228e-02
## 18 2.2254147 18 18.168749 6.256879e-02
## 19 3.0012603 19 19.111255 4.258250e-02
## 20 42.1302143 20 20.000000 4.009435e-15
Equated scores with the observed score equating method
## test4 test1.as.test4 StdErr
## 1 0 -0.01421594 0.03617689
## 2 1 0.95379681 0.05480849
## 3 2 1.90979310 0.06631698
## 4 3 2.86661488 0.07347523
## 5 4 3.83082127 0.07759881
## 6 5 4.80579296 0.07950051
## 7 6 5.79315712 0.07984123
## 8 7 6.79342490 0.07915320
## 9 8 7.80635915 0.07786227
## 10 9 8.83092394 0.07643819
## 11 10 9.86541737 0.07532277
## 12 11 10.90798519 0.07475095
## 13 12 11.95636186 0.07472299
## 14 13 13.00728219 0.07500067
## 15 14 14.05690943 0.07510454
## 16 15 15.10136220 0.07434567
## 17 16 16.13633708 0.07187445
## 18 17 17.15705121 0.06680167
## 19 18 18.15894672 0.05832150
## 20 19 19.13833584 0.04577955
## 21 20 20.09342215 0.02877984
A comparison of equated scores obtained with 2 different chains, bisector and weighted bisector methods.
## theta test4 test1.as.test4 StdErr
## 1 1.741036 17 17.2431 0.09998233
## theta test4 test1.as.test4 StdErr
## 1 1.741036 17 17.06712 0.1618202
## theta test4 test1.as.test4 StdErr
## 1 1.741036 17 17.15485 0.08281067
## theta test4 test1.as.test4 StdErr
## 1 1.741036 17 17.18478 0.07366228
Load the data
Create a dataset for each group and estimate a 2PL model for each group using the R package mirt
library(mirt)
data1 <- dataDIF[dataDIF$group == 1, 1:20]
data2 <- dataDIF[dataDIF$group == 2, 1:20]
data3 <- dataDIF[dataDIF$group == 3, 1:20]
mod1 <- mirt(data1, SE = TRUE)
mod2 <- mirt(data2, SE = TRUE)
mod3 <- mirt(data3, SE = TRUE)
Perform the test for DIF on two groups
##
## Test for Differential Item Functioning
##
## Item parameters tested for DIF: intercept and slope
## Equating method used: mean-mean
## Reference group: T1 Focal group: T2
## Item purification not applied
##
## statistic p.value
## I01 38.605 4.14e-09 ***
## I02 1.954 0.376
## I03 3.888 0.143
## I04 0.025 0.988
## I05 1.753 0.416
## I06 1.291 0.524
## I07 3.433 0.180
## I08 0.181 0.914
## I09 0.686 0.710
## I10 0.517 0.772
## I11 3.237 0.198
## I12 0.891 0.640
## I13 0.464 0.793
## I14 1.296 0.523
## I15 1.591 0.451
## I16 0.962 0.618
## I17 2.856 0.240
## I18 0.159 0.924
## I19 0.627 0.731
## I20 1.861 0.394
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Perform the test for DIF on three groups
##
## Test for Differential Item Functioning
##
## Item parameters tested for DIF: intercept and slope
## Equating method used: mean-mean
## Reference group: T1 Focal groups: T2, T3,
## Item purification not applied
##
## statistic p.value
## I01 164.585 <2e-16 ***
## I02 3.331 0.504
## I03 4.862 0.302
## I04 5.300 0.258
## I05 3.368 0.498
## I06 1.323 0.858
## I07 4.605 0.330
## I08 1.556 0.817
## I09 1.449 0.836
## I10 4.122 0.390
## I11 3.874 0.423
## I12 5.300 0.258
## I13 2.859 0.582
## I14 2.770 0.597
## I15 3.640 0.457
## I16 4.158 0.385
## I17 5.924 0.205
## I18 2.080 0.721
## I19 2.123 0.713
## I20 7.399 0.116
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
It is possible to change the reference group, the equating method used, and to apply purification.
res_diftest3 <- dif.test(est.mods = list(mod1, mod2, mod3), reference = 2,
method = "Haebara", purification = TRUE)
res_diftest3
##
## Test for Differential Item Functioning
##
## Item parameters tested for DIF: intercept and slope
## Equating method used: Haebara
## Reference group: T2 Focal groups: T1, T3,
## Item purification applied. Significance level = 0.05
##
## statistic p.value
## I01 168.026 <2e-16 ***
## I02 2.553 0.635
## I03 2.921 0.571
## I04 4.263 0.372
## I05 3.395 0.494
## I06 0.648 0.958
## I07 2.796 0.592
## I08 2.796 0.593
## I09 1.668 0.796
## I10 2.405 0.662
## I11 1.972 0.741
## I12 4.116 0.391
## I13 1.721 0.787
## I14 2.381 0.666
## I15 4.668 0.323
## I16 2.148 0.709
## I17 4.457 0.348
## I18 1.301 0.861
## I19 2.900 0.575
## I20 3.497 0.478
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
The identity test performs a statistical test to verify if the chain equating coefficients from one form to itself are A=1 and B=0.
data(est3pl)
test <- paste("test", 1:5, sep = "")
mod3pl <- modIRT(coef = est3pl$coef, var = est3pl$var, names = test, display = FALSE)
direclist3pl <- alldirec(mods = mod3pl, method = "Haebara")
pth3 <- paste("test", c(1:5,1), sep = "")
chainec_circle <- chainec(direclist = direclist3pl, pths = pth3)
summary(chainec_circle)
## Path: test1.test2.test3.test4.test5.test1
## Method: Haebara
## Equating coefficients:
## Estimate StdErr
## A 1.0868759 0.063585
## B 0.0022126 0.041833
##
## Identity test
##
## path: test1.test2.test3.test4.test5.test1
## statistic = 2.349279, df = 2, p-value = 0.3089304
The null hypothesis A=1 and B=0 is not rejected.
It is also possible to performs a statistical test to verify if the chain equating coefficients that link the same two forms are equal.
In the following example test 1 and 5 are linked through two different paths giving two different pairs of equating coefficients. The example uses the 3PL models and the Haebara method, though other options are possible.
pth3 <- paste("test", 1:5, sep = "")
chainec3 <- chainec(direclist = direclist3pl, pths = pth3)
ecall <- c(chainec3, direclist3pl["test1.test5"])
summary(chainec3)
## Path: test1.test2.test3.test4.test5
## Method: Haebara
## Equating coefficients:
## Estimate StdErr
## A 1.06595 0.056590
## B -0.50368 0.042982
## Link: test1.test5
## Method: Haebara
## Equating coefficients:
## Estimate StdErr
## A 1.00935 0.029052
## B -0.51886 0.027863
##
## Scale drift test
##
## link: test1.test5
## statistic = 1.623208, df = 2, p-value = 0.4441452
The null hypothesis of equality of the equating coefficients is not rejected.
These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.