This main function estimates true subscores using different methods based on original CTT scores.

Description

This function estimates true subscores using methods introduced in studies of Haberman (2008) <doi:10.3102/1076998607302636> and Wainer et al. (2001) <doi:10.4324/9781410604729>. Hypothesis tests (i.e., Olkin' Z,Williams's t, and Hedges-Olkin's Z) are used to determine whether a subscore or an augmented subscore has added value. Codes for the hypothesis tests are from Sinharay (2019) <doi: 10.3102/1076998618788862>.

Usage

CTTsub(test.data, method = "Haberman")

Arguments

Value

References

Haberman, S. J. (2008). "When can subscores have value?." Journal of Educational and Behavioral Statistics, 33(2), 204-229. doi:10.3102/1076998607302636.

Sinharay, S. (2019). "Added Value of Subscores and Hypothesis Testing." Journal of Educational and Behavioral Statistics, 44(1), 25-44. doi:10.3102/1076998618788862.

Wainer, H., Vevea, J., Camacho, F., Reeve, R., Rosa, K., Nelson, L., Swygert, K., & Thissen, D. (2001). "Augmented scores - "Borrowing strength" to compute scores based on small numbers of items." In Thissen, D. & Wainer, H. (Eds.), Test scoring (pp.343 - 387). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. doi:10.4324/9781410604729.

Examples

# Transferring original scored data to a list format
# that can be used in other functions.
test.data<-data.prep(scored.data,c(3,15,15,20),
                     c("Algebra","Geometry","Measurement", "Math"))
#----------------------------------------------
# Estimating subscores using Haberman's methods       
CTTsub(test.data,method="Haberman") # Estimating subscores using Haberman's methods 

# Obtaining original correlation for the three methods
CTTsub(test.data,method="Haberman")$Correlation  

# Obtaining disattenuated correlation for the three methods
CTTsub(test.data,method="Haberman")$Disattenuated.correlation  

# Obtaining PRMSEs for the three methods
CTTsub(test.data,method="Haberman")$PRMSE  

# Obtaining descriptive statistics summary for estimated subscores  
CTTsub(test.data,method="Haberman")$summary 

# Obtaining raw subscores  
CTTsub(test.data,method="Haberman")$subscore.original  

# Obtaining subscores that are estimated as a function of the observed subscores 
CTTsub(test.data,method="Haberman")$subscore.s 

# Obtaining subscores that are estimated as a function of the observed total score 
CTTsub(test.data,method="Haberman")$subscore.x  

# Obtaining subscores that are estimated as a function of 
# both the observed subscores and the observed total score.
CTTsub(test.data,method="Haberman")$subscore.sx  

#-------------------------------------------      
# Estimating subscores using Wainer's method
CTTsub(test.data,method="Wainer") 
       
# Obtaining descriptive statistics summary for subscores
CTTsub(test.data,method="Wainer")$summary   

# Obtaining original subscores
CTTsub(test.data,method="Wainer")$subscore.original 

# Obtaining subscores that are estimated using Wainer's augmentation method  
CTTsub(test.data,method="Wainer")$subscore.augmented  
        

The 2011 TIMSS Grade 8 Mathematics Assessment Dataset

Description

The TIMSS dataset used in Dai, Svetina, and Wang (2017) (doi:10.3102/1076998617716462). It contained responses from 765 students to 32 items with 6 to 9 items on each of the subscales of (1) number (Q1 to Q9), (2) algebra (Q10 to Q18), (3) geometry (Q19 to Q24), and (4) data and chance (Q25 to Q30). Omitted responses were treated as incorrect.

Usage

data("TIMSS11G8M.data")

Format

A data frame with 765 observations on the following 32 variables.

Q1

a numeric vector

Q2

a numeric vector

Q3

a numeric vector

Q4

a numeric vector

Q5

a numeric vector

Q6

a numeric vector

Q7

a numeric vector

Q8

a numeric vector

Q9

a numeric vector

Q10

a numeric vector

Q11

a numeric vector

Q12

a numeric vector

Q13

a numeric vector

Q14

a numeric vector

Q15

a numeric vector

Q16

a numeric vector

Q17

a numeric vector

Q18

a numeric vector

Q19

a numeric vector

Q20

a numeric vector

Q21

a numeric vector

Q22

a numeric vector

Q23

a numeric vector

Q24

a numeric vector

Q25

a numeric vector

Q26

a numeric vector

Q27

a numeric vector

Q28

a numeric vector

Q29

a numeric vector

Q30

a numeric vector

Q31

a numeric vector

Q32

a numeric vector

Source

Dai, S., Svetina, D., & Wang, X. (2017). "Reporting subscores using R: A software review." Journal of Educational and Behavioral Statistics. 42(2), 617-638. doi: 10.3102/1076998617716462.

Examples

data(TIMSS11G8M.data)
# maybe str(TIMSS11G8M.data); plot(TIMSS11G8M.data) ...

Estimating true subscores using Yen's OPI

Description

This function estimates subscores using Yen's Objective Performance Index (OPI; Yen, 1987) <https://www.ets.org/research/policy_research_reports/publications/paper/1987/hrap>. Yen's OPI (Yen, 1987) is a procedure combining Bayesian method and item response theory (IRT; Embretson & Reise, 2000 <https://psycnet.apa.org/record/2000-03918-000>; Reckase, 1997 <doi: 10.1177/0146621697211002>). This method pulls an examinee's performance on a certain objective (i.e., subscale) towards his/her total test performance in order to get a more stable and precise objective subscore estimate.

Usage

Yen.OPI(test.data)

Arguments

Value

References

Embretson, S. E., & Reise, S. P. (2013). "Item response theory". Mahwah, NJ: Lawrence Erlbaum Associates, Inc. https://psycnet.apa.org/record/2000-03918-000.

Reckase, M. D. (1997). "The past and future of multidimensional item response theory". Applied Psychological Measurement, 21(1), 25-36. doi: 10.1177/0146621697211002.

Yen, W. M. (1987, June). "A Bayesian/IRT index of objective performance". Paper presented at annual meeting of the Psychometric Society, Montreal, Quebec, Canada. https://www.ets.org/research/policy_research_reports/publications/paper/1987/hrap.

Examples

 
        test.data<-data.prep(scored.data,c(3,15,15,20),
                             c("Algebra","Geometry","Measurement", "Math"))
        
        Yen.OPI(test.data)

Computing Yen's Q3 statistic for unidimensional Rasch, 1-, 2-, and 3-PL logistic IRT models

Description

This function calculates Yen's Q3 statistics as introduced in Yen (1984) <doi: 10.1177/014662168400800201> and Yen (1993) <doi: 10.1111/j.1745-3984.1993.tb00423.x> for unidimensional Rasch, 1-, 2-, and 3-PL logistic IRT models to assess the local independence assumption.

Usage

Yen.Q3(scored.data, IRT.model = "2pl")

Arguments

Value

References

Yen, W. M. (1984). "Effects of local item dependence on the fit and equating performance of the three-parameter logistic model." Applied Psychological Measurement, 8(2), 125-145. doi: 10.1177/014662168400800201.

Yen, W. M. (1993). "Scaling performance assessments: Strategies for managing local item dependence. " ournal of educational measurement, 30(3), 187-213. doi: 10.1111/j.1745-3984.1993.tb00423.x.

Examples

 
        Yen.Q3(scored.data,IRT.model="2pl")
        
        Yen.Q3(scored.data)$Q3
        Yen.Q3(scored.data)$Q3.weighted

This function prepares data into a required list format

Description

This function generates a list of data sets using the scored original data set, which can be used as objects in subscore computing functions.

Usage

data.prep(scored.data, subtest.infor, subtest.names = NULL)

Arguments

Value

A list that contains item responses of all subtests and the entire test. The list is then used by other functions (e.g., CTTsub) in the package to obtain subscores.

Examples

        subtest.infor<-c(3,15,15,20) 
        subtest.names<-c("Algebra","Geometry","Measurement", "Math")
        # This math test consists of 3 subtests, which have 15 algebra 
        # items, 15 geometry items, and 20 measurement items.
        test.data<-data.prep(scored.data, subtest.infor, subtest.names)

Sample scored data

Description

This dataset contains responses of 150 examinees to three subscales. These subscales consist of 15, 15, and 20 items respectively.

Usage

data("scored.data")

Format

A data frame with 150 observations on the following 50 variables.

V1

Item 1

V2

Item 2

V3

Item 3

V4

Item 4

V5

Item 5

V6

Item 6

V7

Item 7

V8

Item 8

V9

Item 9

V10

Item 10

V11

Item 11

V12

Item 12

V13

Item 13

V14

Item 14

V15

Item 15

V16

Item 16

V17

Item 17

V18

Item 18

V19

Item 19

V20

Item 20

V21

Item 21

V22

Item 22

V23

Item 23

V24

Item 24

V25

Item 25

V26

Item 26

V27

Item 27

V28

Item 28

V29

Item 29

V30

Item 30

V31

Item 31

V32

Item 32

V33

Item 33

V34

Item 34

V35

Item 35

V36

Item 36

V37

Item 37

V38

Item 38

V39

Item 39

V40

Item 40

V41

Item 41

V42

Item 42

V43

Item 43

V44

Item 44

V45

Item 45

V46

Item 46

V47

Item 47

V48

Item 48

V49

Item 49

V50

Item 50

Details

A dataset containing responses of 150 examinees to a total number of 50 items on three subscales (15, 15, and 20 items respectively).

Examples

data(scored.data)
# maybe str(scored.data); plot(scored.data) ...

Estimating true subscores using Wainer's augmentation method

Description

This function estimates subscores using Wainer's augmentation method (Wainer et. al., 2001) <doi:10.4324/9781410604729>. The central idea of this procedure is that, the estimation of subscores will be improved by shrinking the individual observed subscores towards some aggregate values (i.e., group mean subscores). The extent of the shrinkage depends on the closeness of the subscale being estimated with other subscales as well as reliabilities of all the subscales. Wainer's augmentation is a multivariate version of Kelly's formula (Kelly, 1947) <https://www.hup.harvard.edu/catalog.php?isbn=9780674330009>. For details of Wainer's augmentation subscoring method, please refer to Wainer et al. (2001) <doi:10.4324/9781410604729>.

Usage

subscore.Wainer(test.data)

Arguments

Value

References

Wainer, H., Vevea, J., Camacho, F., Reeve, R., Rosa, K., Nelson, L., Swygert, K., & Thissen, D. (2001). "Augmented scores - "Borrowing strength" to compute scores based on small numbers of items" In Thissen, D. & Wainer, H. (Eds.), Test scoring (pp.343 - 387). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. doi:10.4324/9781410604729.

Kelley, T. L. (1947). Fundamentals of statistics. Harvard University Press. https://www.hup.harvard.edu/catalog.php?isbn=9780674330009.

Examples

 
       test.data<-data.prep(scored.data,c(3,15,15,20),
                            c("Algebra","Geometry","Measurement", "Math"))
        
        subscore.Wainer(test.data)
        
        subscore.Wainer(test.data)$summary
        subscore.Wainer(test.data)$subscore.augmented

Computing correlation indices for subscores and the total score.

Description

This function computes Cronbach's Alpha and Stratified Alpha (Cronbach et al., 1965) <doi: 10.1177/001316446502500201>. Disattenuated correlations are also provided.

Usage

subscore.corr(test.data)

Arguments

Value

References

Cronbach, L., Schonenman, P., & McKie, D. (1965). "Alpha coefficients for stratified-parallel tests." Educational and Psychological Measurement, 25, 291-282. doi: 10.1177/001316446502500201.

Examples

# Transferring scored response data to the required list format
test.data<-data.prep(scored.data,c(3,15,15,20),
                     c("Algebra","Geometry","Measurement", "Math"))
  
#Estimate true subscores using Haberman's method based on observed subscores     
subscore.corr(test.data) 
       
subscore.s(test.data)$summary
subscore.s(test.data)$correlation

Computing subscores using Haberman's method based on observed subscores.

Description

This function estimate true subscores based on observed subscores, using the method introduced by Haberman (2008) <doi:10.3102/1076998607302636>.

Usage

subscore.s(test.data)

Arguments

Value

References

Haberman, S. J. (2008). "When can subscores have value?." Journal of Educational and Behavioral Statistics, 33(2), 204-229. doi:10.3102/1076998607302636.

Examples

# Transferring scored response data to the required list format
test.data<-data.prep(scored.data,c(3,15,15,20),
                     c("Algebra","Geometry","Measurement", "Math"))
  
# Estimate true subscores using Haberman's method based on observed subscores     
subscore.s(test.data) 
       
subscore.s(test.data)$summary
subscore.s(test.data)$Correlation
subscore.s(test.data)$Disattenuated.correlation
subscore.s(test.data)$PRMSE
subscore.s(test.data)$subscore.s

Computing subscores using Haberman's method based on both observed total scores and observed subscores.

Description

This function estimate true subscores based on both observed total scores and observed subscores using the method introduced by Haberman (2008) <doi:10.3102/1076998607302636>.

Usage

subscore.sx(test.data)

Arguments

Value

References

Haberman, S. J. (2008). "When can subscores have value?." Journal of Educational and Behavioral Statistics, 33(2), 204-229. doi:10.3102/1076998607302636.

Examples

       test.data<-data.prep(scored.data,c(3,15,15,20),
                            c("Algebra","Geometry","Measurement", "Math"))
       
       subscore.sx(test.data) 
       subscore.s(test.data)$Correlation
       subscore.s(test.data)$Disattenuated.correlation
       subscore.sx(test.data)$summary
       subscore.sx(test.data)$PRMSE
       subscore.sx(test.data)$subscore.sx

Computing subscores using Haberman's method based on observed total scores.

Description

This function estimates true subscores based on observed total scores using the method introduced by Haberman (2008) <doi:10.3102/1076998607302636>.

Usage

subscore.x(test.data)

Arguments

Value

References

Haberman, S. J. (2008). "When can subscores have value?." Journal of Educational and Behavioral Statistics, 33(2), 204-229.doi:10.3102/1076998607302636

Examples

       test.data<-data.prep(scored.data,c(3,15,15,20), 
                            c("Algebra","Geometry","Measurement", "Math"))
       
       subscore.x(test.data) 
       
       subscore.x(test.data)$summary
       subscore.x(test.data)$PRMSE
       subscore.x(test.data)$Correlation
       subscore.x(test.data)$Disattenuated.correlation
       subscore.x(test.data)$subscore.x

A list of objects that include both test information and subscores.

Description

This list consists of four objects. The first three objects are item responses on the three subscales (algebra, geometry, and measurement). The fourth object is the response data on the total test.

Usage

data("test.data")

Format

The format is: A list with 4 objects:

$ Algebra :'data.frame': 150 obs. of 15 variables:

$ Geometry :'data.frame': 150 obs. of 15 variables:

$ Measurement:'data.frame': 150 obs. of 20 variables:

$ Math :'data.frame': 150 obs. of 50 variables:

Details

Algebra: Responses of 150 participants to 15 items; Geometry: Responses of 150 participants to 15 items. Measurement: Responses of 150 participants to 20 items; Math: Responses of 150 participants to 20 items.

Examples

data(test.data)
# maybe str(test.data); plot(test.data) ...