Package 'TestDataImputation'

Title: Missing Item Responses Imputation for Test and Assessment Data
Description: Functions for imputing missing item responses for dichotomous and polytomous test and assessment data. This package enables missing imputation methods that are suitable for test and assessment data, including: listwise (LW) deletion (see De Ayala et al. 2001 <doi:10.1111/j.1745-3984.2001.tb01124.x>), treating as incorrect (IN, see Lord, 1974 <doi: 10.1111/j.1745-3984.1974.tb00996.x>; Mislevy & Wu, 1996 <doi: 10.1002/j.2333-8504.1996.tb01708.x>; Pohl et al., 2014 <doi: 10.1177/0013164413504926>), person mean imputation (PM), item mean imputation (IM), two-way (TW) and response function (RF) imputation, (see Sijtsma & van der Ark, 2003 <doi: 10.1207/s15327906mbr3804_4>), logistic regression (LR) imputation, predictive mean matching (PMM), and expectation–maximization (EM) imputation (see Finch, 2008 <doi: 10.1111/j.1745-3984.2008.00062.x>).
Authors: Shenghai Dai [aut, cre], Xiaolin Wang [aut], Dubravka Svetina [aut]
Maintainer: Shenghai Dai <[email protected]>
License: GPL (>= 2)
Version: 2.3
Built: 2024-11-04 04:19:26 UTC
Source: https://github.com/cran/TestDataImputation

Help Index

EM Imputation

Description

This function imputes for all missing responses using EM imputation (see Finch, 2008) <doi: 10.1111/j.1745-3984.2008.00062.x>. The Amelia package (Honaker et al., 2011 <doi: 10.18637/jss.v045.i07>) is used for the imputation. Integrated scores are then obtained by rounding imputed values to the closest possible response value.

Usage

EMimpute(test.data, Mvalue = "NA", max.score = 1, round.decimal = 0)

Arguments

test.data

Test data set (a data frame or a matrix) containing missing responses. Missing values are coded as NA or other values (e.g., 8, 9).
Mvalue

Missing response indicators in the data (e.g. "NA", "8", "9", etc.). Mvalue="NA" by default.
max.score

The max possible response value in the test data. By default max.score=1 (i.e.,binary test data).
round.decimal

The number of digits or decimal places for the imputed value. The default value is 0.

Value

A data frame with all missing responses replaced by integrated imputed values.

References

Finch, H. (2008). "Estimation of Item Response Theory Parameters in the Presence of Missing Data." Journal of Educational Measurement, 45(3), 225-245. doi: 10.1111/j.1745-3984.2008.00062.x.

Honaker, J., King, G., & Blackwell, M. (2011). "Amelia II: A program for missing data." Journal of statistical software, 45(1), 1-47. doi: 10.18637/jss.v045.i07.

Examples

EMimpute(test.data, Mvalue="NA",max.score=1,round.decimal=0)

This main function imputes for missing responses using selected method

Description

This function imputes for all missing responses using the selected imputation method. Integrated scores are obtained by rounding imputed values to the closest possible response value.

Usage

ImputeTestData(
  test.data,
  Mvalue = "NA",
  max.score = 1,
  method = "LW",
  round.decimal = 0
)

Arguments

test.data

Test data set (a data frame or a matrix) containing missing responses. Missing values are coded as NA or other values (e.g., 8, 9).
Mvalue

Missing response indicators in the data (e.g. "NA", "8", "9", etc.). Mvalue="NA" by default.
max.score

The max possible response value in test data. By default max.score=1 (i.e.,binary test data). max.score = 2 if the response categories are (0, 1, 2), etc. Note: For IN and RF, the lowest response value should be zero (i.e., incorrect).
method

Missing response imputation methods.
"LW" (by default) represents listwise that deletes all examinees who reported missing responses (see De Ayala et al. 2001 <doi:10.1111/j.1745-3984.2001.tb01124.x>)
"IN" means treating all missing responses as incorrect (see Lord, 1974 <doi: 10.1111/j.1745-3984.1974.tb00996.x>; Mislevy & Wu, 1996 <doi: 10.1002/j.2333-8504.1996.tb01708.x>; Pohl et al., 2014 <doi: 10.1177/0013164413504926>).
"PM" imputes for all missing responses of an examinee by his/her mean on the available items.
"IM" imputes for all missing responses of an item by its mean on the available responses.
"TW" imputes for all missing responses using two-way imputation (if an examinee has no response to all items, the missing responses are replaced by item means first; see Sijtsma & van der Ark, 2003 <doi: 10.1207/s15327906mbr3804_4>). "RF" imputes for all missing responses using response function imputation (Sijtsma & van der Ark, 2003 <doi: 10.1207/s15327906mbr3804_4>).
"LR" imputes for all missing responses using logistic regression (for binary responses) and polytomous regression (for polytmous responses) with mice package (Van Buuren & Groothuis-Oudshoorn, 2011 <doi: 10.18637/jss.v045.i03>).
"PMM" imputes for all missing responses using predictive mean matching with mice package (Van Buuren & Groothuis-Oudshoorn, 2011 <doi: 10.18637/jss.v045.i03>).
"EM" imputes for all missing responses using EM imputation with the Amelia package (Honaker et al., 2011 <doi: 10.18637/jss.v045.i07>). The imputed values are then rounded to the closest possible response value. (see Finch, 2008 <doi: 10.1111/j.1745-3984.2008.00062.x>).
round.decimal

The number of digits or decimal places for the imputed value. The default value is 0.

Value

A data frame with all missing responses replaced by integrated imputed values.

References

De Ayala, R. J., Plake, B. S., & Impara, J. C. (2001). "The impact of omitted responses on the accuracy of ability estimation in item response theory." Journal of Educational Measurement, 38(3), 213-234. doi:10.1111/j.1745-3984.2001.tb01124.x.

Finch, H. (2008). "Estimation of Item Response Theory Parameters in the Presence of Missing Data." Journal of Educational Measurement, 45(3), 225-245. doi: 10.1111/j.1745-3984.2008.00062.x.

Honaker, J., King, G., & Blackwell, M. (2011). "Amelia II: A program for missing data." Journal of statistical software, 45(1), 1-47. doi: 10.18637/jss.v045.i07.

Lord, F. M. (1974). " Quick estimates of the relative efficiency of two tests as a function of ability level." Journal of Educational Measurement, 11(4), 247-254. doi: 10.1111/j.1745-3984.1974.tb00996.x.

Mislevy, R. J., & Wu, P. K. (1996). " Missing responses and IRT ability estimation: Omits, choice, time limits, and adaptive testing. " ETS Research Report Series, 1996(2), i-36. doi: 10.1002/j.2333-8504.1996.tb01708.x.

Pohl, S., Gräfe, L., & Rose, N. (2014). "Dealing with omitted and not-reached items in competence tests evaluating approaches accounting for missing responses in item response theory models. " Educational and Psychological Measurement, 74(3), 423-452. doi: 10.1177/0013164413504926.

Sijtsma, K., & Van der Ark, L. A. (2003). "Investigation and treatment of missing item scores in test and questionnaire data." Multivariate Behavioral Research, 38(4), 505-528. doi: 10.1207/s15327906mbr3804_4.

Van Buuren, S., & Groothuis-Oudshoorn, K. (2011). "mice: Multivariate imputation by chained equations in R." Journal of statistical software, 45(1), 1-67. DOI: 10.18637/jss.v045.i03.

Examples

ImputeTestData(test.data, Mvalue="NA",max.score=1, method ="TW",round.decimal=0)

Item Mean (IM) Imputation

Description

This function imputes for all missing responses of an item by its mean (i.e., IM) on the available responses. Integrated scores for items are obtained by rounding their means to the closest possible response value.

Usage

ItemMean(test.data, Mvalue = "NA", max.score = 1, round.decimal = 0)

Arguments

test.data

Test data set (a data frame or a matrix) containing missing responses. Missing values are coded as NA or other values (e.g., 8, 9).
Mvalue

Missing response indicators in the data (e.g. "NA", "8", "9", etc.). Mvalue="NA" by default.
max.score

The max possible response value in test data. By default max.score=1 (i.e.,binary test data).
round.decimal

The number of digits or decimal places for the imputed value. The default value is 0.

Value

A data frame with all missing responses replaced by Integrated item means.

Examples

ItemMean(test.data, Mvalue="NA",max.score=1,round.decimal=0)

Listwise Deletion (LW)

Description

This function deletes examinees who report missing responses (see De Ayala et al. 2001) <doi:10.1111/j.1745-3984.2001.tb01124.x>.

Usage

Listwise(test.data, Mvalue = "NA")

Arguments

test.data

Test data set (a data frame or a matrix) containing missing responses. Missing values are coded as NA or other values (e.g., 8, 9).
Mvalue

Missing response indicators in the data (e.g. "NA", "8", "9", etc.). Mvalue="NA" by default.

Value

A data frame with no missing responses.

References

De Ayala, R. J., Plake, B. S., & Impara, J. C. (2001). "The impact of omitted responses on the accuracy of ability estimation in item response theory." Journal of Educational Measurement, 38(3), 213–234. doi:10.1111/j.1745-3984.2001.tb01124.x.

Examples

Listwise(test.data, Mvalue="NA")

Logistic Regression (LR) Imputation

Description

This function imputes for all missing responses using logistic regression (for binary responses) or polytomous regression (for polytomous responses). The mice () function with default settings from the mice package (Van Buuren & Groothuis-Oudshoorn, 2011 <doi: 10.18637/jss.v045.i03>) is used to impute for the missing responses.

Usage

LogisticReg(test.data, Mvalue = "NA", max.score = 1)

Arguments

test.data

Test data set (a data frame or a matrix) containing missing responses. Missing values are coded as NA or other values (e.g., 8, 9).
Mvalue

Missing response indicators in the data (e.g. "NA", "8", "9", etc.). Mvalue="NA" by default.
max.score

The max possible response value in test data. By default max.score=1 (i.e.,binary test data).

Value

A data frame with all missing responses replaced by integrated imputed values.

References

Van Buuren, S., & Groothuis-Oudshoorn, K. (2011). "mice: Multivariate imputation by chained equations in R." Journal of statistical software, 45(1), 1-67. DOI: 10.18637/jss.v045.i03.

Examples

LogisticReg(test.data, Mvalue="NA",max.score=1)

Predictive mean matching (PMM)

Description

This function imputes for all missing responses using predictive mean matching. The mice () function with default settings from the mice package (Van Buuren & Groothuis-Oudshoorn, 2011 <doi: 10.18637/jss.v045.i03>) is used to impute for the missing responses.

Usage

micePMM(test.data, Mvalue = "NA")

Arguments

test.data

Test data set (a data frame or a matrix) containing missing responses. Missing values are coded as NA or other values (e.g., 8, 9).
Mvalue

Missing response indicators in the data (e.g. "NA", "8", "9", etc.). Mvalue="NA" by default.

Value

A data frame with all missing responses replaced by integrated imputed values.

References

Van Buuren, S., & Groothuis-Oudshoorn, K. (2011). "mice: Multivariate imputation by chained equations in R." Journal of statistical software, 45(1), 1-67. DOI: 10.18637/jss.v045.i03.

Examples

micePMM(test.data, Mvalue="NA")

Person Mean Imputation (PM)

Description

This function imputes for all missing responses of an examinee by his/her mean (i.e., PM) on the available items. Integrated scores for examinees are obtained by rounding their means to the closest possible response value.

Usage

PersonMean(test.data, Mvalue = "NA", max.score = 1, round.decimal = 0)

Arguments

test.data

Test data set (a data frame or a matrix) containing missing responses. Missing values are coded as NA or other values (e.g., 8, 9).
Mvalue

Missing response indicators in the data (e.g. "NA", "8", "9", etc.). Mvalue="NA" by default.
max.score

The max possible response value in test data. By default max.score=1 (i.e.,binary test data).
round.decimal

The number of digits or decimal places for the imputed value. The default value is 0.

Value

A data frame with all missing responses replaced by person means.

References

Sijtsma, K., & Van der Ark, L. A. (2003). "Investigation and treatment of missing item scores in test and questionnaire data." Multivariate Behavioral Research, 38(4), 505-528.DOI: 10.1207/s15327906mbr3804_4.

Examples

PersonMean(test.data, Mvalue="NA",max.score=1,round.decimal=0)

Response Function Imputation (RF)

Description

This function imputes for all missing responses using the response function imputation (Sijtsma and van der Ark, 2003 <doi: 10.1207/s15327906mbr3804_4>).

Usage

ResponseFun(test.data, Mvalue = "NA", max.score = 1, round.decimal = 0)

Arguments

test.data

Test data set (a data frame or a matrix) containing missing responses. Missing values are coded as NA or other values (e.g., 8, 9).#'
Mvalue

Missing response indicators in the data (e.g. "NA", "8", "9", etc.). Mvalue="NA" by default.
max.score

The max possible response value in test data (the lowest response value should be 0). By default max.score=1 (i.e.,binary test data). max.score = 2 if the response categories are (0, 1, 2), etc.
round.decimal

The number of digits or decimal places for the imputed value. The default value is 0.

Value

A data frame with all missing responses imputed with response function.

References

Sijtsma, K., & Van der Ark, L. A. (2003). "Investigation and treatment of missing item scores in test and questionnaire data." Multivariate Behavioral Research, 38(4), 505-528. DOI: 10.1207/s15327906mbr3804_4.

Examples

ResponseFun(test.data, Mvalue="NA",max.score=1,round.decimal=0)

Example test data

Description

This dataset contains binary responses of 775 participants to 20 items.Missing responses are coded as NA.

Usage

data("test.data")

Format

A data frame with 775 observations on the following 20 items.

Item_1

a numeric vector

Item_2

a numeric vector

Item_3

a numeric vector

Item_4

a numeric vector

Item_5

a numeric vector

Item_6

a numeric vector

Item_7

a numeric vector

Item_8

a numeric vector

Item_9

a numeric vector

Item_10

a numeric vector

Item_11

a numeric vector

Item_12

a numeric vector

Item_13

a numeric vector

Item_14

a numeric vector

Item_15

a numeric vector

Item_16

a numeric vector

Item_17

a numeric vector

Item_18

a numeric vector

Item_19

a numeric vector

Item_20

a numeric vector

Details

A test data that contain binary responses of 775 participants to 20 items.Missing responses are coded as NA.

Examples

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

Treat missing responses as incorrect (IN)

Description

This function replaces all missing responses by zero (see Lord, 1974 <doi: 10.1111/j.1745-3984.1974.tb00996.x>; Mislevy & Wu, 1996 <doi: 10.1002/j.2333-8504.1996.tb01708.x>; Pohl et al., 2014 <doi: 10.1177/0013164413504926>);).

Usage

TreatIncorrect(test.data, Mvalue = "NA")

Arguments

test.data

Test data set (a data frame or a matrix) containing missing responses. Missing values are coded as NA or other values (e.g., 8, 9).
Mvalue

Missing response indicators in the data (e.g. "NA", "8", "9", etc.). Mvalue="NA" by default.

Value

A data frame with all missing responses replaced by '0'.

References

Lord, F. M. (1974). " Quick estimates of the relative efficiency of two tests as a function of ability level." Journal of Educational Measurement, 11(4), 247-254. doi: 10.1111/j.1745-3984.1974.tb00996.x.

Mislevy, R. J., & Wu, P. K. (1996). " Missing responses and IRT ability estimation: Omits, choice, time limits, and adaptive testing. " ETS Research Report Series, 1996(2), i-36. doi: 10.1002/j.2333-8504.1996.tb01708.x.

Pohl, S., Gräfe, L., & Rose, N. (2014). "Dealing with omitted and not-reached items in competence tests evaluating approaches accounting for missing responses in item response theory models. " Educational and Psychological Measurement, 74(3), 423-452. doi: 10.1177/0013164413504926.

Examples

TreatIncorrect(test.data, Mvalue="NA")

Two-Way Imputation (TW)

Description

This function imputes for all missing responses using two-way imputation. Integrated responses are obtained by rounding imputed values to the closest possible response value. If a case showed missingness on all the variables (i.e., empty record), the missing values are replaced by item means first. see Sijtsma and van der Ark (2003) <doi: 10.1207/s15327906mbr3804_4>;

Usage

Twoway(test.data, Mvalue = "NA", max.score = 1, round.decimal = 0)

Arguments

test.data

Test data set (a data frame or a matrix) containing missing responses. Missing values are coded as NA or other values (e.g., 8, 9).#'
Mvalue

Missing response indicators in the data (e.g. "NA", "8", "9", etc.). Mvalue="NA" by default.
max.score

The max possible response value in test data. By default max.score=1 (i.e.,binary test data).
round.decimal

The number of digits or decimal places for the imputed value. The default value is 0.

Value

A data frame with all missing responses replaced by integrated two-way imputed values.

References

Bernaards, C. A., & Sijtsma, K. (2000). " Influence of imputation and EM methods on factor analysis when item nonresponse in questionnaire data is nonignorable." Multivariate Behavioral Research, 35(3), 321-364.DOI: 10.1207/S15327906MBR3503_03.

Examples

Twoway(test.data, Mvalue="NA",max.score=1,round.decimal=0)