deltaPlotR (version 1.2)

deltaPlot: Delta Plot method for dichotomous DIF

Description

This command computes the Delta plot statistics for dichotomous differential item functioning, with all associated output (Delta points, perpendicular distances). The modified Delta plot is also available, as well as several item purification techniques.

Usage

deltaPlot(data, type="response", group, focal.name, thr="norm", purify=FALSE, 
 purType="IPP1", maxIter=10, alpha=0.05,extreme="constraint", 
 const.range = c(0.001, 0.999), nrAdd = 1, save.output=FALSE,
 output=c("out","default"))
## S3 method for class 'deltaPlot':
print(x, only.final=TRUE, ...)

Arguments

data

numeric: either (a) the data matrix with item responses and group membership, (b) the two-column matrix of proportions of correct responses per item and per group, or (c) the two-column matrix of Delta scores. See Details

type

character: the type of data argument. Possible values are "response" (default), "prop" and "delta". See Details.

group

integer or character: a single value for locating the group membership column in the data argument. Ignored if type is not "response". See Details.

focal.name

numeric or character: the value used in the group membership column to refer to the focal group. Ignored if type is not "response". See Details.

thr

numeric or character: the threshold for flagging items as DIF. Can be a positive numeric value or "norm". See Details.

purify

logical: should item purification be performed? (Default is code{FALSE}). See Details.

purType

character: the type of purification process to be run. Possible values are "IPP1" (default), "IPP2" and "IPP3". Ignored if purify is FALSE. See Details.

maxIter

integer: the maximum number of iteration in the purification process (default is 10). Ignored if purify is FALSE.

alpha

numeric: the significance level for calculating the detection threshold (default is 0.05). Ignored if thr is numeric.

extreme

character: the method used to modify the extreme proportions. Possible values are "constraint" (default) or "add". See Details.

const.range

numeric: a vector of two constraining proportions. Default values are 0.001 and 0.999. Ignored if extreme is "add".

nrAdd

integer: the number of successes and the number of failures to add to the data in order to adjust the proportions. Default value is 1. Ignored if extreme is "constraint".

save.output

logical: should the output be saved into a text file? (Default is FALSE).

output

character: a vector of two components. The first component is the name of the output file ("out" by default), the second component is either the file path or "default" (default value). See Deta

x

an object of class "deltaPlot", typically the output of the deltaPlot function.

only.final

logical: should only the first and last steps of the purification process be printed? (default is TRUE. If FALSE all perpendicular distances, parameters of the major axis, and detection t

...

other generic parameters for the plot or the print functions.

Value

• A list of class "deltaPlot" with the following arguments:
• Props
  the matrix of proportions of correct responses, or NA if type is "delta".
• adjProps
  the restricted proportions, in the same format as the output Props matrix, or NA if type is "delta".
• Deltas
  the matrix of Delta scores.
• Dist
  a matrix with perpendicular distances, one row per item and one column per run of the Delta plot. If purify is FALSE, only a single column is returned.
• axis.par
  a matrix with two columns, holding respectively the intercepts and the slope parameters of the major axis. Each row refers to one step of the purification process. If purify is FALSE, only a single row is returned.
• nrIter
  the number of iterations invloved in the purification process. Returned only if purify is TRUE.
• maxIter
  the value of the maxIter argument. Returned only if purify is TRUE.
• convergence
  a logical value indicating whether convergence was reached in the purification process. Returned only if purify is TRUE.
• difPur
  a matrix with one column per item and one row per iteration in the purification process, holding zeros and ones to indicate which items were flagged as DIF or not at each step of the process. Returned only if purify is TRUE.
• thr
  a vector of successive threshold values used during the purification process. If purify is FALSE, a single value is returned.
• rule
  a character value indicating whether the threshold was "fixed" by the user (i.e. by setting thr to a numeric value) or whether it was computed by normal approximation (i.e. by setting thr to "norm").
• purType
  the value of the purType argument. Returned only if purify is TRUE.
• DIFitems
  either "No DIF item detected" or an integer vector with the items that were flagged as DIF.
• adjust.extreme
  the value of the extreme argument.
• const.range
  the value of the const.range argument.
• nrAdd
  the value of the nrAdd argument.
• purify
  the value of the purify argument.
• alpha
  the value of the alpha argument.
• save.output
  the value of the save.output argument.
• output
  the value of the output argument.

Details

Angoff's Delta plot (Angoff and Ford, 1973) is a straightforward test-score method to detect DIF among dichotomously scored items. Proportions of correct responses are computed first per item and per group of respondents, and are successively transformed onto z-scores and then onto $\Delta$ scores. The pairs of $\Delta$ scores can be displayed onto a scatter plot, called the Delta plot, and the majr axis of the ellipsoid of Delta points is derived. Eventually, items whose perpendicular distance (from the major axis) is too large are flagged as DIF. See Angoff and Ford (1973) for further details. The data must be passed through the argument data and can be of three types. Each type is defined by the type argument and can take three values: "response", "prop" and "delta".

• Iftypeis"response", the inputdataconsist in a matrix with one row per respondent and$J+1$columns, where$J$is the number of items. In the colmuns coding for the items, only possible entries are 0 (for incorrect responses), 1 (for corect responses) andNA(for missing values). The extra column is used to define group membership: all respondents of the reference group take the same value (either numeric or character), and all respondents in the focal group take the same (numeric or character) value but different from the reference group. Note that the group membership column can be located anywhere in the data set (not especially in first or last position).
• Iftypeis"prop", the inputdataconsist in a two-column matrix with one row per item. Each row contains the proportions of correct responses, respectively in the reference group (first column) and in the focal group (second column).
• Iftypeis"delta", the inputdataconsist in a two-column matrix that is similar to that provided with the"prop"type of input, but with the Delta scores provided instead of the proportions of correct responses.

If the type of input is either "prop" or "delta", not anymore input information is required and the arguments group and focal.bname are ignored. Otherwise, the group membership column in the data matrix is specified by giving to argument group either the column number (1 for first column, etc.) or the column name (provided the data matrix has argument names). Moreover, the focal group is specified by giving to the argument focal.name the value that was used in the group membership column to code for the focal group. If the input type is not "delta", then extreme proportions of correct responses (either provided when type is "prop" or computed from the data if type is "response") are adjusted by specifying the arguments extreme, const.range and nrAdd with appropriate values. See the adjustExtreme function for further details (note that the cuyrrent extreme argument corresponds to the method argument in this function). The threshold for flaging items as DIF can be of two types and is specified by the thr argument.

1. It can be fixed to some arbitrary positive value by the user, for instance 1.5 (Angoff and Ford, 1973). In this case,thrtakes the required numeric threshold value.
2. Alternatively, it can be derived from the bivariate normal approximation of the Delta points (Magis and Facon, 2012). In this case,thrmust be given the character value"norm"(which is the default value). This threshold equals$$\Phi^{-1}(1-\alpha/2) \; \sqrt{\frac{b^2\,{s_0}^2-2\,b\,s_{01}+{s_1}^2}{b^2+1}}$$where$\Phi$is the density of the standard normal distribution,$\alpha$is the significance level (set by the argumentalphawith default value 0.05),$b$is the slope parameter of the major axis,$s_0$and$s_1$are the sample standard deviations of the Delta scores in the reference group and the focal group, respecively, and$s_{01}$is the sample covariance of the Delta scores (see Magis and Facon, 2012, for further details).

Item purification can be performed by setting the argument purify to TRUE (by default it is FALSE so no purification is performed). The item purification process (IPP) starts when at least one item was flagged as DIF after the first run of the Delta plot, and proceeds as follows.

1. The intercept and slope parameters of the major axis are re-calculated by removing all DIF that are currently flagged as DIF. This yields updated values$a^*$,$b^*$,$s_0^*$,$s_1^*$and$s_{01}^*$of the intercept and slope parameters, sample stanbdard deviations and sample covariance of the Delta scores.
2. Perpendicular distances (for all items) are updated with respect to the updated major axis.
3. Detection threshold is also updated. Three possible updates are possible: see below.
4. All items are now tested for the presence of DIF, given the updated perpendicular distances and major axis.
5. If the set of items flagged as DIF is the same as the one from the previous loop, stop the process. Otherwise go back to step 1.

Unlike traditional DIF methods, the detection threshold may also be updated since it depends on the sample estimates (when the normal approximation is considered). Three approaches are currently implemented and are specified by the purType argument.

1. Method 1 (purType=="IPP1"): the same threshold is used throughout the purification process, it is not iteratively updated. The threshold is the one obtained after the first run of the Delta plot.
2. Method 2 (purType=="IPP2"): only the slope parameter is updated in the threshold formula. By this way, one keeps the full data structure (i.e. neither the sample variances nor the sample covariance of the Delta scores are modified) but only the slope parameter is adjusted to lessen the impact of DIF items.
3. Method 3 (purType=="IPP3"): all adjusted parameters are plugged in the threshold formula. This approach completely discards the effect of items flagged as DIF from the computation of the threshold.

See Magis and Facon (2013) for further details. Note that purification can also be performed with fixed threshold (i.e. specified by the user), but then only IPP1 process is performed. In order to avoid possible infinite loops in the purification process, a maximal number of iterations must be specified through the argument maxIter. The default maximal number of iterations is 10. The output contains all input information, the Delta scores and perpendicular distances, the parameter of the major axis and the items flagged as DIF (if none, a character sentence is returned). In addition, the detection threshold and the type of threshold (fixed or normal approximation) is provided. If item purification was run, several additional elements are returned: the number of iterations, a logical indicator whether the convergence was reached (or not, meaning that the process stopped because of reaching the maximal number of allowed iterations), a matrix with indicators of which items were flagged as DIF at each iteration, and the type of item purification process. Moreover, perpendicular distances are returned in a matrix format (one column per iteration), as well as successive major axis parameters (one row per iteration) and successive thresholds (as a vector). The output is managed and printed in a more user-friendly way. When item purification is performed, only the first and last steps are displayed. Specifying the argument only.final to FALSE prints in addition all intermediate steps of the process (successive perpendicular distances, parameters of the major axis, and detection thresholds). The output can be saved into na text file by specifying the argument save.output to TRUE (by default the output is not captured). If so, the argument output can be specified as a vector of two character values. The first one gives the desired name of the text file, and the second one specifies the directory where the file will be saved (full path is required). By default, the output will be saved in the current working directory as "out.txt" file.

References

Angoff, W. H. and Ford, S. F. (1973). Item-race interaction on a test of scholastic aptitude. Journal of Educational Measurement, 10, 95-106. Magis, D., and Facon, B. (2012). Angoff's Delta method revisited: improving the DIF detection under small samples. British Journal of Mathematical and Statistical Psychology, 65, 302-321. Magis, D., and Facon, B. (2013). Item purification does not always improve DIF detection: a counter-example with Angoff's Delta plot. Educational and Psychological Measurement, 73, 293-311.

See Also

adjustExtreme

Examples

# Loading of the verbal data
 data(verbal)
 attach(verbal)

 # Excluding the "Anger" variable
 verbal <- verbal[colnames(verbal)!="Anger"]

 # Basic Delta plot, threshold 1.5, no item purification
 res <- deltaPlot(data=verbal, type="response", group=25, focal.name=1, purify=FALSE, 
                  thr=1.5)

 # Equivalent writing
 res <- deltaPlot(data=verbal, type="response", group="Gender", focal.name=1, 
                  purify=FALSE, thr=1.5)

 # Using proportions of correct responses as input
 dataRef <- verbal[verbal[,25]==0,1:24]
 dataFoc <- verbal[verbal[,25]==1,1:24]
 p0 <- colMeans(dataRef)
 p1 <- colMeans(dataFoc)
 res.1 <- deltaPlot(data=cbind(p0,p1), type="prop", purify=FALSE, thr=1.5)

 # Using Delta values as input
 Delta <- 4*qnorm(1-cbind(p0,p1))+13
 res.2 <- deltaPlot(data=Delta, type="delta", purify=FALSE, thr=1.5)

 # 'norm' threshold
 res <- deltaPlot(data=verbal, type="response", group="Gender", focal.name=1, 
                  purify=FALSE, thr="norm")

 # Keeping the first 10 items to exhibit DIF
 data <- verbal[,c(1:10,25)]
 deltaPlot(data=data, type="response", group=11, focal.name=1, purify=FALSE, thr="norm")
   # Item 8 is flagged as DIF

 # Item purification with the three processes
 res0 <- deltaPlot(data=data, type="response", group=11, focal.name=1, purify=TRUE, 
           thr=1.5, purType="IPP1") 
 res0 # No DIF item detected

 res1 <- deltaPlot(data=data, type="response", group=11, focal.name=1, purify=TRUE, 
           thr="norm", purType="IPP1") 
 res1 # Item 8 flagged as DIF after 2 iterations

 res2 <- deltaPlot(data=data, type="response", group=11, focal.name=1, purify=TRUE, 
           thr="norm", purType="IPP2") 
 res2 # Item 8 flagged as DIF after 2 iterations

 res3 <- deltaPlot(data=data, type="response", group=11, focal.name=1, purify=TRUE, 
           thr="norm", purType="IPP3") 
 res3 # Items 6, 7 and 8 flagged as DIF after 4 iterations

 # Printing the full results of item purification
 print(res, only.final=FALSE)
 print(res0, only.final=FALSE)
 print(res1, only.final=FALSE)
 print(res2, only.final=FALSE)
 print(res3, only.final=FALSE)