- In the IRT training section, the calculation and display of item information curves was fixed. (Thanks to Hynek Cigler for reporting this issue.)
- Calculation of minimal scores in ItemAnalysis() function was fixed.
- Display of ICC in IRT models is being updated to respond to updates in mirt package.
- In the Regression section of the app, missplaced SE for intercepts and slopes were fixed.
- New function
plot_corr()to compute and visualize an item correlation matrix was added. and implemented within the ShinyItemAnalysis interactive app. - New datasets
HCIgrads,HCItestretest,HCIprepostandHCIdatawere added. startShinyItemAnalysis()function was fixed to finish job persisting in the backgroud . after the app is closed.- Figures can now be downloaded at once in DDF section of the app.
- Equations were unified in Regression, IRT and DIF sections of the app, with p being the index for person and i being the index for an item.
- Training section was added to DIF section.
- Figures in Regression section are now displayed in plotly mode in the shiny app.
- In IRT/Rasch section, the scaterplots for standardized total score vs. factor score now displayed in plotly mode in the shiny app. (Needs to be applied to other IRT subsections.)
- IRT/Rasch section was restructured, and it now offers both the intercept/slope as well as the IRT parametrization. (Needs to be applied to other IRT subsections.)
- Regression/Multinomial newly offers also the IRT parametrization.
- Selected R code was updated.
- Unnecessary dependencies were removed.
- Summary tab was updated within the app, interactive help with formulae is newly offered.
- Lines for interpretation of IIC for selected theta were added in IRT training section.
- styler package was applied on .R files, documentation was updated.
- ggWright() plot was updated to more general description.
- References were updated.
- Display of Cronbach's alpha in PDF/HTML reports was corrected.
- Item slider bug in ordinal regression models in the app was fixed.
- DIF matching criterion was synchronized for Summary and Item subtabs in the Generalized logistic regression method, DIF/Fairness section of the app.
- Bug in display of table for Regression / Model comparison was fixed.
startShinyItemAnalysis()function was updated to allow starting the application as a background job in Rstudio (default setting). This permits the user to work and try sample R code examples in the console.- Downloadable tables are newly provided in shiny app in DIF sections for ordinal DIF.
- DDplot is now displayed in plotly mode in the shiny app.
- New function
recode_nr()for recognizing and recoding not-reached responses was introduced. ItemAnalysis()function was updated with new columns for ratio of missed and unreached items, shortened column names and ordering.- Traditional item analysis table in Item analysis section of the app has been expanded to include item reliability, item validity, and percentage of missing and unreached for each item.
DDplot()function now offers argumentcriterion. Withoutcriterionspecification, difficulty-discrimination plot is provided. Whencriterionis specified, difficulty-validity plot is provided instead, depicting item criterion validity.- Difficulty-validity plot is newly offered in section Validity/Criterion_Validity/Items
- Documentation of
DDplot()was updated. - All datatable outputs now have proper bootstrap style/theme.
plotDIFLogistic()now correctly plots matching criterion when item purification is applied.
- Use of DIF matching criterion other than Total score is now enabled in shiny app in DIF sections.
- Downloadable tables are newly provided in shiny app in DIF sections.
- Iterations of purification process are now displayed in downloadable tables in the DIF sections of shiny app.
- DIF method comparison table is provided in DIF section of the shiny app.
- Help page for the
ShinyItemAnalysispackage was added. - NEWS file is now formatted using
markdown. - On attach message was updated.
- Wording, Figures and page structure were updated in shiny app in DIF sections, histogram of total scores now better compares the groups.
THIS IS A CRAN VERSION
It includes versions 1.3.1-1 - 1.3.1-8 and following changes:
- Mismatch in specification of H0 and H1 was corrected in
DIF/Fairnesssection and inReportsection. plotDIFLogistic()now correctly plots matching criterion.- Logistic method in
DIF/Fairnesssection now correctly plots best fitting model when p.adjust.method is applied. - Logistic method in
DIF/Fairnesssection now excludes missing values.
- New dataset
LearningToLearnwas added to the package. - New toy datasets
LearningToLearn6andLearningToLearn9were added to the app. - Upload of the DIF matching criterion is now available in Data section of the app.
- D-DIF index calculation was added into DIF/MH/Item to evaluate DIF effect size
- OR calculation in DIF/MH was edited.
plotDIFLogistic()function was updated to accept class ofLogistikas input.plotDIFLogistic()now accepts character input foritemargument.- Summary and Item inputs in DIF/Logistic are connected now. Model is now fitted only once.
- Citation and list of References were updated.
- Authors' details were updated.
- Due to changes in
difNLR, functionddfORD()was renamed todifORD(). - About page was updated.
- Error messages for
DistractorAnalysis()andplotDistractorAnalysis()were unified. - Copyright year in footer is now responsive.
- Description of toy datasets was updated.
- styler package was used for all functions. Documentation was updated.
- For
plotDIFLogistic()it is now possible to turn off plotting empirical probabilities. This can be done usingdraw.empirical = FALSE. - Selected R code was updated in DIF/MH
- Threshold for discrimination can be now specified in DDplot()
function via argument
thr. Default value is 0.2. - Warning and error messages were updated for
plotDistractorAnalysis()function whenkeyis missing. - Width of inputs was unified in
DIF/Fairnesssection. - Wording was updated.
- Some typos were fixed.
- Text in reports was updated.
- Some examples were set to run to allow for automatic checks of the package.
- Typos in
plotAdjacent(),plotCumulative()andplotMultinomial()were fixed.
- Functions
DistractorAnalysis()andplotDistractorAnalysis()were updated to not requiringkeyinput.
- Bug in key uploading was fixed. When nominal data was uploaded and then ordinal (without cut-score specified in file), key was kept from nominal data.
- Bug in ordinal data was fixed. Recalculation of factors to numeric is done.
- Some calculations with
NAvalues were fixed. - Double file input
"key"was unified.
- Item names for toy dataset
Sciencewere updated. - Basic summary table for ordinal data was updated.
- Table for MH in PDF reports was fixed.
- Header of sections were fixed in PDF reports.
- IRT parameter estimates table was fixed in PDF reports.
- Bug in
plotMultinomial()for data without NAs was fixed.
- DIF/DDF detection methods for ordinal data were added. DIF/Fairness section was updated.
- Tabs in Regression were renamed.
- Warning when group vector is missing was added into Method comparison tab
in
DIF/Fairnesssection. - Method comparison tab for
DIF/Fairnesssection was updated.
- In DIF section, Summary table for total scores now displays correct number of respondents in groups (N).
- Bug in Adjacent model in Regression tab was fixed. Matching criterion and parametrization inputs now have effect on model.
- Bug in
plotMultinomial()was fixed. Function now displays more than 12 response patterns as linetype.
- Range for x-axes in
plotAdjacent()andplotCumulative()were updated to avoid removing data. - Some typos were fixed.
- Cronbach's alpha with CI was added into reports.
- Bonferroni correction for multiple comparison was added as an option for correction methods in Regression, DIF and Reports tabs.
- Warning for table for parameters in Bock's IRT model was modified.
- Wright map now considers item names when option Keep item names.
- Some wording was updated.
- Some typos were fixed.
- Minimal number of groups for gULI in DDplot was set to 2 in both Item analysis and Report tabs.
- Mantel-Haenszel test was added into reports.
- Error for DIF/Logistic was customized.
- Text in reports in DIF section was updated.
- Tables in DIF section now includes both p-values and adjusted p-values when correction method is applied.
- Warning for fancyhdr latex package was solved.
THIS IS A CRAN VERSION
It includes versions 1.3.0-1 - 1.3.0-18 and following changes:
- In
ItemAnalysis()function, documentation for item reliability and item validity indices has been added (Reported by Lot Fonteyne).
- Bug in
plotDistractorAnalysis()was fixed. X-ticks are now displayed even with discrete = T argument.
- Inputs in Validity/Correlation overlaped when window was narrow. Fixed.
- Generate and Download buttons overlaped in Reports when window was narrow. Fixed.
DistractorAnalysis()andplotDistractorAnalysis()now handle discrete matching and groups specified via argument cut.points.- Distractor plots in Validity/Items tab now properly handle discrete criterion and the number of groups is correctly descreased when criterion does not have unique cuts.
- Bug in Raju in
DIF/Fairnesstab was fixed. Plots for items are now displayed for dataBov dataset.
- Checking of unique cuts for distractor plots was added into Reports. Warning is displayed when cuts are not unique and slider is updated.
- Some typos were fixed.
- Bug in Bock's model for binary items was fixed in reports.
- Histogram when some levels of total scores are missing were fixed in Summary and DIF tab.
- Legend was added into IRT plots when downloaded.
- IRT models now respect ordering of items.
- Some typos in IRT models tab were fixed.
- DIF summary table bug was fixed. Table is disaplayed when some DIF methods cannot be fitted.
- DDF parameter tables were fixed for data BOV and Czech Trivia.
- With
difNLRpackage version 1.2.3 plots for DDF are displayed even for binary data. Plots were also fixed for data Bov. - PDF Reports now work locally and on shinyapps.io.
- Dendrogram now properly replaces item names when Keep item names is not selected in Data upload section.
- Authors' and contributors' details were updated.
- Function
plotMultinomial()was updated to return smooth curves for any matching criterion.
- Bug in Bock's nominal IRT model was fixed. ICCs are now properly displayed for all binary data. Error for non-displayable table of parameters was further improved. Tables for items are displayed even in case that table for all items is not.
- DIF items are now correctly written in reports.
- Corrplot now properly replaces item names when Keep item names is not selected in Data upload section.
- In
DIF/Fairnesssection, histograms of Total scores for groups are now properly displayed also for ordinal items.
- Legends for multinomial and distractor plots were improved.
- Authors' and contributors' details were updated.
- Bug in Bock's nominal IRT model was fixed. ICCs are now properly displayed for Czech Trivia dataset. Error for non-displayable table of parameters was improved.
- Bug in
DistractorAnalysis()was partially fixed, need to be revised.
- Reports:
- long tables have now header on each new page
- legend is now visible for IRT plots even when dataset consists of more than 50 items
- Matching criterion in
plotMultinomial()was relaxed to decrease size of reports.
- Reports are now rendered even in case that multinomial model cannot be fitted for some items.
- With option Keep item names, names of items are properly displayed in Report for DIF detection methods.
- Model comparison table returns now also partial results when some models cannot be fitted.
- Bug in plotMultinomial() was fixed. In case of only two categories observed, plot is displayed properly now.
- Bug in
DistractorAnalysis()was fixed. In case of possibly discrete matching, matching is cutted based on its factors.
- Uploaded data can be now unloaded without restarting app.
- Distractor plot for validity analysis was updated to allow for lowering the number of categories in cases when cuts are not uniquely defined. Application was updated accordingly.
- Errors and warnings were customized.
- In all dichotomous models the number of EM cycles can be increased in Settings.
- Seed was set for validity boxplot for reproducibility of jitter points.
- All static DOI links were updated. (Thanks to Katrin Leinweber)
- Some typos were fixed in reports.
- Option for displaying session info in reports now functions properly.
- Option to display average score in DDplot in reports fixed.
- New function
plotCumulative()which plots category probabilities functions estimated by \code{multinom()} fromnnetpackage was added. It is now used in application. - New function
plotAdjacent()which plots category probabilities functions estimated by \code{multinom()} fromnnetpackage was added. It is now used in application. - New function
plotMultinomial()which plots category probabilities functions estimated by \code{multinom()} fromnnetpackage was added. It is now used in application.
- Selected R codes for cumulative, adjacent and multinomial models were updated.
- Option for matching criterion (total vs Z-score) in multinomial model was added.
- Equations in DIF section are now center-aligned.
- Alerts for analysis where data with missing values are omitted were added.
DDplot()now can displayed average item score as estimate of difficulty for ordinal data. See argument average.score.
- Correlation structure now use pairwise complete observations for calculation of correlation.
DistractorAnalysis()andplotDistractorAnalysis()functions were updated to allow for lowering the number of categories in cases when cuts are not uniquely defined. Application was updated accordingly.
- Tables and equations of polytomous models in Regression section are now center-aligned.
- With option
Keep missing valuesunfunctioning analyses were fixed (missing data removed, warning message added)
- Multinomial regression plot is now displayed using smooth function
- With option
Keep missing valueswarning message is now displayed for those analyses for which missing values are omitted. - For ordinal data, Cronbach's alpha is now calculated from original ordinal (non-binarized) data both in Reliability and in Item analysis section.
- Busy indicator now properly displays above input tabs.
- Bug when using matching in
DistractorAnalysis()was fixed.
- New ordinal toy dataset SCIENCE was added.
- Each tab is now open at the top of the page.
- Adjacent logistic regression model was implemented for ordinal data.
- Both adjacent and cummulative regression models allow for selection of matching criterion (total score, standardized total score) and parametrization (classical slope/intercept and IRT parametrization).
- SD in Regression section was changed to SE.
- Keep missing values button is now functioning and it is applied for ordinal and binary data.
- Cumulative logistic regression model was implemented for ordinal data.
- Delta method is now applied for calculation of SE in IRT parametrization in cumulative logistic regression.
- Bug in slider in histogram of total scores was fixed. Range is now correctly displayed when NA values are present.
- Fixing versions.
- Math mode is now stable in tables. (Thanks to Stéphane Laurent)
- UI for polytomous models in Regression section was added.
- Traditional item analysis table is done via ItemAnalysis() function. It now includes also average score.
startShinyItemAnalysis()automatically launches default web browser.- Sample code and wording was updated for traditional item analysis table in app and also in reports.
ggWrightMap()now can display user-specified names of items (Reported by Matthew G R Courtney)plotDIFLogistic()now can display user-specified names of reference and focal group.- Some typos were fixed.
- Analysis for ordinal items has been added. This includes use of ordinal data in sections Summary (histogram of Total scores, success rate), Reliability, Validity (correlations, predictive validity),
- Equations for 3PL and 4PL regression models were fixed.
- Math symbols are now displayed using MathJax.
- Exercises were added in IRT section / Polytomous Training.
- DIF section now includes method comparison.
- Labeling of correction methods in DIF section was improved.
- Ability parameters are now available also for Bock's model.
THIS IS A CRAN VERSION
It includes versions 1.2.9-1 - 1.2.9-3 and following changes:
- Reference to a new paper on
ShinyItemAnalysisin The R journal was added.
- Histograms for total scores now handle cut-scores out of range of observed values
- Bug in reports when using delta plot with purification was fixed.
- Cut-scores for histograms are properly displayed in reports.
- Year in reports was updated.
- Sections About, References and Settings are displayed only with icon and are right-aligned.
- Margins and padding were updated.
- t-test was added for comparison of total scores between groups into DIF section and into reports.
- Histograms for total scores are now displayed in plotly mode.
- Some math mode was fixed in IRT section.
- Some typos were fixed.
- Selected R code was updated in DIF/Total scores.
- Animate function for sliders was slowed down.
- Generalized ULI index is now correctly displayed in traditional item analysis table in reports.
- Item names are now correctly displayed in reports.
- Item names are now correctly displayed in DDF section.
- Function
plotDIFLogistic()now handles user-specified matching criterion.
- Legend of DDplot was modified.
- Wording in PDF/HTML reports was edited.
- IRT/Training/Dichotomous models - plots are correctly plotted.
- Download buttons in IRT/4PL works now.
- TIF were converted into plotly.
- Section Validity/FA is not accessible for now.
- Selected R code for IRT/Bock's model was updated.
- Wording in IRT models was updated.
- Wording in Reports was updated.
- Legend for plots of IRT models was modified in Reports.
- Ability estimates summary table is now round in Reports.
- Some typos were fixed in Reports.
- Colours of IRT summary plots correspond to model for Items.
- IRT section of shiny app was updated and now provides new and more shiny figures in plotly
- PDF and HTML reports were updated and now allow
- different types of correlation in corrplots
- different types of discrimination indices in DDplots
- summary table for ability estimates in IRT section
- Dendrogram is now correctly rendered.
- Function
ItemAnalysis()was added. - IRT section now allows download of ability estimates
- Help pages for functions were updated.
- Key for ordinal data for binarization is now correctly used. Values greater or equal to value of the key are set to 1.
- When NULL data is uploaded, default dataset GMAT is in use.
- Scree plot was moved into the new tab
Factor analysis. - Dendrogram was added into
Correlation structuretab. - Option for global cut-score for ordinal data was added.
- R package ggdendro is now used.
- Removal of items with only 0s or 1s now works.
- Removal of observation with NA group now works.
- Text in Data section was updated.
- List of references was updated.
- Selected R code in
Correlation structuretab was updated.
- Bug in IRT models/Training/Dichotomous models text of theta in plotly was fixed.
-
Reorganization of data upload and renaming data variables. test_answers()> nominal() correct_answers()> binary() scored_test()> total_score() test_key()> key() DIF_groups()> group() criterion_variable -> criterion()
And creating new variables: ordinal() ordinal data maximal() maximal values for ordinal data minimal() minimal values for ordinal data z_score() standardized total score, no need to use scale(total_score()) in each function
- Updating text in Data section.
- Editing boxes for data upload in Data section.
- Bug in Regression/Nonlinear 4P IRT Z (rounding in interpretation) was fixed.
- Bug in IRT models/Training/Dichotomous models text of theta in plotly was fixed.
- Bug in IRT models/Training in plotly a ModeBar was fixed.
- Data upload now allows also ordinal data, min/max values can be specified.
- Validity/Correlation structure now allows selection between Pearson/Spearman/Polychoric correlation, Polychoric is default for binary data
- DDplot now allows also ordinal items and offers different types of discrimination indices to be displayed
- Download of tables (as .csv files) is now provided for standard scores, correlation matrix, traditional item analysis, and for parameter estimates in IRT models.
- Data upload now better points to "Upload data" button.
- Precision was re-set for IRT models/Training tab. Now = 0.1.
gDiscrim()function has been extended to calculate Upper-Lower Index (ULI) also for ordinal items
- Bug in IRT dichotomous models comparison is fixed. (Reported by Jean-Luc Kop)
- Bug in IRT dichotomous models (3PL and 4PL) was fixed. Tables of parameters are rendered even in case that EM cycles terminated without convergence. In such a case SD of parameters are set to NA.
- Bug in
plotDIFLogistic()function was fixed. Function now returns also item.name argument. - Bug in
plotDIFirt()function was fixed. Function now returns also item.name argument.
- Global setting for figure downloads was created. This can be changed in Setting section.
THIS IS A CRAN VERSION
It includes versions 1.2.7-1 - 1.2.7-7
- Section Reliability was added.
- Split-half methods including also Revelle's beta were added
- Cronbach's alpha method with CI was moved to section Reliability.
- Spearman-Brown formula was added.
- New function ggWrightMap replaced wrightMap from WrightMap package.
- Section IRT models
- Selected R code for Rasch and 1PL models was updated.
- Reference list was updated.
- Section DIF/Fairness
- Subsection Logistic IRT Z is deprecated.
- Subsection Nonlinear IRT Z is now called Generalized logistic.
- Subsection Generalized logistic includes now wide range of models and different specification of DIF type.
- Subsection SIBTEST was added, including SIBTEST method and Crossing-SIBTEST method.
- Reference list was updated.
- Some wording was updated.
- Selected R code was updated.
- Function theme_shiny is deprecated.
New function
theme_app()was created and applied for all graphics, including R functions in package.
- Paddings were adjusted.
- Acknowledgments were updated.
- Crashing of the app when upload no data was fixed.
- Data page
- Design of data page was changed.
- User can now specify encoding of missing values.
- Basic summary was added.
- Phone version
- collapsible menu closes when clicked
- Plotly files in IRT training section are now displayed with mode bar due a bug in plotly package.
- Some typos were fixed.
- Link to google form for feedback was added into About section.
- New R icon was added into footer.
THIS IS A CRAN VERSION
It includes versions 1.2.6-1 - 1.2.6-12
- Mirror at Charles University is no longer in use.
- Function
DDplot()- Limits for y-axis were adjusted when negative values of discrimination are detected.
- Warning for negative discrimination values were added.
- Reports
- Some unused parts of reports were excluded.
- UI
- Some parts were splitted from main ui.R file
- IRT
- Exercises in IRT training were extended.
- Data
- Bug in data upload was fixed. Now length of criterion variable is correctly checked.
- Packages - shinyBS is now required.
- Data
- HCI dataset was added into package.
- Description of dataMedical was improved.
- Description of training datasets was improved.
- About
- Page was improved.
- IRT
- Helps for interactive exercises were added.
- Typos were fixed.
- IRT
- Expected item scores were removed from IRT training NRM.
- Interactive exercises were added into IRT training - dichotomous models.
- Data
- Graded data od Medical data set was added.
- In all dataMedical data sets the criterion variable was renamed to StudySuccess.
- IRT
- Expected item scores were added into polytomous models in IRT training.
- Selected R code was updated in IRT training.
- IRT section
- Colors of sliders and curves in IRT training section were adjusted.
- Regression
- NLR 4P model was added into model comparison.
- README file was updated. Typos were fixed.
- Design of figures was unified.
- Download button options were unified.
- Selected R code syntax was improved in some sections (IRT only partionally, DIF not).
- Selected R code was improved for first 4 sections.
- Some wording was improved.
- IRT section
- Wording for IRT training was added and improved.
- Selected R code syntax was improved in some sections. This can be
done with
txt <- gsub("\n", "
", txt) txt <- gsub(" ", " ", txt) txt where txt is R code in "". - theme_shiny was created as a default theme for all figures. Now it is used only partionally.
- README file was updated. Typos were fixed. Links were added. Hexbin was added.
- References were updated.
- IRT section
- Selected R code was added into polytomous models.
- Equations were added into polytomous models.
- Download buttons were added into polytomous models.
- 4PL model was added.
- 4PL model was added into model comparison.
- Sliders in IRT training are now colored with respect to line colors.
- Regression section
- Section is newly organized.
- NLR 4P model was added.
- Reports
- 4PL IRT model was added as an option.
- IRT equation is generated based on selected model only.
- Unnecessary files were deleted. Some cleaning of files was done.
- IRT section
- Selected R code in model comparison was updated. Typos were fixed.
- Regression section
- Interpretation of NLR 3P model was extended.
- Syntax for NLR 3P model was improved.
- IRT section
- IRT section is newly organized.
- IRT training section was expanded by polytomous models. GRM, GPCM and NRM were added.
- Syntax for IRT training was improved.
- IRT section
- Sliders steps are now 0.01 in dichotomous models.
- Numeric inputs were removed from IRT training section.
- Selected R code was extended in IRT training section by item information function plot.
- References were updated.
- Mirror web link was updated.
- Reports with DDF but none DDF item are now correctly generated.
- Hexbin
- was added into footer.
- is now as shortcut icon.
- Paddings
- in large devices was adjusted.
- in small devices was adjusted.
- Unnecessary files were deleted.
- Reports
- Some wording in reports was improved.
- PDF report text is fully justified.
- Subsections characteristic curves is not visible when none DIF or DDF item is present.
- Some typos were fixed.
- Some parameters were renamed. The unneecessary ones were removed.
- UI for
DIF/Fairnesssection was updated. - References were added into About page.
- New design of reports is now available. Design of PDF reports was updated.
- Wording in report was improved.
DIF/Fairnesssection in report was improved.- New hexbin for
ShinyItemAnalysiswas created and added into reports. - Sample code in
DIF/FairnessIRT Lord and IRT Raju was extended to 1PL, 2PL and 3PL models. - Bugs fixing
- Wording in regression plots was fixed.
- Examples in
DIF/FairnessIRT Lord and IRT Raju were fixed. - Function
plotDIFirt()now can be also used for only one item.
- Reference list was updated.
- CITATION file was modified.
- In Summary page the description for summary table was added. The sample code was updated.
- Wording in summary statistics in Summary section was improved.
- Description of Session info section in reports was added.
- Correlation plots now offer reordering by hierarchical clustering. Several linkage methods are offered. This is also included in reports.
- Confirmation of data upload was added. Also warnings and errors are displayed in case that something went wrong when uploading.
- In IRT training
- parameter theta (latent ability) is now included
- the interpretation for characteristic curves is provided
- horizontal and vertical lines for specified theta were added into plot for characteristic curves
- selected R code was added.
- In multinomial regression bug when only two choices in data are present was fixed.
- Link for Adela's website was added.
- Columns for BIC and its adjusted version were swapped in table for IRT models comparison. Text above table was also updated.
- In IRT section dpi of downloaded plots was set to 300.
- In IRT training downloaded plots also include legend with parameters.
- In all IRT models table with parameters includes all parameters a, b, c, d.
- In IRT training equations were moved above interpretation.
- Reports were mildly updated.
- Resolution of figures in reports was increased (now 300ppi)
- Dependencies were updated.
- Some typos fixed.
- Legends of plots were unified and placed in better way for
NLR IRT ZDDF- function
plotDIFLogistic()
- Bug in item names for
DDFsubsection ofFairness/DIFsection was fixed.
THIS IS A CRAN VERSION
- Report's customized options were extended. Reports now include also dynamic description.
- Author and dataset name can be now filled and add into reports.
- Dataset
difMedicalwas renamed toMSAT-Baccording todifNLRpackage. Its description was mildly updated. - Purification options were added into
Nonlinear IRT ZandDDFsubsections ofDIF/Fairnesssections. - Purification options was added
DDFanalysis inReportssection when customized setting is selected.
- Size of plots and font in plots was changed for reports.
- Plots in HTML report are now centered.
- Wording of reports' default option was updated.
- References were updated in
Referencesection and also in README file. - Wording of Spearman test in
Validitysection was mildly updated.
- Design of reports was edited.
- Several bugs were fixed:
- Bug when creating report for uploaded data was fixed.
- Bug when creating report for data without predictive criterion or DIF group was fixed.
ShinyItemAnalysisnow usesplotlypackage!
- Design and options of
Reportpage were updated. - Layout of IRT curves training page was mildly changed. There are now two curves in each plot. Also text area input was added and synchro- nized with sliders. Colour of sliders now correspond to colour of curve.
- Several bugs were fixed:
- Interpretation of regression models was fixed.
- Interpretation of text about correlation in Predictive validity was fixed.
ShinyItemAnalysisnow used alsodata.tablepackage.- Data page was divided into two subpages to avoid long rendering of tables. when changing datasets.
- Function
gDiscrim()to calculate generalized discrimination was implemented. - Function
DDplot()was edited to be able to handle generalized discrimination. Double slider was added intoTraditional analysis page. - Traditional analysis table was extended by adding column with generalized discrimination. This is done also in reports.
- Some pages has now their own source code files:
Summary- Summary.RValidity- Validity.RTraditional analysis- TraditionalAnalysis.RRegression- Regression.RIRT models- IRT.RDIF/Fairness- DIF.IRT
- Last three rows in regression comparison table are now bold.
- Reference list was extended.
- The subpage
DistractorsofItem analysispage was mildly edited. - Generalized DD plot was removed from
Distractorsubpage. It is now available onTraditional item analysissubpage. - Item patterns frequency plot is now made with
ggplot2package. - Names of downloaded plots were customized and their resolution was
increased and size of text was descreased in some pages:
- Summary
- Validity
- Item analysis
- Regression
THIS IS A CRAN VERSION
ShinyItemAnalysiscan now handle missing values. Newly there is an option to handle NA values as zeros in scored data.- Bug in WrightMap was fixed.
- We have new contributor - Lubomir Stepanek!
- New online mirrors were added.
- Busy indicator and tooltips were added.
- Structure of application was mildy updated.
- Datasets
dataMedical,dataMedicalkey,dataMedicaltestwere updated. Group variable and criterion variable were added. - The newest versions of some packages are now required (
difNLR v 1.0.2,difR v 4.7,ggplot2 v 2.2.1,mirt v 1.24,shiny 1.0.3andshinyjs v 0.9). - DIF/Fairness
- Option of item purification was added into some DIF detection methods - Delta plot method, Mantel-Haneszel, Logistic regression with total score, Lord's test and Raju's test.
- Default option of p-values adjustments in DIF detection methods
was set to
none. - Bugs in DIF Logistic IRT Z and DIF Nonlinear IRT Z tables of parameters were fixed.
- IRT models
- IRT models are now fitted only by mirt package.
- Training plots of item characteristic and information curves were added into IRT models section.
- Pearson's correlation coefficient between factor scores and standardized total score was added for Bock's nominal IRT model.
- Reports
- New options for reports were added. By default no IRT models and no graphs of correlation structure are fitted.
- Reports are now described in detail.
- The settings of reports are now also offered in Report section.
- Generation of reports is now done part by part. Once it is finished, the download button is offered.
- Message boxs with warning for generation and download of reports were added.
- Validity analysis:
- Validity analysis of predictive outcome is now included on Validity page in Predictive validity section.
- Now it is possible to upload predictive outcome.
- Dataset
GMATandGMATtestfromdifNLRpackage include now criterion variable which can be used in Predictive validity section. - Functions
DistractorAnalysis()andplotDistractorAnalysis()were extended bymatchingargument.
- Item analysis
- Barplot of answer selection proportion was added into Distractor subsection.
- Diagram of custom discrimination - new version of DD plot - was added into Distractor subsection.
- CITATION file was added.
- Maximum upload size was set to 30MB.
- Item names were fixed and unified.
- Some typos were fixed.
- Readme file was updated.
- Reports design was slightly updated.
- Code was slightly updated to correspond to new versions of packages
difNLR, mirt andggplot2. - Selected code was updated.
- Package dependencies were updated (packages
knitrandxtableare now used in report generation). - Minor bug in DIF IRT (Lord and Raju) plots and downloads was fixed.
- Designs of footer and navigation bar were updated.
- The version for small media was designed.
- Weblink for version 1.1.0 was created.
- Reference list was expanded and links to online versions of articles were added.
THIS IS A CRAN VERSION
- New overall appearence using shinythemes package.
- New datasets
dataMedical,dataMedicalkey,dataMedicaltestwere added into the package. - New dataset is offered in example datasets inside the shiny app.
- Bug in Distractor plots for uploaded data was fixed.
- Calculation of Success Rate in Standard Scores was fixed.
- Differential Distractor Functioning detection method was added.
- Syntax of
difNLRfunctions was edited to response to new version of package. - Function
plotDIFirt()was updated. - Correlation structure section was added including correlation plot and scree plot.
- Pages structure was changed and updated.
- IRT models are now fitted by mirt package.
- Subsection IRT model selection was added into IRT models section.
- Wrigh maps were added for Rasch and 1PL IRT model.
- Bug in data upload was fixed.
- Subsection regression model selection was added into Regression section.
- Typos were fixed.
- Selected R code was updated.
- Description of data including uploading own data set was extended.
- Description of Standard scores was added.
- Lower panel was updated.
- Rasch model was added into IRT models section.
- Pearson correlation coefficient of factor scores and Z-scores were added for IRT models
- Reference list was expanded.
- Bug in multinomial regression for scored data was fixed.
- Bug in DDF for datasets withou DIF items was fixed.
- Selected code was updated.
THIS IS A CRAN VERSION
- New graphical representation via
ggplot2package - Cronbach's alpha was added
- Distractor analysis was extended
- More DIF detection methods were added Mantel-Haenszel test, Logistic regression, Lord's and Raju's statistic for IRT models all are now used via difR package
- Delta plots were extended: fixed and normal threshold are now allowed
- Multiple comparison correction methods were added into DIF detection procedures
- More examples of datasets were added
from
difNLRpackage - References were added
- New functions were added into the package:
-
DDplot()function for graphical representation of difficulties and discriminations. -DistractorAnalysis()function and its graphical representationplotDistractorAnalysis(). -plotDIFirt()andplotDIFLogistic()functions for plotting characteristic curves based on IRT models and based on logistic regression.