TiaPlus -- Revision history Build 314 (January 2014) ========= Added second extended POtablesExt export ('hekjesbestand') Build 313 (June 2013) ========= Minor bug fixes Build 312 (June 2012) ========= Fixed a bug where a Asymptotic Glb coefficient that could not be calculated was handled incorrectly. Build 311 (April 2012) ========= Fixed a bug in reading .CSV files where TiaPlus ignored leading blanks in the Identification string field, causing subgroup specifications to fail. Implemented a warning for users exceeding the upper item maximum score limit (max. allowed unweighted item score = 254) Some improvements in error handling related to GLB calculation. Build 310 (March 2012) ========= Fixed the calculation of the item mean score, which was misreported if the item had a item minimum score > 0. Fixed a bug in the calculations for the Percentiles Table. That is a table following the Frequency Distribution Table. Discarded the requirement that the PC on which TiaPlus gets installed has a minimum screen resolution. Build 309 (2010) ========= (on request) added subgroup numbers to the grid on Wizard Screen 2 and subtest numbers to the grid on Wizard Screen 4 Corrected an oversight regarding the TiaPlus help file location. TiaPlus should now find the TiaPlus1.hlp file immediately without user intervention. Windows 7 users might want to install a Microsoft tool to be able to view TiaPlus' HTML-Help. See http://support.microsoft.com/kb/917607/en-us Implemented a more efficient way to calculate Variance-Covariance matrices. Added support for MS Excel .xlsx file format data files (Office 2007 or later). If Office 2007 (or later) is not present on your PC then using .xlsx files requires the installation of the MS Office 2007 System Driver (Data Connectivity Components, free download) from http://www.microsoft.com/downloads/details.aspx?FamilyID=7554F536-8C28-4598-9B72-EF94E038C891&displaylang=en Build 308 ========= The standard error of measurement for a test will now be based on the asymptotic GLB coefficient, if available. Otherwise Cronbach's is used to calculate this value. Found a degenerate case in which calculating the GLB coefficient caused TiaPlus to stall at max CPU consumption. Fixed. Added detection of possible keying errors in MC items in the Flags.txt file Changed the default installation directory for TiaPlus to C:\Cito\TiaPlus\ in order to avoid UAC and data storage problems when running on a Vista or Windows 7 operating system. Expanded the size of item labels that TiaPlus can handle (internally) to 19 positions, however output files will only show the leftmost 9 positions. Expanded the max. size of the item database Id field to 22 characters. Found and fixed a bug causing TiaPlus to crash with a message indicating that a database record cannot have duplicate fields or keys. Reworked the Predominant Item Information dialog to make it more user friendly. Build 307 ========= Found and fixed an error not handling long path names (>= 68 characters) correctly. Improved the routine to detect Unicode files (having two bytes per character in stead of one). Although Unicode files process normally in TiaPlus you may want to load the file in TextPad and then save it (Save As... dialog) in PC (file format) ANSI (encoding). Added reporting the value for the correlation between Total Test Scores and External Criterion Scores if the "Item scores - External Criterium correlation" was selected. Per user request, changed the display order of TiaPlus jobs in dialogs (e.g., in Database Maintenance or in Template usage) to 'Most recent On top'. Fixed an error in the labelling of output files, ocurring when two or more subsequent analyses (different data sets) are done in one TiaPlus session. Found and fixed an error in the GLB calculation routine; in rare cases (e.g., with a test length of 3) the value of the GLB coefficient appeared to fall be below the value of Cronbach's Alpha coefficient. Build 306 ========= Added a test to see if TiaPlus has sufficient rights (write to disk) on locked down PC's. Fixed the routine to Weigh P-values to population ratios Modernised the "Select folder for graphic files" dialog in Wizard Screen 6 (improvement only available for the Windows XP or Vista operating system. Changed TiaPlus to remember the User Defined Norms Scale check mark status between sessions. Fixed a resize error attempting to display the Output Viewer. Corrected the Item Graphics now showing RitH values if that option was chosen. Build 305 ========= Changed TiaPlus to remember its screen position (Wizard) and the WindowState of the Output Viewer screen (normal, minimized, maximized) between sessions. Fixed a bug that prevented Factor Analysis to work properly when Missing as Zero was selected Expanded the 'Template' handling to also include Subgroup information when copying information from another test or database. Increased the maximum number of subgroups to 20 (was 10). Note that a substantial number of testees per subgroup is required to ensure that the analysis makes any sense. Think 50 testees per subgroup minimum, preferably 100 or (a lot) more. TiaPlus now creates an output file summarising the items in a test that may need reviewing. Its name is Flags.txt and this file is located in the same folder as the main analisys output file ( 99999 (5 digits). Now 6 digits are allowed. Build 300 ========= Relaxed the (partial credit) MC2R item scoring rules. Example: with a key of "ADF" now the answers ADF, AFD, DAF, DFA, FAD and FDA are all rewarded 3 points. AD or DA give 2 points etc. This means that there is no ordering requirement anymore. Corrected miscalculating Rir values for items having missing observations where the scoring convention is "Missing as zero". Corrected TiaPlus that it will not fail anymore on larger numbers of persons (N > 32767) when calculating the GLB reliability coefficient or performing Factor Analysis or calculating polychoric correlations. Fixed a bug that TiaPlus would crash if the Test description and / or the Population description text box were left empty. Changed TiaPlus to not alter filename capitalization anymore. So that e.g., C:\Todd's Data Files\Dataset.csv remains C:\Todd's Data Files\Dataset.csv Changed TiaPlus to (also) accept MS Excel spreadsheet files (*.xls) as data input files. This required several significant database changes. This means that as of build 300 TiaPlus will not be able to use older databases. The exception is that importing 'template'-item specification data is still possible. Changed TiaPlus to accept apostrophes (') in file and folder names, like in c:\todd's data files\dataset.csv Changed the TiaPlus data grids to adjust their row heights to the current (enlarged) Windows font size. Build 206 ========= Added some explanations to the users manual regarding to item types. Corrected a flaw in the transformation of scores to numerical grades (Dutch 1 - 10 system). Logo change. In the late nineties the "Cito" organization was renamed to "CitoGroep". Now under new management this move is reverted and we call ourselves "Cito" again. Created a new entry (for CSEM output file name) in Wizard Screen 6 adjusted TiaPlus.INI file and Users Manual accordingly. Created a new CSEM tab in the output viewer, and made further minor changes & fixes in the output viewer Fixed some bugs in Wizard Screen 6 where names (and paths) of files and the Graphics directory were sometimes not remembered correctly between sessions. Created the option to de-activate already defined subgroups (Wizard screen 2) and subtests (Wizard screen 4). When not 'Active' TiaPlus retains the definition but does not perform the calculations. Improved processing speed somewhat (a basic analysis, 156000 students, 180 MC items, in approx. 7 minutes on a 2.4 Ghz Pentium PC with 512 Mb RAM). Reduced the amount of internal RAM used by TiaPlus, to avoid disk swapping as much as possible. Fixed minor bug where TiaPlus refused to see (and use) an empty output directory (folder). TiaPlus now first clears the Wizard screen 3 column 'MC keys' (and 'Max score') before inserting new values supplied from a file, defaulting Max scores for MC items to "1". Fixed a grid initialization error for 'Weigh P-values to population ratios' module. Build 205 ========= Double clicking the output file name field (wizard screen 1) now starts the associated dialog for (re)specifying a file name. Double clicking the input data set name field (wizard screen 1) now starts the associated dialog for (re)selecting a data set. Average Rit (via Z-transform, weighted) value will be reported with test characteristics. Reworked a lot of source code to enable connecting to a MS SQLserver data base while batch processing (undocumented Cito feature) In CSV input data set files now not having a CrLf line termination in the last record is allowed. Corrected display of minimum test score in case of "calamity" items. Now can calculate GLB for subtests (with 90 items or less) even when total number of items in data set is > 90. Slightly improved input data set reading speed (while scoring). Input data sets now may have trailing extra information per record after last item answer/item score. Now can report conditional Standard Errors of Measurement. Includes a data base change. Adjusted the calculation of the average P-value in case the 'Missing as Missing' Option is selected. Adjusted incorrect behavior of several 'cancel' buttons. Build 204 ========= Added a short glossary to the User Manual (Appendix B). Data base change: corrected an error in OMR table (missing record) Corrected an error in Dutch language hotkey combination on wizard screen 1 Added support to perform a web checkup to see if an update is available on the Cito server Added global error checking in scoring module to avoid crashes Added global error checking in testing wizard screen 3 to avoid crashes Build 203 ========= Added the possibility to play a short sound when analysis is ready Reworked a lot of source code and expanded the data base to enable batch processing (undocumented Cito feature) Made the number of samples for the GLB coefficient user definable (via TiaPlus.ini file) reworked the GLB.DLL Build 202 and earlier ===================== Added the EIRF option (Empirical Item response Function table) Added confidence limits on P values option Added item scores - external criterion correlation option.