Windows Versions (at Curtin University)

Excel 2002 / 2003, Version 5.6.4, 29 June 2009.

This is the version for Excel 2002 or 2003. It will not work with Excel 2007 or 2010. The extensive update history for this version is summarised below (on this page).

Excel 2007, Version 5.8.2.1, July 2012.

This version will run with Excel 2007, or with the 32-bit version of Excel 2010. Comments on updates and known problems for this version are here (links to a pdf file).

Excel 2010 & 2013 , Version 5.10.4, 4 May 2014.

This version runs with Excel 2010 and Excel 2013. Comments on updates and known problems for this version are here (links to a pdf file).

Excel version checker

Have a look at this link if you're not sure which version of Excel you have.

Macintosh Version (at Curtin University)

Version 5.9.2.1, 26 August 2013.

For Excel 2011 users.

Student (Also see the soft where? page, and the Student Grants page at ASC.)

Version 5.7.5s, September 2009. For Excel 2007 only.

ASC download site currently has these versions:

Windows Versions

Excel 2002 / 2003: Version 5.6.4 (ASC site is up to date).

Excel 2010 & 2013: Version 5.10.2, 1 September 2013 (ASC site is NOT up to date).

Macintosh

Not available for download from ASC (may be requested from this site; write to support@lertap.com).

Student

12 September

1

Lertap5MacroSetA.xla updated with two new macros: StatsbSortAZ, and NumericFilter1. The first will sort a Statsb report by item difficulty, while the second will create a new dataset populated with records having test scores in (or outside of) a specific score range.

19 June

1

Macintosh version has major update, bringing it to par with the Windows version.

2

Quintile plots now use thicker trace lines, making then easier to visualize.

15 June

New features: The capability to recode columns, or "variables", now exists within Lertap. The new recode macro has special codes, Exclude and Delete, which add to its flexibility. The group differences captured in the breakouts reports are now augmented by an analysis of variance table. Lelp is only a click away.

23 May

Added a new document, "Visual Item Analysis", to attract more users to Lertap's quintile plots. Click here to have a look at it (pdf document, about 400 KB).

25 April

Adjusted the way user-created macros are entered and accessed: now they're in an Excel Add-In called Lertap5MacroSetA.xla. Use the Lelp link in the next topic to become full-bottle on this handy new approach (makes the job of entering user macros quite a bit easier).

10 April

Substantial enhancement: a new menu option, "Macs", has been added to the toolbar. It provides the opportunity for users to have their own macros called directly from the toolbar.

Lelp yourself to a good read (refer to the April 2006 revision entry).

8 March

Substantial enhancements: the MDO option now works so that all item statistics are adjusted for missing data, and MDO is no longer limited to affective subtests: it may also be used with cognitive items. (See 11 September 2004 notes below.)

A did-not-see option is now available to support users of online test generators. Lertap will adjust its item statistics so that they're based only on those students who saw an item. To date this new option has been used to analyse data output by the Angel Learning and Test Pilot systems.

Lelp has the scoop, per usual (refer to the March 2006 revision entry).

3 March

Bug fix. *tst was not working correctly with selection criteria of variable length. See 28 November 2004 below.

18 February

Substantial enhancement: a new option on the Run menu will breakout any test score by groups (for example, by gender, region, study code, and so on), with a corresponding chart automatically included.

A new option on the Shorts menu will see that the group results are displayed in a simulated box and whisker plot, also known as a boxplot.

The histogrammer has increased skills: it will prepare summary graphs for group scores, and can now also "plot" the Log(PROB) values from a response similarity analysis.

Lelp has the latest, as always (see January 2006 revision entry).

10 January

1

Changed the version number to 5.6.

2

Completed the documentation for the enhanced RSA version.

31 December 2005

Season's Greetings to all, and best wishes for 2006.

5 December

Substantial enhancement: Lertap's response similarity analysis, RSA, has been improved with the addition of an inferential statistic to test the "significance" of RSA results. A new RSA report, RSAsig, has been added, and once again the RSAcases report format has been improved. New options now control RSA runs, including an option which allows poorly-performing items to be omitted from RSA calculations.

Documentation for this enhanced RSA version should be ready early in 2006.

11 November 2005

Improvements were made in the formatting of the RSAcases report.

1 November

Minor enhancements were made to catch some common *sub card syntax errors, such as Title=Survey1, which should be Title=(Survey1).

19 September

Substantial enhancement: two new options on the Shorts menu make it easier to create and modify score histograms. Now users can control the number of bars seen in a histogram, and it's no longer necessary to have the Analysis ToolPak Add-In installed.

2 August

Minor enhancement. The RSAtable format was changed so that H-H values at or above the cutoff would have a red "blip" in the table, making them easier to spot. (RSAtables are output as part of the response similarity analysis mentioned at 11 July below. Prior to this change such values often had no blip at all in the table, depending on the overall total number of H-H values in the table.)

19 July

1

Changed the version number from 5.4.6 to 5.5.

2

Minor bug fix. Having production mode set to "yes" would not allow "output an item scores matrix" to run if there was only one subtest.

11 July

Substantial new feature: support for response similarity checking was added. Lertap will now create three new worksheets, and a text file, suitable for analysis by so-called "cheating checkers". Our work features an interface to George Wesolowsky's "excessive similarity checker", SCheck.exe.

Lelp has the latest (see July 2005 revision entry).

9 May

Minor bug. A small but annoying bug was fixed in the quintile plots -- if and when data tables were added to each plot, the height of the graph would reduce, resulting in a squat, scrunched plot. This has been fixed.

8 May

Substantial new feature: a production mode capability was added as an option. When activated, a click on "Interpret CCs lines" will automatically roll through to "Elmillon item analysis", without stopping to suggest that the Freqs report be examined. Additional options make it possible for the roll to include histograms, quintile plots, and IStats reports, one for each subtest. Now a single click will take the place of several. Will be handy for production shops.

As always, Lelp has the latest (see May 2005 revision entry).

28 April

Bug. The Stats1f report's Mean/Max bands were not being calculated correctly. If a user had the MDO option in effect for an affective subtest, the item Mean used in the band was uncorrected— now the adjusted item mean, with missing data excluded, is used, and the Mean/Max bands are accurate (note that this problem did not occur if the MDO option for an affective subtest was not in use).

25 April

Bug. The Stats1f report was found to have incorrect p values for cognitive items, or incorrect % values for affective items, on the "other" line when the MDO option was in effect.  Fixed.

11 April

Changed the version number from 5.4.5 to 5.4.6.

8 April

Minor bug. Excel crash encountered when processing a set of affective subtests, with all items in all subtests having the same response (a very uncommon happenstance).  Fixed.

22 February

Bug. The *mws card was found to fail when asked to work with EMQ items.  Fixed.  (EMQs are mentioned immediately below.)

18 February

1

Substantial new feature: items may now have as many as 26 options.  This means that EMQs, extended matching questions, may be Lertapped.  With this change, the following statement is valid:

Res=(A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q,R,S,T)

This Res= declaration says that items use 20 options, A through T.  Can be A through Z, if wanted.

The format of the *alt card was changed to support the use of more options.  Read about this by looking at the Revisions topic in Lelp (see February 2005 revision entry).

This enhancement was requested by Jerome De Lisle, University of the West Indies (Port of Spain, Trinidad).

2

Substantial new feature: new options have been installed for quintile plots.  Now you can have Excel automatically add data tables to the plots, and you may control the number of items to be quintiled. Read about this by looking at the Revisions topic in Lelp (see February 2005 revision entry).

These options are in response to suggestions received from

Hugh Moore, Deakin University (Melbourne, Australia).

3

Substantial new feature: A cognitive item’s difficulty index may now be adjusted so that people who have not answered the item are excluded from the calculation of the index. Read about this by looking at the Revisions topic in Lelp (see February 2005 revision entry).

This modification was made after e-discussions with Hui Deng of The College Board, New York.

10 February

New upgrading capability: It is now possible for Windows users to get the very latest version of the Lertap5.xls file by visiting the Software page at www.lertap.curtin.edu.au.  (Best to read about this in Lelp before trying it; it’s under the Welcome topic.)

26 January

Corrected a very minor formatting problem which occurred when the “Make a new blank Lertap 5 workbook” option was taken from the New menu: the height of the rows in the new CCs worksheet was not the same for all rows (now it is).

9 January 2005

Installed an error-trap to cover problems which can arise when using the advanced toolbar on Macintosh computers.

31 December 2004

It’s that time again (!). Season’s Greetings to all; best wishes for the new year.

Suggested prime New Year’s resolution: “In 2005, I will provide more feedback to the terrific people at Lertap HQ.”

27 December

1

Adjusted the status bar message so that it advises the time it may take for Lertap to extract eigenvalues from the interitem correlation matrix.  (As mentioned in Lelp, when the number of items exceeds 50 it may take a matter of minutes to get eigenvalues on some computers.)

2

A small error was fixed in the new *tst code mentioned immediately below:

28 November

Substantial enhancement: the capabilities of the *tst “card” were expanded to allow for longer filtering criteria.

Now the *tst card may have a format such as this:

  *tst c12=(fullstrength, lite), c14=(bitter)

The line above would see to it that all Data records having either “fullstrength” or “lite” in column 12, and “bitter” in column 14, will be filtered out to a new data set.

(Two of Western Australia’s finest beers, Emu Export and Matilda Bay, meet these criteria.)

Prior to this change, the *tst filtering criteria could be only a single character long.

This change was undertaken at the request of Fae Mellichamp, Professional Testing Inc., Tallahassee, Florida.

20 November

Addressed a problem related to the sort icon on Lertap’s toolbar.  On very fast Windows computers, and on all Macintosh computers, Lertap’s sorting process will not quite complete its job as expected.  See comments under the “sort” topic in Lelp (the help file).

17 November

1

Substantial enhancement: added the *exc card to make it easier to exclude items from a subtest.  Read up on it by looking at the Revisions topic in Lelp (the help file).

2

Changed the Freqs report so that it indicates the column number used by each item, as well as the item’s actual ID.

3

Changed the way Lertap processes data when a student has not answered any of a subtest’s items – now these students will still get a score if the subtest has items credited to all test takers (this is usually done by using *mws cards with an other= assignment, as discussed in Lelp (the help file)).

4

Bug. Fixed a bug in the IStats report which used incorrect item IDs whenever some items had been removed from a subtest.

9 November

Another technical paper was added to the website, this one in answer to the question: “Why don’t Lertap’s results match those from my old item analysis program?”.

ItemCriterionCorrelations1.doc

31 October

1

Changed the version number from 5.4 to 5.4.5.

2

Added a technical paper to the website having to do with the use and interpretation of eigenvalues, and on methods for improving the value of coefficient alpha.  See it at:

JERM2005a.doc

24 October

Installed an error message to handle matrix inversion problems, should they occur.  (Lertap inverts the IStats correlation matrix in order to derive SMC values, as mentioned below, under 30 September 2004.  Not all correlation matrices are amenable to being inverted (who can blame them?), and here at Lertap HQ we’ve found that matrices which will invert okay under a Windows-based computer will sometimes fail to invert on a Macintosh.  There was an obvious need for an error message.)

20 October

1

Substantial enhancements: more modifications were made to the IStats report:

Another row has been added with item loadings on the first (unrotated) principal component of the correlation matrix.  These loadings are then plotted in ten bands, making it easy to see which items best correlate with the first component.  If the option has been taken to place SMCs on the diagonal of the correlation matrix, this row will then have loadings on the first (unrotated) principal factor.

In addition, Lertap now indicates how high alpha would be if the loadings on the first principal component were used as weights on the item scores.

These enhancements are designed to aid the process of identifying those subtest items which most contribute to the subtest’s internal consistency, as indexed by coefficient alpha.

2

The Stats1f report has been modified so that items whose removal from a subtest would result in an improvement in coefficient alpha are highlighted (by using a bold font).  This has been done in order to make it easier to spot items that don’t seem to be contributing to the subtest’s internal consistency.

8 October

1

A row with the percent of total variance corresponding to each eigenvalue was added to the IStats report.

2

The 8-ball option, used to reformat a Lertap worksheet so that it’s ready for input to another program (such as SPSS), was modified so as to behave properly in the presence of the new SMC bands (mentioned immediately below, under 30 September 2004).

30 September

1

Bug. The eigenvalue extraction routine was found to have an error which affected the derivation of the very last eigenvalue (the other eigenvalues were okay); this was repaired.

2

A modification was made with regard to the calculation and display of SMC values (squared multiple correlations).  Now SMCs are always displayed as an extra row appended to an IStats correlation matrix.  Then, below this row, the SMCs are “plotted” in a series of ten bands, making it easy to see their pattern.

(The IStats worksheet is obtained by using the Run menu, clicking on the option to “Output item scores matrix”.  The option to replace the diagonal elements of a correlation matrix with SMCs is still available, unaffected by this change.)

3

Bug. It was found that Excel’s standard matrix inversion routine would regularly fail when directed to invert a correlation matrix with order greater than 60 (that is, for a data set involving 60 or more items).  The M_INV routine from the matrix analysis package maintained by the Foxes Group is now used instead of Excel’s MINVERSE function.

(For more about the Foxes Group, refer to 2 October 2003 below.)

15 September

A minor correction was required to the work of 11 September 2004.

The Lertap help file, Lelp, was updated with more information about missing data, and how Lertap handles it.

11 September 2004

A substantial change was made so that it’s now possible to filter out missing data when affective item statistics are computed.  This is done by using the MDO control word on a *sub CCs line.

This will be a welcome enhancement to users who collect survey data, and wish to get results based only on people who have actually answered the survey questions.

Read about this change by going to Lelp, and looking at the Revisions discussion under the R&R&R&R topic.

9 September

With the kind permission of Leonardo Volpi, Foxes Group (Italy), the eigenvalue extraction routines found in the Matrix.xla Add-In were copied to Lertap, where they now reside as internal functions and subroutines.  This means that it is no longer necessary to download and install the Foxes Add-In; Lertap is now capable of computing eigenvalues on its own, without requiring an add-in.  (See original reference below, at 2 October 2003.)

30 July

Added an option to the Move+ menu which eases the job of creating a formula score – users who re-scale test scores will find this useful.  For more info, look at the Revisions topic in Lelp.

29 July

Bug. Corrected an error found by Fae Mellichamp, Professional Testing Inc.  The Halve&Hold option would fail if the Data worksheet was not the first sheet registered to a workbook.

9 July 2004

Relaxed the record ID location requirement so that Data records may have their ID in any column of the Data worksheet.  Prior to this revision, Lertap required the record ID to be in either the first or second column of the Data sheet.  Now the ID can even come after the item responses (if wanted).

This change was requested by Barbara Foster, University of Texas.  It’ll be real handy for those users who like to record item responses starting in column 1 of the Data worksheet.

22 June 2004

1

Created a new document with solutions for an Excel limitation: the maximum number of text characters which may occupy a worksheet cell is, for Lertap purposes, 255.  This poses a problem for Lertap users with long *col lines in their CCs worksheets – this doesn’t happen very often, but it’s a big problem when it does.  Have a read of this Word document:

ExcelColumnLimitProblem1.doc

The document’s third solution has suggested work-saving guidelines for all Lertap users who frequently have jobs with multiple subtests.

2

Modified Lertap so that the Windows version offers a link to Lelp whenever the 255-character limit is exceeded.

12 June

Improved Lelp, the Lertap help file, by making it easier to access the System Worksheet topic.  (This topic was formerly classed as a pop-up, and, as such, would not display in the former PDF version of Lelp.)  The System Worksheet topic in Lelp, when viewed in interactive mode (that is, not PDF mode), now has numerous hyperlinks to related Lelp topics.

What’s the System Worksheet?  It’s where Lertap’s various options are set.

Click here to access the web version of Lelp.

21 April

Added references to fresh research from the University of Alberta dealing with the use of classical item statistics to estimate the parameters seen in common IRT models.

Lertap can be used to derive such estimates.  For more information, refer to Experimental Features in Lertap 5.

3 April

1

Made the Student Version a version unto itself, as it used to be.  This was done to eliminate the acknowledged bug mentioned below (see 7 March).

Made minor improvements to the formatting of quintile plots (see 30 October 2003 below), and fixed an error uncovered by Dra. Lila Quero-Muñoz.  The error was found when processing a cognitive subtest with just two items, with one of the items credited to everyone – Lertap was unprepared for this special case, and bombed out.

2

Changed the version number from 5.3 to 5.4.

3

Delivered version 5.4 (@3April2004) to Assessment Systems Corporation (www.assess.com).  For a summary of its main features, mouse click on me.

29 March

Improved the Shorts menu’s options so as to include a link to a new macro which will make very quick work of a line graph.  Now it’s very easy to have Excel make a line-graph of any series of selected cells, for example, of item difficulties, item discrimination coefficients, or eigenvalues.

Updated Lelp (the help file).

16 March

Installed a new document with links to, and summaries of,  Lertap reviews.

7 March

Added a new document on how to score essay and short-answer questions with Lertap.  A click here will take you to it.

Added comments on the Student Version to Lelp (the help file), and to the KnownProblems page.  There are times when Lertap will be made to believe it’s running the Student Version even when users know they’re not (this is an acknowledged bug).

7 February

Some hopefully-helpful hints added to Lelp, the help file, for users of ASC’s FastTEST item banking and test development system.  These hints discuss how to ease the process of exporting item statistics from Lertap so that they may then be imported by FastTEST’s Import Wizard.

3 February

Added support for XCALIBRE users.  (XCALIBRE is an IRT program from Assessment Systems Corporation, ASC.)  There is now a line in the System worksheet which directs Lertap to create a worksheet formatted for use with XCALIBRE.

The Lelp file has also been updated so that it discusses how to create a csv file (comma separated values) suitable for use with another ASC product, FastTEST.

Read all about it in Lelp, the help file.

28 January 2004

Upgraded the Macintosh versions so that they’re basically on par with Windows equivalents.  Mac users should write to support@lertap.com for more information on how to obtain and run Lertap versions.  (The Mac information on the commercial site at Assessment Systems Corporation is not as up to date as that found here.)

In the process, changed the way calls are made to the “Analysis Toolpak” Add-In.  It’s possible that this change may enable Lertap to run at more non-English sites.

17 December 2003

Season’s Greetings: may your holiday stocking be filled with favourite foods and drinks, and good health be with you and yours throughout the new year.  No new Lertap mods are expected before February 2004 at the earliest.  Thanks to all for feedback during 2003.

18 November

Bug. Fae Mellichamp, of Professional Testing, Inc. (Tallahassee, Florida) found an error in the way Lertap and Excel process long *col lines.  It turns out that there is an effective limit of about 255 characters which can be used in a *col line.  While this is still a respectable number of characters, she found a need to create a *col line with almost 400 characters – this caused Excel to display an “overun” error.  (A fix for this problem is now available -- please refer to 22 June 2004 comments above.)

17 November

Delivered version 5.3 (@5November2003) to Assessment Systems Corporation (www.assess.com).

5 November

Lertap’s toolbar was given a modest enhancement.  Now there’s a new drop-down menu, “Shorts”, which makes it possible to quickly change some standard Excel settings, such as the referencing system in use, whether or not a worksheet’s row and column headers are displayed, and whether or not gridlines are displayed on a worksheet.

Of course it’s possible to change these settings, and many more, by using Excel’s Tools / Options menus and tabs – “Shorts” just makes quicker work of the job.

30 October

A substantial enhancement, quintile tables and plots, was added to Lertap’s cognitive test analysis capabilities.  This is a significant leap forward; read about it by working through the Revisions topic in Lelp, the help file.

15 October

The Student version has been brought up to date.  It now has fewer shackles: the former version limited the number of data records, number of items, and number of subtests.  The new version only limits the number of data records (to 20).  Click here to read about how to obtain this version.

2 October 2003

Lertap will now compute SMC values, squared multiple correlations.  And, with the help of a free Excel Add-In from the Foxes Group in Europe, it will also compute eigenvalues.  Details may most easily be found by working through the Revisions topic in Lelp, the help file.

Changed the Windows version number from 5.25 to 5.3.

29 September

Bug. Repaired an error on the Run menu’s options.  If an external criterion analysis was undertaken, any subsequent request to output an item scores matrix would not produce the expected results.

25 September

The Spreader is now programmed to make two passes instead of one.  In the first pass it compares the length of the string to be spread with the length of the string in the previous row.  If there’s a difference notice is given, and a chance to edit the data is provided.

It’s possible to reset The Spreader so that it does what it did before: just make a single pass.

For more information, refer to the Revisions topic in Lelp, the help file.

22 September

The Move+ menu now sports a new option: “Copy a Data column to a DAT worksheet”.  This allows data fields to be added to a workbook’s DAT file.  If interested in reading more, follow the links given under 14 September immediately below.

15 September

“ToHalve&Hold” option added under the Run menu.  This allows a data set to be randomly split into halves, useful for those thinking about calibration and validation samples.  (For more information, refer to the Revisions topic in Lelp, the help file – make sure you’re looking at an up-to-date copy of Lelp.)

14 September

Tetrachoric correlations are now computed, when conditions are right – they’re found towards the bottom of the IStats worksheet.

Also, when conditions are right, Lertap now creates a new worksheet, “DAT”, having item scores recorded in a simple text format, a format suitable for direct input to other programs, such as Bilog.

Read about these changes in Lelp, the help file (search for tetrachorics, or go to the Revisions topic under R&R&R&R).

26 August

With reference to the calculation of cognitive item difficulty, as seen below at 9 July 2003, changed the default calculation when users have activated the CFC (correction for chance) option.  When CFC is active, a cognitive item’s difficulty is defined as the number of people selecting one of the item’s correct answers, divided by N, the total number of test takers.

By “correct answer” is meant an answer which has a positive non-zero weight, that is, an answer which awards points, or fractions of points.

20 August 2003

1

Minor corrections made so that the default missing data assignment for affective items functioned as intended with *mws cards having this format:

      *mws call, *,*,5,4,3,2,1

(This format is very rarely used.)

2

Bug. Affective items with a negative mean were found to cause the Stats f report to fail.  This was repaired.

3

Changed the default missing data assignment for affective items so that it equals the mean of the item response weights.

12 August

Delivered version 5.25 to Assessment Systems Corporation (www.assess.com).

4 August

Made it possible to recalculate the score correlations seen at the bottom of the Scores worksheet.  Now columns can be deleted from the Scores worksheet, and the correlations adjusted accordingly.

This option is available under the Move menu, now named Move+.

31 July

1

Repaired and improved the operation of the advanced toolbar (refer to the Revisions topic in Lelp, the help file.)

2

Enhanced the power of the *mws card so that it can poke weights into the “other” category of an item’s scoring.  An example is shown here:

      *mws c8, 1, 1, 1, 1, other=1

This line will give one point to anyone selecting the item’s four options, and one point to anyone else – in fact, everyone will get a point for this item, even if they don’t answer it (useful when an item is simply to be “credited” to all test takers).

      *mws call, 1, 2, 3, 4, 5, other=1

The line above might be used with affective items; it says that all of the subtest’s items will award one point if a person’s response falls into the “other” category.

Refer to Lelp, the help file, for a discussion of the “other” category – see the Revisions topic as a place to start.

3

Updated the Windows version number to 5.25.

21 July

Version 5.2 delivered to Assessment Systems Corporation (www.assess.com).

20 July

Gave Lertap a bit more smarts.  When the Lertap5.xls file is opened, it checks to see which Excel referencing system was last in use.  Lertap then pokes Excel into R1C1 mode, and later, when Lertap is closed, it resets Excel to A1 referencing if that’s what had been in use.  Thanks to Nathan Carr of California State University, Fullerton, for prompting this change.

16 July

Brought in an example formerly used with Lertap 2, showing how the system may be applied to assess parallel-forms reliability.  (Word file, 16 pages.)

10 July 2003

Bug. Fixed bugs in the routine responsible for setting up Excel / ToolPak histograms.  This routine routinely failed when users requested a histogram of percentage scores.

9 July

Changed the way Lertap computes the difficulty index for cognitive items, as suggested by Nathan Carr.  Most users will not notice any difference until they have cognitive items wherein points are given for more than one of the item’s responses.  For details, see the Revisions topic in the online help file.

8 July

Bug. Nathan Carr uncovered an untrapped error in the module responsible for estimating error variance.  At times this estimate may be negative, but Lertap had not been advised of this possibility, and would crash.  Now there’s a message, and MS(error) is set to zero for this special case.

27 June 2003

Improved the capabilities of the *mws and *alt CCs “cards” so that they may be used to quickly remove items from a subtest.  (See the Revisions section of the online help file.)

3 June

Added a new reference document showing how Lertap 5 may be used in a concurrent validity study.  (Word file, 10 pages.)

30 May

Installed links so that the online help system may be called directly from the Windows vesrion.  These links are referred to as “Lelp” (for Lertap Help).

21 May

Updated chm and PDF versions of the online help system (see following topic).  Made a new website to display the contents of the online help system; briefly tested this site under Windows and Mac.

14 May

Made latest versions of the online help system available via the website.  There are “chm” (compiled HTMLHelp) and PDF versions currently available.  Click here to get copies.

11 April

Placed a draft copy of the forthcoming online Help system on this website.  The Windows version of the file should be completed by the end of May, 2003.  Click here to download this file (it’s large – about 1MB).

15 January

Opened up a mailing list option for users.  It’s available as a hyperlink from the website’s home page.  (Click on the home page link above.)

16 December 2002

All of the hard workers here at Santa’s Lertap workshop paused to wish everyone the best for 2003.  Season’s Greetings.

28 November

An installation help page was added.  A click here will take you to it.

26 October

Created a TestRun page with a series of steps to help users solve problems which sometimes arise when using Lertap with certain computers.

8 September

Added extra functions to the Scores output.  Now the summary statistics given towards the bottom of the Scores worksheet include the Range, IQRange (interquartile range), Skewness, and Kurtosis.

Standard Excel functions are used to calculate these statistics.  To see respective formulas, use Excel’s Help system.

5 September

Modified the System worksheet so that users may control the speed with which “The Spreader” goes about its job.

The Spreader allows a single column to hold all item responses, speeding up data entry, and facilitating importing item responses from other files.  Its use is mentioned in Chapter 3 of the manual.

Prior to this modification, The Spreader paused one second as it stepped from one Data row to the next.  This gave time for users to visually confirm that it was doing what they wanted.  However, with large data sets it was too slow—the pause got in the way.  Now there is no pause at all, but it’s possible to put one back in by making a suitable change in Row 13 of the System worksheet.

The System worksheet is normally hidden.  Chapter 10 of the manual has instructions on how to access and use it.

22 August

Changed the “insufficient data message” so that it covers the case of no data in the first column of the first row of the Data worksheet.

There should always be something in the first column of every Data row.  Lertap thinks it’s come to the last data record when it encounters a row having nothing in its first column.

This change was made in response to feedback from Matthew Webb of Curtin University’s Health Sciences Division.

15 August

Modified the System worksheet so that it allows the use of experimental features to be set by users.  The default setting is “No”, do not use the experimental features.

The System worksheet is normally hidden.  Chapter 10 of the manual has instructions on how to access and use it.

To branch to the Experimental Features page, click here.

13 August

Support for some experimental calculations was added.  These are for internal use only as they relate to an in-house research project, “BABstats”, at Curtin University.

12 August 2002

A sample dataset sent by Nicolaas Claasen, from Pretoria’s Human Science Research Council, uncovered an error in the processing of *mws lines.  The error occurred only when the MDO control word was present on a *sub line, and only when the *mws line made use of asterisks, as in the following example:

*sub Affective, Res=(1,2,3,4,5,6,7), MDO

*mws c5, 1,2,3,4,*,*,*

These CCs lines tell Lertap to turn off the missing data scoring option for affective items (see MDO notes in Chapters 2, 6, and 10 of the manual); they also tell the program that the item whose responses are found in column 5 of the data worksheet do not make use of the last three response codes seen in the Res=() declaration.

*mws lines with asterisks are not used often—this error is unlikely to have affected many users, and has now been corrected.

29 July

Tim Bothell of Brigham Young University had a cognitive test wherein some items used five response codes, A through E, while others used ten response codes, A through J.

He used Res=(A,B,C,D,E,F,G,H,I,J) on his *SUB line in the CCs sheet.

He then wanted to tell Lertap that some items used only the first five of these response codes, whereas others used all ten.  He thought he should use a *ALT line in his CCs worksheet to accomplish this.  (The *ALT line is mentioned in Chapter 5 of the manual.)

He was right, but found Lertap not up to the task.  There was no way to say “10” on the *ALT line.

Now there is.  X equals 10.  If you have a case like Tim’s, use an X on the *ALT line to indicate an item uses 10 response codes.  The following example says that the first five items use six response codes, while the last five use ten response codes:

*Alt 66666 XXXXX

26 June 2002

It’s a Go at Assessment Systems Corporation (!).  All versions are now available from the Lertap area of the ASC site.  The whole manual is available for free from ASC, as are the Student and 30-Day versions.

Beta test sites: you will have to register with ASC in order for your Lertap use to continue.  No charge – send an email to Larry Nelson for comments on how to register with ASC.

24 June

Made a trivial update to one of the message boxes; there was a small inaccuracy related to the instructions found in the commercial version regarding the steps required to obtain a license.  This was spotted by Dave Weiss of Assessment Systems Corporation, St. Paul, Minnesota.

Note: it is expected that the commercial version will become available before the end of June, that is, in just a few days.

20 June

Added another instructor’s aid: a Word file with 15 answer sheets, as returned in a high school chemistry class.  Gives the chance for students to practise setting up data in Excel, and using Lertap to get results.  Suitable for use with the Student Version (or any other version).

20 June

A Student Version has been developed, and will be freely available soon from Assessment Systems (www.assess.com).  It is hobbled, able to process a limited number of items (20), and a restricted number of data records (20), but it’s free.  Instructors may find it useful.  Would-be users who want to quickly take the system for some test drives should like it.

18 June

Included a link to a brief new page which makes reference to some IRT and CTT comparisons.  A click here will take you to it.

17 June

Made various minute cosmetic changes to the system, largely in preparation for the coming launch at Assessment Systems.

3 June

Added an instructor’s aid: a complete sample data set, with responses from 12 principals to a 1983 survey concerning the use of computers in primary schools.

This data set contains the actual questionnaires, as returned by the principals.  Users may print the 12 returns, and enter results in Lertap, comparing their results with the “true” ones provided.  Suitable for use with the Student Version (or any other version).

Click here to branch there.

28 May

Updated the version available for download from the Curtin University website:

www.lertap.curtin.edu.au

This version has been named the “Standard” version.  It should run on both Windows/NT and Macintosh machines, but only for a limited term: sixty days or so.  The number of days left to expiry is now seen by clicking on the smiling yellow face on Lertap’s toolbar.

Users should refer to the KnownProblems page for details regarding problems which may arise when using this system.

It must also be noted that the Standard version is a use-at-your-own-risk version; there are no guarantees at all regarding the performance of the Standard version.

The Standard version has all of the features found in the commercial versions mentioned below – it’s a freebie which (1) will not run forever, and (2), may be withdrawn at anytime.  (When the commercial versions are available, it will be possible to download them on a 30-day trial basis from www.assess.com.)

15 May

Two new commercial versions are just about ready, one for Windows and NT users, another for Macintosh people.  These versions do not have any new features, but they have had a bit of fine tuning – for example, some of the user messages have been improved, and it seems the problems with Mac OS/9 and Mac OS/10 may have been resolved.

6 May

Problems have been noted when working with large data sets.  When there are more than 4000 records in the Data worksheet, we have found that Excel will not always respond to Lertap’s request to “create a new data set which is a copy of the present one”.  We have also found that external criterion runs may bomb out with data sets when n > 4000.  We have work-arounds to suggest for users – write to us at the email address above.

Note added 10 June 2002: these two problems do not arise when using the latest version of Excel for Windows, Excel 10 (this is the version which comes with Office XP; we tested Excel 10, running under Windows XP, with one of the large data sets from la Universidad Central de Venezuela, n=11,190, and had no problems at all).

Note added 8 July 2002: two of us visited central Java last week to work with that province’s education department.  We saw a data set with more than 20 000 (twenty thousand) records successfully processed under Windows 98, using Excel 9 (better known as Excel 2000).

Feb - April

Principal author on medical leave (a quiet time for all).

25 January 2002

Refreshed versions of the Lertap5.xls and Lertap5.zip files were placed on this website, good for 165 days (valid through early July, 2002).  After this time Lertap will be available on a commercial basis from Assessment Systems Corporation, www.assess.com.

12 December 2001

Tim Bothell of Brigham Young University sent a data set with an Istats worksheet included (this worksheet is produced by using the last option in the Run menu).  It displayed several ugly #DIV/0! values.  On examination it was found that these unsightly values corresponded to items having no variance (all students selected the same answer to the item).  This problem was fixed.  The #DIV/0! values have been replaced by 0.00.

It’s once again time to send Season’s Greetings to all.  This year has been an extraordinary one—along with others the world over, Australians sent their condolences to families affected by September events in New York and Washington D.C.

3 October

Corrected an error found by Antoinette Marais, a doctoral student at the University of Alberta: subtests with only one item caused the variance components routine to crash, producing a “runtime error 6” code in Excel.

27 August

Changed the website back to a Curtin University server.  The URL is now:

www.lertap.curtin.edu.au

The WebCentral server mentioned below supports FrontPage extensions, something we have used to demonstrate the efficient use of on-line data collection forms.  The Curtin server does not support FrontPage, and, consequently, a small part of the Lertap website will no longer work as originally designed.  (So why the switch to the Curtin server?  Local support.)

25 July

Changed Lertap’s web server from Curtin University (Perth, Western Australia), to WebCentral (Sydney).  See www.webcentral.com.  The main URL remains www.lertap.com, of course.

This change should improve service to overseas visitors.

1 July

Fixed the routine which makes a blank new workbook so that the rows in the new CCs worksheet all have the same height of 15 points.  (Before row 12 had a disproportionate height as it copied row 12 in the original CCs sheet, a row with a height of some 45 points.)

27 June

Improved the code which handles moving (copying) a column from the Data worksheet to the Scores worksheet.  Now a message appears on the status line whenever the process consumes more than a few seconds (there was no message before, and users could not be sure that Lertap was really at work).

26 June

Re-installed the code which automatically activates Excel’s R1C1 referencing system whenever Lertap starts up (the single line of code which does this had been inadvertently deactivated).

2 June 2001

1

Installed an error trap for *col cards which are missing an appropriate number of C’s.  For example, the following card is missing a C before the number 23:

*col (c2 – 23)

This change was made in response to feedback from Leli Mustiarasari, an Indonesian postgraduate student.

2

Added comments to the Lertap5.xls so that the control cards seen in the CCs worksheet might be more readily understood by beginners.

25 May 2001

Made very (very) minor changes, such as updating the email addresses shown in various screens (current best address is larry@lertap.com).

18 April

Made it possible to download an unzipped copy of the prototype from the website.

9 April

Added code to the macros called by the options on the New menu so that the Runtime error 9 message would not appear when new workbooks are set up. (The cause of this error is not known, but it’s been fairly frequent.)

12 March

Improved the diagnostic message which covers the case of all missing data, that is, the error notice that is produced when a data column is found to be empty over all rows of the Data worksheet. This is generally caused when the *col card points to too many columns.

5 March

1

Made it possible for the Data sheet to have asterisks (*) without causing the response frequency counter to stop every time it found one. Now asterisks are processed as blanks, and their presence is included as part of the count of ‘other’ responses (the count of ‘other’ responses is shown as ? in the Freqs worksheet).

This change was made in response to feedback from Carlos Gonzalez at la Universidad Central de Venezuela. The output from scanners will often include asterisks to indicate that a scan sheet has an invalid response, such as someone shading in multiple item responses. Carlos processes very large data sets, often with as many as 10,000 cases, and having Lertap stop every time it found a * was quite inconvenient.

When it finishes its frequency tally, Lertap now reports the number of asterisks encountered in the Data sheet (if any).

2

Improved the information given when Lertap reads a cell which has more than one character in it. For example, a cell meant to contain a 3 may, instead, have 43 (for some reason).

If the cell’s contents are not changed, Lertap will (1) tally the cell in the ? row of the Freqs sheet (which is probably not bad), and (2) strip off the first character in the cell, and consider it to be the answer to the respective item for purposes of test scoring and item analysis (which is probably not good, but how is Lertap to know what the real response is?).

This change was made after reviewing test results processed by Wally Bzdell of Boston University.

19 February

Made very minor changes to the Syntax worksheet (one of the four worksheets contained in the Lertap5.xls workbook). It now contains a reference to the book, suggesting that users might want to order the book should they require more details.

15 January 2001

Updated the lertap5.zip file on the website so that it’s good for 180 days from the date of the last system generation (or update). Scroll to Row 100 of the Comments worksheet in Lertap5.xls to see the current expiration date.

20 December 2000

Placed a fresh copy of the prototype’s Zip file on the website, good for 90 days instead of the usual 60.  Season’s greetings to all—best wishes for 2001.

12 December

Corrected Elmillon so that an affective subtest having just one item and no variance will process okay, without raising an error message from Excel.

7 December

Changed the declaration of four variables used in the ScatterPlot routine from integer to double precision (to circumvent overflow problems which occurred when using very large numbers, such as age expressed in number of days since birth).

6 December

Fixed an error which had to do with Data cells containing blanks, or spaces.  The system was found to be halting on encountering these, pointing out that processing could not continue until the cell’s contents were fixed.  Now the system rolls along, as it should (and used to), counting blanks as “others”.

Updated Zip file placed on the website.

27 November

First version of the book, “Item Analysis for Tests and Surveys Using Lertap 5”, delivered to printer for processing of 200 initial copies.

24 November

Very minor modifications made to the prototype; new Lertap5.zip file available on the website.

23 November

Updated Chapter 9, Lertap, Excel, and SPSS to website.  More updates to Chapter 8 also.

22 November

Updated Chapter 8, interpreting affective test results, placed on the website.  New version of Chapter 1 also.  Minor updates to Chapter 10 too.

Zip files of the three largest chapters made available via the website.

21 November

Found and fixed an error which occurred when running an external criterion analysis, with the selected subtest (the one whose items are to be correlated with the external criterion) being an affective subtest—Lertap tried to create a U-L report, which was not on, and caused a freeze.  This problem reflects the fact that this was the first run of an affective test with an external criterion since the U-L report routine was added.

16 November

Placed first draft of Chapter 1, Lertap 5 Alive, on the website.

15 November

1

Found and fixed an unexpected error: if a *col card had no hyphens, Lertap would fail.  For example, it choked on this card:

*col (c16,c30,c36)

2

Placed an initial draft of Chapter 8, interpreting affective test results, on the website.

14 November

Continued work on documentation.  Chapter 7 now thought to be finished, as is Chapter 2.

Appendix A, the original Lertap 2 Quiz from 1973, now ready (added to the website).

13 November

Added the latest draft of Chapter 9, “Lertap, Excel, SPSS, and such” to the website (dated 10 October).

Began sending chapters to proof reader.  At this stage, only Chapters 2, 7, and 10 are considered close to being at final draft stage. 

10 November

Added the first draft of Chapter 7, interpreting cognitive test results, to the website.

Cleaned up the website a bit.  Dates after chapters now indicate when the chapter was last updated.  All of the chapters are still drafts, and most have not yet been proofread.

8 November

Placed a new version on the website, from where it may be downloaded.  This version has correct Option Base 1 statements in all modules, hopefully addressing the bug notice immediately below.

7 November

Potential bug notice:

Some sites, such as Proinvesca in Caracas, and Burapha University in Thailand, report being unable to use the version of 2 November, placed on the website 6 November.  The system gets started, but then fails with a “subscript out of range” message, runtime error 9.

Made sure Option Base 1 statements were present in all Visual Basic modules (some modules were lacking this statement), and recompiled.

Will this solve the “subscript out of range” errors?  In theory it should, but it raises the question: why did the 2 Nov system work at some sites, and not others?  The Option Base 1 error was one which should have kept it from working anywhere (in theory).

6 November

Made it possible to download the prototype, as a Zip file, from the website:

http://education.curtin.edu.au/systems/

(This site may also be accessed by going to www.lertap.com.)

2 November

Added Peng-Subkoviak estimates for the proportion of consistent decisions made in mastery testing situations (their p₀ and kappa estimates).  These are displayed on the Stats1ul sheet, after the B-K statistics mentioned below.

This enhancement was another suggested by staff at the University of Alberta.

1 November

Improved the B-K output so that error variance is displayed, along with a value for calculating the 68% confidence interval.

31 October 2000

Added the Brennan-Kane (B-K) index of dependability to the end of the Stats1ul report.  (This is a type of reliability indicator for mastery tests—see Brennan & Kane in the references.)

Lertap uses the general variance-components method for calculating this index; items are not assumed to be dichotomously scored.

Thanks to Todd, Keith, and Tess at Alberta for suggesting this addition.

30 October

Improved the error message given when keyed-correct answers are found to not match with a subtest’s Res=() declaration.

27 October

Changed the U-L statistics for the “mastery” case so that percentages are shown instead of frequencies.  This change was suggested by staff at Alberta.

Installed a comment in cell R1C1 of the Stats1ul sheet which indicates the number of people in each group.  Excel uses a small red triangle to flag cells which have comments—to see the comments, simply let the mouse pointer hover over the cell.

26 October

1

Updated Chapter 2 with new text on the biserial correlation calculations.

2

Changed Chapter 3 so that its description of how the Spreader operates is up to date (see note at 12 October below).

25 October

Updated Chapter 10 with new text on the biserial correlation calculations; also added a small section on correction for chance scoring.

24 October

Added another discrimination index, the biserial correlation coefficient, to the full stats report (Stats1f).  It appears as r(b).  This addition was suggested by Todd Rogers, Alberta.  (Lertap 2 was the last version to include biserial coefficients; they were not seen in Lertap 3, nor 4.)

23 October

1

Problems with item difficulty calculations were reported from Alberta.  Staff there found that an item with more than one right answer could have more than one entry in the item difficulty bands which appear at the end of the Stats1f sheet.  This was fixed.

At the same time, another change was made: the “diff” index on the brief stats reports, Stats1b, used to equal an item’s mean—now it’s the item’s difficulty, where difficulty is defined as the number of people who got some points for their response, divided (of course) by the total number of people.

2

Updated Chapter 10 to reflect the changed item difficulty calculations.

20 October

Ed van den Berg relayed a problem from San Carlos University—a user had a Data worksheet with data starting in column 2 of each record.  Column 1 was empty.

Lertap thinks an empty first column means it’s come to the end of the data, so this user got no results, and wondered why.

An error message was installed to catch this type of omission.

18 October

Repaired the bug shown below at 14 October 2.

17 October

1

Fixed an error in the histogrammer: if the Scores sheet was missing data in a record’s ID field (the first column of the Scores sheet), the histogrammer would fail, saying that it could not “find a row with ‘n’”.  It’s not good to have empty ID fields, and Lertap will still stop processing if it finds a record with an empty ID—now, however, the error message given by Lertap is more accurate.

2

Repaired the bug below (George’s bug).

14 October

1

Known bug notice:

When a subtest has fewer than 3 items, the summary statistics which appear at the end of the full statistics reports, such as Stats1f, will be incorrect in the area of the alpha figures.  With prototypes dated before 14 October, this bug crashes the system with a “division by zero” error.

George Kretke, Boulder Valley Schools (Colorado), found this bug.

Fixed 17 October.

2

Known bug notice:

When an mws card such as the following is used with cognitive subtests, the upper-lower report, such as that seen in Stats1ul, will show correct results, but in the wrong columns.

mws c5, *, *, 1, 0

About this card: it says that the first two options in a subtest’s Res=() specification are not used by the item whose responses are found in column 5 of the Data worksheet (this form of the *mws card is not often used).

Fixed 18 October.

12 October

Changes made to the operation of The Spreader.

The Spreader will now walk down the rows of the Data sheet, spreading stuff to the right automatically, row by row.  It still checks to make sure that it’s not going to overwrite anything.

To stop The Spreader, press the Esc key at any time.  To restore the Status Bar so that it says “Ready”, click on the yellow smiley face on Lertap’s toolbar, then on the OK button.

Also, The Spreader is now deactivated when the user is in one of the first two rows of the Data sheet.  This prevents titles and headers from accidentally being spread.

Thanks to Ed van den Berg, San Carlos University, for help on this feature.

10 October

1

Made very minor change to the XL4 exporting routine, accessed via the 8-ball on Lertap’s toolbar, so that it will work when the Data worksheet’s name, Data, is all lower case (that is, “data” instead of Data).

2

Added a new version of Chapter 10, Computational Methods, to the website (use link in following cell to jump there).

4 October

Added Chapters 5 & 6 to the website.

Click here to jump to the web page.

3 October

1

Fixed errors in the computation of the Upper-Lower summary group statistics.  Errors had been occurring for both internal and external criteria statistics.

2

Fixed the brief stats display for cognitive subtests so that all responses with weights greater than zero are underlined (it was the case that only one response would be underlined, no matter how many “right” answers an item might have).

3

If a *sub card has a title=() declaration, but no name=(), now the subtest’s name will be the same as its title.

26-29 September 2000

Updated the website with new drafts of several chapters from the coming (new) Guide to Lertap 5 Use and Interpretation.

Click here to jump to the web page.

25 September

Found and fixed an error which could arise when a cognitive subtest was processed with no *sub card.  A programming error was producing an incorrect message, and turning off the subtest’s U-L item analysis.

19 September

1

Fixed an error in the histogram routine which produces the Excel chart.

2

Fixed an error in the main program which halted processing during the computation of total test scores (this error was created by faulty wiring in the new U-L item analysis module).

3

Added to the Computational Methods document.  It now has a table of contents, and a new section with data from time trials.

4

Added information on the “Call” word now available on *mws cards to two documents: Control Cards for Cognitive Subtests, and Control Cards for Affective Subtests.

The “Call” word for *mws cards is mentioned below at 14 September.

15 September

1

Added to the Computational Methods and References documents.

The Computational Methods document now includes a section on the new Upper-Lower item analysis routine.

2

Prepared to watch the opening ceremony for the Sydney 2000 Olympics.  Surely Lertap users all around the world will be doing the same (forcing themselves away from their computers, dragging themselves to local stores to ensure an adequate supply of snacks and beverages, no doubt keeping a copy of Lertap documentation at hand to read during the commercials).

14 September

1

Fixed an error which stopped the program from processing results when the number of data records was greater than 2000.

2

Made it possible to use a card such as the following:

*mws Call, -3, -2, -1, 0, 1, 2, 3

The change here is in the ability to use the word “all” as a column indicator on the *mws card.  This means that the new weights will be applied to all the items in the subtest.

A card such as the above would be useful, for example, when found after this sub card:

sub aff, res=(1,2,3,4,5,6,7)

In this case, a response of “1” to any item would get a weight of –3, while a response of “7” would get a weight of 3.

12 September

A new U-L (Upper-Lower) item analysis routine was brought online.  This is an addition to the traditional correlation-based methods seen in the other item analysis sheets, such as those named “Stats1f” and “Stats1b” (the full and brief item statistics displays).

Now a “Stats1ul” summary is created for every cognitive test, with item performance data based on a comparison of two groups: the upper 27% of the test takers, and the lower 27% of the test takers.

This analysis may be kicked into a “mastery” mode by including the word mastery on a * sub card, such as seen in the following example:

sub name=(Sem 1 test), title=(Sem1), mastery

A card such as this one will force the U-L routine to define two groups, an upper group, the “Masters”, corresponding to those having percentage scores of 70 or above, and a lower group, the “Others”, corresponding to everyone else.  The mastery level may be changed to, for example, 80 by using a card such as the following:

sub name=(Sem 1 test), title=(Sem1), mastery=80

7 September

1

Minor change to the brief item stats display, and also to the item response charts made from it.

2

Started design of new upper-lower / mastery—non-mastery item stats sheet for cognitive subtests.

Added parameter to the Systems sheet at cell (10,2) which controls whether or not this sheet will be standard output.  Default setting is Yes.

6 September

1

Made additions to the Computation Methods document which explain the functioning of Lertap’s hidden “Systems” worksheet, and how to change the user level so that the Liberty Bell icon is made available on the toolbar.  These new sections are at the end of the document.  (The Liberty Bell is mentioned below: see comments under 12 April.  The quickest way to get to the Computational Methods document is via the hyperlink at 30 August below.)

2

Further investigated the bug with Excel 97 charts, first mentioned at 12 July below.

The problem has been acknowledged by Microsoft, and it’s possible to fix it if users have SR-2 of Excel 97 installed (this can be checked by using Excel’s Help / About Microsoft Excel option).  The fix is very effective, but requires users to modify the registry on their computers.  If you’d like details, click here.

5 September

1

Fine tuning of the new scatterplot routine, with more tests passed under Excel 97 and Excel 2000, Windows 98 and NT4.

2

Changed the data set which comes with the main program file, Lertap5.xls.  Now the Data and CCs sheets correspond to the Lertap Quiz data set which was originally distributed with Lertap 2.  This data set was also sent out with Lertap 3, where it was known as “QuizData”.

1 September

Encountered a bug when testing the new histogrammer under Excel 97 on a student shared computer running Windows 98.  Installed an error catch and message.  (Seems to be a problem related to file sharing.)

31 August 2000

Added a scatterplotter.  Tell Lertap which two columns of the Scores sheet you’re interested in, and Lertap will ask Excel to make a chart containing a scatterplot of the two respective variables.  Excel scatterplots, like its histograms, are quite well formatted, and in many respects are easier for users to format than are the equivalents in SPSS—once made, these scatterplots (like the histograms) may be tailored to individual needs—right click on the chart itself, or double left click on things like titles and score labels.

Tested under Excel 97 and Excel 2000 (in fact, this routine was made under 97 and then imported to 2000).

30 August

1

Minor changes to the operation of the histogrammer, such as displaying new messages in the status bar area.

2

Updated the Computational Methods document with a new section on “Histograms” (it appears towards the end of the document).  Click here to access this document; if asked for a password, click Cancel.

29 August

Completed initial work on a new histogrammer.  If the user has the “Analysis ToolPak” add-in installed, then a dinkum histogram is made, having all the properties of an Excel chart.  This histogram is in addition to the former Lertap2-style histogram, and seems a worthy, colourful, addition.

Lertap2-style histograms are now in worksheets with names such as Histo1L, Histo2L, while the new fancy Excel histograms are in worksheets named Histo1E, Histo2E, and so on.

Tested with Excel 97 and Excel 2000.

How to install the Analysis ToolPak add-in?  See Excel’s on-line help.  Note that there seem to be two add-ins which begin with the name “Analysis ToolPak”.  Ignore the one which includes “VBA” in its name.

24 August

Fixed the Excel4 file converter so that it also works with the new IStats worksheet (see 21 August).  Tested exporting to SPSS.

23 August

Finished a new document, “Computational Methods Used in Lertap 5”, and made it available via the web site.  (Click here—if asked for a password, click Cancel.)

22 August

Changed the labels on some of the item statistics sections.  Now it is more apparent where the point-biserial coefficients are.  (Columns which were headed with just “r” before now have “pbI” instead.)

21 August

Created a new routine which creates an item scores and statistics matrix for any given subtest.  This matrix is on a new worksheet called “IStats”.

The statistics include item intercorrelations, as well as the average item intercorrelations (useful in factor analysis, as in SPSS).

20 August

Completed routine which allows columns to be copied from the Data worksheet to the Scores worksheet, and vice versa.

17 August

Installed the “External criterion analysis”.  Now it is possible to correlate all items of any subtest with an ‘external criterion’.  The external criterion is a score, any score, taken from the Scores worksheet.

This new routine is activated via the Run menu on Lertap’s toolbar, where a new option has now been added.  Works with both affective and cognitive subtests.

14 August

1

Improved the calculation of a cognitive item’s dif (difficulty) and disc (discrimination) values so that they are correct even when the item has more than one correct answer (more exactly: when more than one response has a weight greater than zero).  Disc values are now allowed to show as negative values; prior to this negative Disc figures were changed so as to show as zero.

In previous versions of Lertap the dif and disc values were not correct when an item had more than one correct answer.  (Mention of this was made in the manuals.)  This limitation has now been addressed in Lertap 5.

2

Corrected an error made at 8 August when the calculation of standard deviation values was changed.  The error had to do with incorrect signs for the point-biserial correlations seen in full stats sheets (such as Stats1f).  Note: in Lertap5 these correlations are simply denoted as “r”; in previous versions this “r” was labelled pt-bis.

3

Added the new routine for calculating alpha values, as described below (11 August), for cognitive subtests.

11 August

1

Added a new routine to calculate coefficient alpha values if an item were to be removed from an affective subtest.  This information now appears at the end of the full stats sheet (for example, at the end of Stats1f).  It makes it possible to very quickly see what a subtest’s alpha value would be if any particular item were missing from the subtest (or deleted, as SPSS terms it).

The same facility should be added to cognitive subtests within a few days.  (Done; added 14 August.)

2

Checked new alpha calculations (described above) with SPSS figures for the same data set.  When doing this, it’s important to realise that Lertap and SPSS have different ways of handling missing data.  Lertap and SPSS results are likely to differ if a data set has missing data.  Another point to note, in this regard, is that the two systems do not calculate variance and standard deviation figures in the same way.  Lertap values will be slightly smaller than SPSS equivalents when the number of cases is relatively small (less than 30, for example)—this is because Lertap divides the sum of squared deviations by “n”, not “n-1”.  This is done in keeping with the general philosophy of Lertap: it’s a descriptive package, not an inferential one.  Using “n” is correct when one wishes to describe the data on hand; “n-1” should be used in those cases where one says “I’ve got a sample from some large population, and wish to have an estimate of the population’s variance, or standard deviation”. 

3

Began to update the References page.  (Click here to link to it.)

10 August

Corrected the computation of the item-criterion correlation coefficient.  This needed to be done after the change in s.d. calculation mentioned below, under 8 August.  Re-checked all calculations for both cognitive and affective items by going outside of LERTAP.  The computational error, now corrected, was of a magnitude of n/(n-1), where n is the number of data records.  For a data set with n=30, for example, item-criterion correlation coefficients were 1.03 times higher than the correct value.  With n=100, they were 1.01 times higher.

Note: by item-criterion correlation is meant the correlation between an item and the criterion score.  The usual case is to have the subtest score serve as the criterion (in which case the analysis is said to use an internal criterion)—for the “usual case”, then, the correlation referred to here is under the “r” column in the full stats sheet (such as Stats1f), or the “disc.” Column in the brief stats sheet for cognitive items, which becomes the “cor.” Column for affective brief stats.

8 August

1

Added a new control word, “scale”, for use on the *sub card.  For cognitive subtests, the result is the computation of a z-score for each test taker.  For affective instruments, the result is a scaled score equal to the original raw score divided by the number of items, something which could also be called a normalised score.  In both cases, cognitive and affective, the new scaled score appears immediately to the right of the original score in the Scores worksheet.

Examples of use:

*sub name=(a new exam), title=(newx), scale

*sub aff, scale, title=(att1), name=(attitude 1)

This word will be useful in numerous affective surveys, such as the MSLQ, the Motivated Strategies for Learning Questionnaire in use here at Curtin, and the CAQ, the Computer Attitudes Questionnaire in use at Burapha University and the University of North Texas.  Both of these instruments commonly summarise results by normalising scale scores.

2

Changed the computation of the standard deviation figure so that it is a “population” statistic, not a “sample” statistic.  Prior to this change most s.d. figures seen in the worksheets were sample values—the new s.d. values will be slightly lower than the old ones when the number of data records is less than, say, 100.  The difference between the two values can be seen in the full stats sheets under Summary Statistics, where both s.d. statistics are displayed (look, for example, at the Stats1f worksheet).

If users are comparing output from the new version with that from previous versions, it should be noted that both Lertap 2 and Lertap 3 used sample s.d. figures for most internal calculations.  Lertap 3 differed from Lertap 2 in that it displayed the population value as the s.d. for subtests.

3

Updated documents on the Web site to reflect the new “scale” control word (the two documents mentioned under 25 July).

7 August

1

Registered as an Internet domain.  The URL www.lertap.com maps to http://education.curtin.edu.au/systems/ and the email address larry@lertap.com routes to l.nelson@curtin.edu.au.

2

Placed hyperlinks to all user documents on the Web site.

25 July 2000

Created two new documents, Lertap 5 Control Cards for Cognitive Subtests, and Lertap 5 Control Cards for Affective Subtests.  (These may be seen via the Web site mentioned above.)

23 July

1

Fixed problem encountered if the res=() specification on a *sub card has a zero as the first character.  Res=(0,1), for example, now seems to be okay.

2

Fixed problem related to #NULL! Values found in spreadsheets exported from SPSS 9.0.  These no longer cause a system crash—however, execution is suspended.

3

Improved the handling of bad cell data so that the user is told which cell(s) contains questionable data, and given the chance to fix the cell on the spot (processing is terminated).

21 July

1

Minor modification to the message given whenever a Data cell is found with questionable contents.

2

Addition of another syntax example to the sheet mentioned at 18 July, this one covering use of the *tst card.

20 July

Continued to add more error messages designed to catch common mistakes in Lertap’s control cards.

19 July

Put in more explicit diagnostic error messages related to parsing the *key, *pol, *alt, and *wgs cards.  These are in response to feedback from Piet Abik, Pontianak, Indonesia.

18 July

Added a new sheet with control card syntax examples to the Lertap5.xls workbook.  Now users may remind themselves of some syntax style on line, without having to refer to printed documentation.

16 July

Minor change in Stats1f display—the value of alpha is now selected so that it’s the first thing seen when a full stats sheet is activated.

15 July

1

Corrected an error which occurred whenever an affective subtest had just a single item—if this subtest was the last in a series of affective subtests, all item correlations in all subtests were said by Lertap to be 1.00.  This problem was uncovered whilst processing survey results collected by Nanta Palitawanot in Thailand.

2

Minor alteration to charts stats display for cognitive items—see 8 July (1)—the display mentioned is now right-justified so as to line up better with the items stats.

14 July

1

Fixed processing of affective items so that there’s no crash when an item has no variance.

2

Changed the Freqs display so that an item is never freaked more than once.  It had been the case that an item’s response frequencies would be tallied every time the item’s data column was referenced in a *col card—it was thus possible for an item to be freaked multiple times.

13 July

Added a speedometer in response to Luis Pulido who, along with Carlos Gonzalez, is processing data sets with thousands of cases in Caracas.  Now Lertap gives a greater indication of what it’s doing—it updates the Excel Status Bar as it reads and writes records.  Users must have View Status Bar turned on, of course.  As it turns out, the speedometer actually slows down the processing by a few seconds, just a few, but I think it’s worth it as it is reassuring to see the cases ticking over—makes the user less doubtful as to whether or not the system is working.

12 July

1

More tests with charts under Excel 97.  Could not get Excel to allow more than 47 charts per workbook (not per worksheet—47 was the maximum number which could be obtained for any workbook, no matter what the number of subtests).  Installed a better error message for this bug; a true fix seems some way off.  The problem has not yet appeared under Excel 2000, but to date Excel 2000 charts tests have been limited to data sets having only about 100 items.  (See more comments at 6 September above.)

2

Noticed a few problems running with a group of capable Thai graduate students at Burapha University: if an affective item has no variance (all respondents select the same option), the system gives an overflow error—the fix for this is already in the code brought in from Lertap 3, and needs now to be activated.

If there’s no space after the *col or *key, the system does not run correctly—this again points to the need to improve the messages given for syntax checks of the CCs lines—there are several checks currently in the code, but the messages they put out to users are inadequate.

Some students asked about a Help file, and whether one would be built in.  A Help file was started months back, but abandoned as it was felt to require too much overhead to support; having a Help file makes system installation an issue whereas at the moment it’s not (I think one of the most attractive features of the prototype is that it’s so easy to boot up—there is no install procedure whatsoever).

11 July

Installed fixes for the Regional Settings problem.  Consider these control cards:

*col (c3-c12)

*sub wt =0.5

*key BBBBD CACBB

*mws c3, .25, .5, 1, 0.75

If the Regional Settings use a comma as a decimal separator, instead of a decimal point, the Wt= assignment above would not work correctly, nor would the three decimal values seen on the *MWS card above.  Now they will.  Decimal values must still be entered as seen here (above), with a decimal point used as the decimal separator, but Lertap replaces the decimal point with the correct decimal separator, whatever it may be, as it interprets the CCs cards.

To date tested only with Spanish Regional Settings (which use a comma as the decimal separator).  If the Regional Settings are set to English there is no problem.

Tested with Excel 97 and Excel 2000.

9 July

Started to work on the Regional Settings problem.  Confirmed that the present prototype will not correctly parse *MWS cards such as the below (8 July) if the decimal separator is not a decimal point.  Note that this problem will also apply to the Wt= assignment on the *Sub card.

8 July

1

Minor formatting changes made in the display of the charts summary indices—it was the case that cognitive items with digits as response codes had their questionable distracters display as a number instead of a string.  For example, if res=(1,2,3,4,5), and distracters 2 and 4 were questionable, charts would display ?  24.00 instead of ?  24

2

Luis Pulido advises a bug when using MWS cards which attempt to assign non-integer weights, as in, for example,

MWS c5, 1.00, 0.75, 0.50, 0.25

Apparently the weights do not convert properly when Lertap passes them to Excel—but this is probably true only when the user’s Regional Settings specify that the decimal separator is something other than a decimal point.  Expected to affect beta testers in South America and Indonesia.  Must be fixed.

Fixed 11 July.

5 July

1

Added the missing data option from Lertap 2 for affective subtests, MDO.  In Lertap 2, the presence of the control word MDO on a *SUB card resulted in the weight for “other” responses being set to the middle of the scale of response weights used by the item.  For example, in Lertap 2 if Res=(1,2,3,4,5) and a person did not answer an item, a weight, or score, of 3 would be assigned if MDO was present.

Now MDO is back in Lertap 5, but its functioning is reversed.  MDO is assumed, by default, to be present for every affective subtest.  To defeat this, MDO is added somewhere on the SUB card.  For example,

sub aff, mdo, n=(Blah blah), t=(blah), res=(1,2,3,4,5,6,7)

will result in assigning all those with an “other” response a score of 0 (zero) on this subtest’s items.  If the card were

sub aff, n=(Blah blah), t=(blah), res=(1,2,3,4,5,6,7)

then all those with an “other” response would get a score of 4 (four) on this subtest’s items.

What is an “other” response?  It is said to happen whenever a respondent’s answer to an item is not a member of the response code set used by the item, that is, not found in the subtest’s res=( … ) declaration.  This typically includes missing responses.

2

Decreased the font size used in Charts.  Now it’s set to 10, the size seen in most other worksheets produced by the program.

Note: feedback is needed on the charts.  They were thought to be working well, but when tested more thoroughly with Excel 97 it was noticed that “insufficient memory” messages were encountered with subtests having more than 30 items.  This is expected to be a difficult problem to trace—the programming for Lertap 5 is done under Office 2000, and Office 97 does not have access to VBA code modules written under 2000; debugging for Excel 97 is not at all straightforward.

When tested in Thailand, with Thai versions of Windows, it has been noticed that the charts are incorrectly sized—they’re too small.  The Thai versions of Windows use a non-standard character set, and this, perhaps, is the cause.  It was noted that applying the Office 97 Language Switcher, changing from Thai back to English, cleared up this problem—the charts were then correctly sized.

30 June 2000

1

Adjusted the charts routine so that the chart cannot extend too far to the right (in some cases the chart was hiding the items stats).

2

Changed the Comments sheet so that it mentions just one Web site: education.curtin.edu.au/systems/

26 June

Installed code to ensure that the Comments sheet always displays whenever the main file, Lertap5.xls, is opened or activated.

19 June

Enhanced the charts display by adding summary stats indices to the right of each chart.  For cognitive items these indices include diff., disc., ?, and other, while for affective items they consist of pol., mean, s.d., cor., and other.  This enhancement effectively means that the charts worksheet totally replicates the data found in a brief stats worksheet—the difference is that in charts item response frequencies are displayed as bars in a 3-D graph, not numbers.

16 June

Added the charts capability, and an icon on the toolbar to activate it.  This utility takes a brief stats page, and turns it into pictures, that is, into bar charts, one for each item, which indicate item response frequencies.  Tested under Excel 2000 and Excel 97 (but see notes at 5 July).

13 June

1

Found the error in the histogrammer which created problems for users of Excel 97.  Was using VBA’s ROUND() function, which works with Excel 2000, but not with older versions.  Solution: change all calls to APPLICATION.ROUND() which forces the system to use Excel’s in-built function instead of VBA’s (VBA = Visual Basic for Applications).

2

Changed processing of the *tst card so that when used Lertap would not automatically delete all the worksheets which may have been generated by the user.  Note: the *tst card should be the first card in the CCs file.  If not then the CCs sheet in the new workbook will be missing its top line.

12 June

Incorporated a number of small-ish enhancements to improve the user interface: made Excel stop trying to save the main Lertap5.xls file (there is no need to save it); added a trap to the Run options so that they would not produce an error message when the user was looking at Lertap5.xls instead of her or his own workbook; forced Excel to recalculate after running Elmillon.

9 June

1

Corrected internal handling of dates.  Users whose Windows regional settings specified a language other than English could not run the system until they changed the regional settings.

Note added 12 July: system tested with Buddhist calendar, and found to work okay.

2

Corrected an error which occurred when an affective subtest had only one item.

3

Changed the Excel file to Lertap5.xls instead of xlt.  The file no longer works as a template, which should make its use a bit more natural to many users (?).

29 May

Added a new drop-down menu, “New”, to the toolbar, and protected the template’s workbook, and all its worksheets—now, to set up a new workbook, users work through the New drop-down menu.  Added a Comments worksheet to the template with some explanation of this.  (This is a major change to the way the program operates.  It should make it easier to use.)

25 May

As a result of processing a 55-item survey with 10 scales (subtests), changed the way the program behaves so that it: (a) no longer stops on each *COL card; (b) does not automatically delete worksheets when users choose Run / Interpret CCs lines; (c) has the Delete worksheets icon showing at the very left of the toolbar, no matter what the user level is set at (see comments at 18 April for the matter of user levels).

24 May

Added support for the *TST card.  This card is used to cull specified records from a Lertap workbook, placing them in a new data sheet in new workbook.  An example of the format is:

*TST c9=(A,B,C), c22=(M)

This card would result in a new data worksheet whose records had an A B or C in column 9, and an M in column 22.

When this card is used, it should be the first card in the CCs worksheet.  The new workbook will appear to be the same as the original; it will have a CCs worksheet identical to the original, and it will also have a Data worksheet, but with a reduced number of records.

This card first appeared in Lertap 2.  Its format has changed a bit, but it does essentially the same job, the major difference being that the new card results in a new workbook—in Lertap 2 it simply filtered the data records, reading only the ones which met the criteria.

Our immediate use of this card at Curtin University will be to allow us to break out subgroups of students, so that we have separate workbooks for those in our elementary childhood, primary, and secondary streams.  To do this will require three unique applications of the *TST card; the first time the card’s specification will be

*TST c6=(E)

then *TST c6=(P), and finally *TST c6=(S).  It would be possible to make the application of this card identical to that seen in Lertap 2 (in this case, new workbooks would not result, or would possibly be an option); I will wait to hear from users as to the need for this, if any.

23 May

Corrected some minor formatting problems in item statsf and statsb output (affective item results had some small formatting problems).

17 May

Installed Excel 4 file converter.  Works by clicking on the Eight Ball, a new icon towards the right end of Lertap’s toolbar.

Will convert any worksheet of any workbook to the Excel 4 format.  If the worksheet is the Data sheet, the first row will be stripped off in the conversion.  This means that the converted file will have column headers in its first row.

If the worksheet is the Scores sheet, or the Sorted sheet, the first row will be stripped off, as will the bottom rows (where the summary statistics and correlations are found).  Here again this is done so that the new file will have its first row with column headers, followed by data, and only data.

The resultant Excel 4 worksheet will import easily to SPSS.

In SPSS, do a File / Open, and ask it to look for xls files.  Tell SPSS that you want it to “Read variable names”; it will pick these up from the first row of the incoming Excel 4 worksheet.  Note that SPSS is much more particular about variable names than is Lertap.  It will often find Lertap column headers which it doesn’t like, but it will explain this, and provide information about how it’s changed the headers to names which it can work with.

It bears repeating that SPSS picks up Excel 4 worksheets with little hassle.  You do not need to have ODBC installed.  You do not need to specify field names to bring over.  This makes the process very straightforward, and it’s even fairly easy to get new students to do.

Note on adding columns to a Lertap 5 worksheet: let’s say you have had Lertap produce test scores, and want these to be converted to Excel 4 format—however, you wish you could bring over some columns from the Data worksheet first.

Can you add columns to Lertap’s Scores worksheet?  Of course.  Go to the Data worksheet, select the column you want to copy, return to the Scores worksheet, and paste into an empty column.  The Data and Scores worksheets have the same format, with the first two rows being used for header information; this makes copying and pasting columns easy.

8 May

Enhanced the *MWS card so that it may use asterisks to denote unused item responses.  For example,

*MWS C12, 0, *, 0, 1

would signal that the second response code is not used by the item whose responses are found in column 12 of the Data worksheet.  If Res=(A,B,C,D), this would mean that the item does not use response B.  (This feature was added to maintain compatibility with Lertap 2.)

4 May

Created several new web pages, and placed them at:

http://edweb.curtin.edu.au/systems/

They will soon move to:

http://education.curtin.edu.au/systems/

The new pages include two examples, “ChemQuiz”, and “Ed503 Survey”.

1 May

Made it possible for the *MWS cards of Lertap 2 to be used with this new version.  MWS means “multiple weighting specification”; it will be the fastest way for most users to double-key an item.

In Lertap 2, a card of *MWS 25,0,1,0,1,0 would mean that the second and fourth responses, or alternatives, for Item 25 were to be given 1 point, with the other three responses getting 0 points.  (When there are two “correct answers” to an item, the item is generally referred to as being “double-keyed”.)

In this version, Lertap 5, the syntax is almost the same, except that the item number is replaced by Cx, where x is the column number of the Data sheet where the item’s responses are found.  For example, if Item 25’s responses were found in column 41 of the Data sheet, then the corresponding *MWS card would be: *MWS C41,0,1,0,1,0

Another example:

*MWS C23, 0.0, 0.5, 1.0, 0.0, 0.5

here the second, third, and fifth responses to the item whose responses are found in column 23 of the Data sheet are getting non-zero weights.

The weights may be any real number, positive or negative, but must use the decimal separator common to English-speaking countries, which is a full stop, also known as a decimal point, and, in the U.S., as a period.

There is another way to multiply-key items: change their weights in the respective Sub worksheet.  Using *MWS “cards”, however, will be simpler for most users.

Note: see comments at 8 July for problems related to using the *MWS card when Regional Settings are set to use a decimal separator which is not a decimal point.

29 April

Fixed an overflow error in the Elmillon routine which makes the total test score (this error occurred when the number of cases exceeded ten thousand).

28 April

Installed routine to compute score intercorrelations.  These Pearson product-moment coefficients appear at the bottom of the Scores worksheet.

27 April

Added an ability to sort Scores, activated by a new menu bar icon.  (Makes a copy of the Scores worksheet, determines the area on this sheet where column headings and scores are written, selects this area, and then calls Excel’s normal Data / Sort dialog box for the user to complete.  The original Scores sheet remains unaltered.)

19 April

1

The program will now calculate a Total score when the number of subtests is greater than 1.  By default, subtest scores are simply added together.  It is possible to turn this thing off by using a WT=0 assignment on the *SUB card, exactly as done with Lertap 2.  For example:

*SUB Res=(A,B,C,D), WT=0, Name=(Badgers), Title=(UW)

The use of WT=0 in the example above means that this subtest’s score will not form part of the Total score.

Examples of other valid WT assignments: WT=.5  WT=2  WT=1.25  WT=-.8

(The last example indicates that the weight may be negative.  Note: see comments at 9 July; there are problems with the Wt= assignment when the system decimal separator is something other than a decimal point.)

2

Fixed an error in the Histogrammer, which would crash with negative scores (Carlos Gonzalez of UCV found this when processing a job with over 13 thousand students, and two subtests, an affective one with 66 items, and a cognitive test with 10 items; apparently the affective test scores could be negative—the histogrammer plotted the cognitive scores okay, but bombed out with the others).

3

Enhanced the Histogrammer by having it restrict the plotted range to no more than 50 intervals (should result in a “graph” which will fit on a single page when printed; this is true when the page has an A4 size--testing is required to see if the standard page sizes used in other countries will work).

18 April

Made a number of changes to make it easier to run the program.  There is a setting in the System worksheet which defines UserLevel (this is not new; it’s been there for some time); when it’s set to 1, numerous messages now appear when the user clicks on Run options.  The menu bar has been changed so that there are fewer options for this level of user, and, should the user opt to Interpret the CCs lines after they have already been interpreted, Lertap now automatically deletes all results sheets (Freqs, Sub, Stats1f, Stats1b, ….).

When UserLevel is set to 2, more buttons appear on the menu bar, including a new one, to the left of the smiley face, which will delete all the results sheets with one click.  Results sheets are Freqs, Stats, and Histos (primary sheets are Data and Subs; in the present version, these are the only sheets required to generate all the results sheets).

The system ships with UserLevel = 1.  To change it, unhide the System worksheet, and follow your nose.  If you change this setting, you must close the workbook and reopen it for the change to be effected.

I hope these changes result in a system which is much easier for “everyday” users to apply.  By everyday user I mean someone who wants to get results in a hurry, and has no interest in getting into the Sub sheets to make alterations.

17 April

1

Fixed an error in the processing of Sub sheets with non-standard names.  A standard name for a Sub worksheet is Subx, where x is a number.  Sub1, Sub2, …, Sub87 are examples of standard names.  Sub1a, SubWhoHa, and Sub87Revised, on the other hand, would be examples of non-standard names.

2

Added “How it Works” comments to the User Guide.  (If you were wondering how on earth non-standard Sub names get created, see this new section.)

3

Added a “Write subtest scores?” parameter to a subtest’s Sub worksheet.  The default setting is TRUE.  If it’s set to FALSE, the subtest’s scores will not be saved in the Scores worksheet (they will be written only long enough for the routines which create the stats worksheets, full and brief, to use them—after which they are zapped).

4

Activated the “Active?” setting for subtest items.  By default this is set to TRUE, meaning that each item is included when it comes to deriving subtest scores and stats.  If it’s set to FALSE, the item is excluded from both scores and stats.  Use this, for example, to take a “bad” item out of the subtest, or just to see what effect removing an item has on subtest reliability.

There is another way to set this parameter if the subtest involved is a cognitive one: use a value of 0 (zero) for the item in the subtest’s *WGS string.  This maintains compatibility with Lertap 2.  For example, the following *WGS string, which would correspond to a subtest with 5 items, has a zero in Item 3’s position:

*WGS 11011

When I want to test the effect of removing items from a subtest, I first use Excel to make a copy of the subtest’s Sub sheet.  In the copy I then make the changes I want, and follow this by applying the Liberty Bell.  See the “How it Works” section in the User Guide for more comments.

5

Wrote this comment: to find the impact created by applying different item weights, follow the suggestion just given, in the paragraph above, to make a copy of the respective Sub sheet.  Then, in the little boxes of item “Weight(s)”, make the changes you want, and ring the Liberty Bell.

14 April

Fixed the Data worksheet so that it does not contain macros of its own.  This makes it possible to rename any sheet to Data, and still have the little data-entry icons appear on the menu bar.  In fact, now all icons on the menu bar are always there; it had been the case that the icons which displayed depended on which type of worksheet was being used.  This fix also addresses the problem of copying the data sheet to a workbook which is not using the macros from the Lertap template.  While this could be done before, error messages would appear whenever a copied Data sheet was closed (or opened, for that matter).

13 April

Who would work on the 13^th?

12 April

1

The Lertap 2 control words of CFC and PER may now be used on *Sub control cards.  CFC activates the correction for chance option for cognitive tests, while PER causes percent scores to be output.

2

Support for affective subtests was added.  Whether or not a subtest is cognitive or affective is signalled by using the AFF control word on the *Sub card.  When *Sub contains AFF, the subtest is classed and processed as an affective one.  If AFF is not present, then the subtest is assumed to be cognitive.

The output for affective tests consists of a “full stats” worksheet, and also a “brief stats” worksheet.  The brief stats sheet provides a quick, and colourful, means of quickly seeing how items performed.  The full stats sheet provides more details on item responses than have appeared in any previous version of Lertap.

The “brief stats” formats now seen for both types of test, cognitive and affective, result from feedback from Ed van den Berg of San Carlos University.  I feel these quick-summary sheets have added greatly to the utility of the system.  I know already that I will now find it easier to discuss results with students and colleagues who ask me for assistance in processing their classroom tests.  The brief stats sheets print quickly, and look very good on my colour printer (an Epson Stylus 800).  (For that matter, the full stats sheets print well enough too – they’re just long.  Note that Excel 2000 has an excellent print preview button which makes it a simple process to adjust page breaks.)

3

Added another button to the menu bar.  It has a Liberty Bell on it (without the crack).  This button allows for processing subtest results one by one.  (The Run drop-down on the menu bar has, as its second option, a call to the Elmillon item analysis routine to act on all subtests presently active.  Use the Liberty Bell when you want to focus attention on just one particular subtest.)

7 April

Added another data entry assistant, “Spread’em”.  This takes the contents of a cell and breaks it up into individual characters, distributing them to the right.  This makes it possible to enter a string of item responses in a single cell, and then spread the string’s characters, one by one, over the cells to the right.  This routine checks to make sure there’s nothing to the right which would otherwise get overwritten.  The spreader is activated via a new icon on the menu bar.  (Also see 5 April above, Note 2.)

6 April

Made numerous small changes to the statistics output in the two forms of item stats worksheets (brief, and full).  Also changed some of the formatting found in these sheets.

5 April

1

Added colour highlighting to selected cells of certain worksheets.

2

Added a data-entry assistant to make it easier to enter strings of responses, making it unnecessary to seat single responses by using the <Tab>, <Enter>, or <RightArrow> key.  This utility is activated by clicking on the (new) icon to the right of the smiley face on the menu bar.  This icon appears only when in the Data sheet.  Note that this utility is trained so that it will not overwrite data.  (Also see 7 April notes below.)

4 April

1

Fixed Elmillon menu option so that it’s always on.

2

Added additional item stats display, called “brief item stats”.  This display can be turned off via a new option in the System worksheet, but I suspect users will want to keep it as it summarises crucial item data very quickly.  Now each subtest has full (Elmillon) item stats in a worksheet called “Stats1f”, and brief stats in a sheet called “Stats1b”.  This enhancement was requested by Ed van den Berg of San Carlos University (Philippines).

31 March

1

Added some error messages to the histogrammer.  A check is made to see that subtest stats appear at the base of the Scores sheet.  If they don’t the histogrammer will not run, but there’s no crash.

2

Added control for histogram bar length scaling; see the System worksheet.  This will let Carlos Gonzalez see “better” histograms when he, or any other user, has very large data sets.  The default is to trim the length of the bars so that they are never longer than 200 symbols.  This is done by resetting the number of cases represented by each symbol.

3

Added a menu bar button to hide / unhide the Sub worksheets.

4

Added error messages when all Subs have been deleted and Elmillon is called.

30 March 2000

1

Added an About button to the menu bar (smiley).

2

Changed the histogram button on the menu bar.

3

Fixed problems with *ALT card processing.  In theory it is now possible to run without this card.

4

Add information messages to CCs parsing, such as highlighting *COL cards as they are parsed.

5

Started to update the two webs sites, edweb.curtin.edu.au/systems and education.curtin.edu.au/~nelsonl