A few historonics

The first version of Lertap was developed for the Venezuelan Ministry of Education in the years 1971 through 1972. The Ministry was then embarking on a national assessment program, with emphasis on mathematics and language achievement. The Kuhlmann-Anderson aptitude, or "IQ", test was also used on a national scale by the Ministry, and a general-purpose item analysis program was required, one which could handle conventional achievement tests, and the Kuhlmann-Anderson.

At the time, the development of the Ministry's assessment centre was under the direction of Rogelio Blanco, with Richard Wolfe, of OISE (Ontario), overseeing the technical services part of the operation. Richard created a general front-end to set up data sets for subsequent analyses, using the PL/I programming language. I contributed what amounted to the first version of Lertap, programmed in FORTRAN II. It picked up data sets pre-processed by the PL/I program, and output classical item statistics. The first Lertap, locally called "DIEitem", could not only handle the idiosyncrasies of the Kuhlman - Anderson test, but could also entertain multiple tests within the same data set. Thus one could submit a data set with results from the mathematics test, the Spanish-language test, and the Kulhmann-Anderson test, all strung together in a lengthy input string.

Work on DIEitem was supported by the Ford Foundation, and by the la Organización de los Estados Americanos, OEA.

The first English-language version

In 1973 work on the second version began at the University of Colorado, home of the Laboratory of Educational Research. The PL/I front end was replaced by another, written in FORTRAN, which featured the use of a set of free-form control cards to describe a job. These control cards included ones called *TST, *FMT, *SUB, and so on (I mention them as a reminder to those who spent much time with this version). Free-form control cards were not widely used in those days, and, in this regard, Lertap 2 could be considered as being slightly ahead of its time.

Lertap 2 also introduced support for processing affective tests. The late Bob Conry of the University of British Columbia provided strong support for the "aff" subtest capability, while Ken Hopkins and Gene Glass, at LER in Boulder, did all possible to encourage the development of the overall package.

The work started at LER was continued when I assumed my first position at the University of Otago, in Dunedin, New Zealand, late in 1973. By the end of 1974 Lertap 2 was stable, and in use in a variety of centres in Canada and the United States. Lertap 2 was blessed, if I can use that term, with a complete user guide. I draw attention to this fact as the same could not be said of the IBM PC version which later emerged when the first microcomputers appeared.

I am grateful for the Lertap 2 support provided by several people at Otago, especially Dan McKerracher, Department of Education, and Brian Cox, Computing Centre. The user guide which emerged from Otago was a fun document to write, and seemed to be quite well received. Brian Cox saw that it had a fetching cover, featuring his favourite Burroughs computer, and some of his staff.

Use of Lertap grew steadily in the 70s. During these years I held posts at Boston University, and la Corporación Venezolana de Guayana.

An apple a day

Late 1980 saw me back in Otago, experimenting with the new Apple ][ microcomputer. A year later I began the development of Lertap 3 in earnest, using a CP/M card on an Apple, and then on an Osborne 1 system. I found the BASIC 80 language to be capable of speedy performance, and used it to produce a suite of interlinked modules which would load and unload themselves in just 56K of core memory.

In early 1983 a working version of Lertap 3 was ready, and it was accompanied by a new user guide every bit as complete as that created for Lertap 2. Barbara Calvert keenly supported the development of this version, putting some of the resources of the Department of Education behind the effort. A few hundred copies of the user guide were printed, and made ready for distribution.

Enter IBM's decision to produce a microcomputer of its own. Enter another LER alumnus, Evelyn Brzezinski of the Northwest Regional Educational Laboratory, Portland, and Larry Erikson of National Computer Systems, Minneapolis. By late 1983 Lertap 3 had been altered so as to operate within IBM micros, and National Computer Systems had purchased it. NCS repackaged the system as two stand-alone programs, MicroTest1, and MicroSurvey1.

I was free to continue the development of Lertap 3, and did so, at a reduced pace. By the late 80s this version was in use at many sites, with the NCS versions finding a home in many (many) more. It was unfortunate that a user guide for this version was never thoroughly developed. The guide printed at Otago covered the pre-IBM version, but the operation of this version differed much, and IBM users found it to be of limited use.

In 1987 I began to circulate a series of brief user help sheets from Curtin University. These were later assembled as a small book, Lertap 3 General Notes, printed by Curtin University's printing services.

Lertap 3 remains a potent data analysis system. The scope of analyses it supports includes those related to cognitive and affective tests, general surveys, and classroom gradebooks. Its data preparation facilities include a module for complete date entry verification. And, it can handle results from the Kuhlmann-Anderson tests. (Not that they're used that much anymore, but Form B of the K-A is complex, making more exacting demands of a test-analysis program.)

In 1992 Piet Abik translated Lertap 3 to the Indonesian language, and it was later purchased by Indonesia's Ministry of Education and Culture for country-wide use in secondary schools, with the support of Bambang Irianto.

Lertap 2, Lertap 3, and the NCS derivatives came to be used throughout the world. I am aware of some of the corners they've reached by references in publications, and by new would-be users emailing me to ask about the system's availability.

When Microsoft released the Windows 3 operating system, in 1992 (in Australia), it was soon clear that Lertap had to move to Windows. People started to write to ask if the Windows version would be ready soon. I worked at it, off and on, until 1997.

A dud, dudes

I refer to this work as Lertap 4. It was never finished, It came to have a nifty facility for processing survey results, but not much more. My error was in believing that I could, on my own, rebundle all of Lertap 3's power in a Windows package. In order to appreciate the scope of this desire it would be necessary to understand both the extent of Lertap 3's capabilities, and the nature of programming under Windows. Suffice it to say that the idea of building a stand-alone Windows version of Lertap 3 is one I had to give away.

Por fin, me pegó la luz: Excel

So, you might ask, what's this new version of Lertap, then? Lertap 5? Isn't it running within Windows? Yes, certainly. It will also run on a Macintosh system, or, for that matter, any platform which supports a recent version of Excel.

What is different about this new version is that it is built on Microsoft's spreadsheet flagship, Excel. Using Excel has freed me from having to develop user interfaces related to data entry and maintenance, and, to my pleasant surprise, writing output to Excel worksheets is entirely straightforward compared to writing output to generic Windows forms. There are numerous spots where data in a Lertap 5 worksheet are simply passed to Excel; this has freed me from having to re-structure Lertap 2 & 3 modules, and it has undoubtedly resulted in faster program execution.

I have, furthermore, gone back to Lertap 2's method of job definition. Instead of having users answer a multitude of dialog boxes, they define their jobs by using control "cards", ones which are nearly identical to those first seen in Lertap 2 almost 30 years ago. A retrograde step? I think not. The parsimony of using a control "card" job definition language is remarkable. I know that users can master this language; it's not that extensive, and it has a track record of success.

The result is a system which asks almost no questions of users. This is a vast change from both LERTAP 3 and the design seen in Lertap 4. I did not intentionally set out to produce such a system; I didn't know it might be feasible. Yet here it is. Users enter their results in the Data worksheet, type up their control "cards" in a worksheet called "CCs" (for Control Cards), and click on the Run button. Not once are they asked a job definition question of any sort.

Lertap 5 credits

I gratefully acknowledge the support of Curtin University of Technology's Faculty of Education, and its Division of Humanities, which made it possible for me to set aside several months of development time in the year 2000 without having to be concerned with classes and committee work. A special thanks to Graham Dellar.

Nanta Politawanont of Burapha University, Thailand, and Suchada Kornpetpanee, also of Burapha, provided a home away from home in 2000, letting me use several of their Thai data sets to debug initial sections of Lertap 5 code.

Todd Rogers of the University of Alberta, along with his doctoral students, especially Keith Boughton and Tess Dawber, have provided invaluable guidance, putting Lertap 5 through the ringer on numerous occasions, and at times pointing the way to code modifications.

Nurhadi Amiyanto of the Government of Central Java, Indonesia, sponsored a Lertap workshop series in 2002 which saw the software tested on a few Jateng data sets; at the time, with over twenty thousand students, this was Lertap's most substantial data processing challenge.

Carlos Gonzalez, of la Universidad Central de Venezuela, has been behind Lertap 5 all the way, often sending sample data sets and testing new features.

After some initial reluctance, David Weiss of Assessment Systems Corporation, creator of Iteman, XCALIBRE and a vast quantity of other psychometric resources, saw the light, and gave Lertap 5 a prominent role in his shop in mid 2002.

Professor Ken Hopkins

Finally, a dedication. It is an honour to dedicate this version to Professor Kenneth D. Hopkins of the University of Colorado. He has been teacher, mentor, and friend since 1971. His books and publications dealing with classical item analysis have provided the bases for Lertap's development, and his frequent feedback over the years has done much to bring the system to its present state. Many will be the number who join me in wishing him a long and pleasant retirement. Australians would say "Onya, Ken, you done good". Indeed. LERTAP is, I hope, at least a small credit to his teaching and writing.

Larry Nelson
Faculty of Education, Language Studies, & Social Work
Curtin University of Technology
May, 2005