Our usual base is in the southwest of Western Australia, a place with quite a nice climate, plenty of sunny beaches, good cricket teams and all that, but from time to time we take a break, stay inside, and do a bit of actual work.
Here are some of the things we've been up to most recently.
•Added a Rasch capability for item calibration
Lertap now has built-in Rasch skills with a focus on calibrating dichotomous items, such as those found in multiple-choice tests. Read up on it by visiting this topic. Suggested exercises for possible use with classes are mentioned here.
•Changes made in the Stats reports
Stats_f reports now include the 95% confidence interval for coefficient alpha. In Stats_b reports, negative item discrimination values, and negative item-scale correlations, are now shown in red. In the bands reports, at the very bottom of Stats_f sheets, positive alpha changes (usually unexpected and unwanted) are also now wearing red.
•Updated two chapters in the manual
Chapter 8 has to do with processing surveys and other "affective" instruments.
Chapter 9 gets into how to share Lertap data with other systems, such as SPSS, SAS, Xcalibre, and Pearson VUE.
•Improved video resources
The video resources page has been updated and enhanced; links to additional "shows" have been added.
Response Similarity Analysis, "RSA", is used to see if the test answers given by a pair of students might be "suspiciously similar". The Windows version of Lertap5 (220.127.116.11) has been enhanced so that RSA users may change some of the run settings on the fly, without having to make changes in Lertap's System worksheet. The Mac version (5.10.99) had run-time errors which have now been fixed.
The Windows version has also seen changes made to the sort order in RSAsig and RSAcases reports. These reports were formerly sorted on their "sigma" column -- now the sort is on the "H-H Index" column. A discussion and example is found here.
•Started to move to a new web domain, Lertap5.com
Initiated the process of moving all Lertap5 material to a new domain: lertap5.com. The former domain, "www.larrynelsonstuff.com", will continue to be operational until March 2020. A re-router is now in place; visitors to the former domain will be automatically redirected to the new one. In theory, web browsers, such as Chrome, Firefox, Edge, and Safari will update their URLs by themselves (in theory).
•Updated PearsonVUE topics
The pdf document describing how to use Lertap5 with data from PearsonVUE has been updated with the inclusion of a practical example.
•Added more DIF resources
The "GimmeABreak" paper has been updated with some new sections, and new examples, related to "DIF", differential item functioning. A new paper is also available discussing how to link Lertap with the difR package (has extensive DIF resources).
•Became aware of some important new Microsoft security settings
The "Faster opening and closing" item below mentions how Excel seems to have slowed down, and in quite a pronounced way, a noticeable way. It turns out that Microsoft added enhanced virus scanning capabilities to Windows 10 (back in September 2018) and Lertap's main workbook, Lertap5.xlsm, will be scanned for viruses every time it's opened unless some special action is taken.
Lertap5.xlsm has lots of "macros"; scanning them for malicious code takes time -- read this topic to find out how to tell Windows 10 that it doesn't need to scan the Lertap5.xlsm workbook. You may find that Lertap will open much more rapidly.
Note in mid-2019: Microsoft may have addressed this issue behind the scenes. Worry about this matter only if it takes more than 15 seconds for Excel to open Lertap5.xlsm.
•Faster opening and closing
What happened to Excel in 2018? It seems to have slowed down, needing sometimes well over a minute to open Lertap, and even tens of seconds to close it. Open and close times were never an issue before.
So: we got in, tickled and twisted some areas of Lertap code, and it is better; now, once it has been installed, it will open substantially faster, and close in a jiffy. The improvements started to appear in version 18.104.22.168 (early December 2018).
•IRT users get even more support
Building on work done at the University of Western Ontario, a new script for use with Rstudio is ready. It's for those wanting to get IRT statistics and graphs for dichotomous test items. Go to this page and follow links found there. Also, admire this topic about a Rasch capability now built into Lertap itself.
•Added a paper on using R and RStudio
These two systems, R and RStudio, can receive data and appropriate programming code from Lertap and do wondrous things. This new paper has the scoop.
•Added a "cleaned quantiles" ability
Have a look at these packed plots when you have a chance. They're new. And, then again, they're really not new: all they are are conventional quantile plots with the trace lines for the distractors removed. The idea is to make it easier to spot trends: does the response trace for an item start low on the left, say below 0.20, and steadily climb, getting over 0.80 or better on the right without taking a dip anywhere? If so we've got a discriminating item and, were we to try and fit some sort of IRT model, we might have cause to be optimistic for an okay fit.
A link to the complete "updates summary" page is here.