Quote Originally Posted by noetsi View Post
Given that I purchased EQS not MPLUS Bentler must still make some money (Mplus won't let you use the student discount if you have a PHD even if you are in another graduate program. EQS will. A bizare rule lol and buying either of these without a student discount is impossible for moi).
Bentler had a collection of grants from the NIH that ran out like 2 or 3 years ago. those grants helped fund EQS and make it what it is today. he thought that the company he founded (Multivariate Software Inc) could support itself though the sales of EQS and whatnot, but when everybody started ditching him for Muthen and Mplus, he pretty much stopped doing any updates for EQS. if you see it in his website, EQS has had no substantial updates (only maintenance ones) since 2012. right now he's focused on IRT software (through EQSIRT in his new company) but i think he's too late to jump into this bandwagon (business-wise speaking). back in the day when the choice was LISREL or LISREL, EQS came as a blessing because it implemented a different model parameterization (the Bentler-Weeks model) that made the coding and analysis of data a lot easier. the problem with the Bentler-Weeks model is that it does not allow for the most advanced forms of latent variable modelling (IRT, finite mixture models, etc.) but Muthen's parameterization does (which is an improvement on Joreskog's LISREL model) so it simply makes economic sense to ditch all software at the expense of Mplus. you don't need to get EQS to do regular SEM and EQSIRT to do Item Response Theory because Mplus does both. you don't need to buy HLM or MLWin to do multilevel models because Mplus can do that too. the only real challengers at this point to Mplus' control of the market are R and STATA. no other piece of software can do as much as Mplus does with the exception of those two at the present moment. so yeah, i could see how Bentler and Multivariate Software Inc would be prone of offering discounts and special sales to students as long as they buy EQS and keep on using it, even though the number of people who use EQS to run SEM models dwindles every day.


Quote Originally Posted by noetsi View Post
My guess is that when they crunch those numbers they come up with what their political superiors want. Or keep crunching them till then. My research dealt with how productivity was influenced by politics, it was on implementation of new productivity methods such as Sick Sigma). I know in testing of high school students, politics trumps IRT
oh but of course that happens! when you play the Game of Thrones (aka the game of politics) you need to make sure your (usually fake) claims are as substantiated as much as possible by whichever fake analyses sound more impressive. it's like when Republicans say "the unemployment rate is terrible" and then Democrats say "the unemployment rate is good". your average citizen would (rightfully) ask "how can these opposing conclusions be obtained from the same data? it makes no sense!" but your skilled data analyst would notice that each political power focuses solely on certain statistics that fit their political narrative and ignore the other ones, especially the way in which unemployment is measured (which, of course, influences anything you'd like to say about it). Education is no different in this extent. Value-added/multilevel models allow politicians to push for policies related to teacher accountability (usually opposed by the teacher unions) by showing how "scientific" (notice the " please) analyses support their claims. but if you ask the analysts defending the teacher union they'll tell you these value-added/multilevel models are wrong. who is right? who is wrong? well, everybody and nobody. it all depends on who you wish to believe. personally, all i want is a stable paycheque