Educational Insanity

As the result of a quick Twitter back-and-forth, I told @JeffNugent and @mcglaysia that I would write a blog post about the research linking technology integration and student achievement.  A couple of points before I get into it:

• I have been the lead investigator on many (maybe a dozen or so?) studies aimed at examining the relationship between technology integration and student achievement.  These studies ranged from small studies (one or two schools) to federally-funded, statewide investigations.  So, I have a pretty decent practical understanding of this body of research and how the work gets done.  Some day I’ll write about the politics and the nitty gritty of this sort of work.  For now, though, I’ll just say that the old joke is more true than it is funny: “educational research is like sausage.  If you like to consume either one, you don’t want to watch it being made.”
• This pool of literature is deep and getting deeper all the time.  I can’t possibly get to everything.  In fact, I’m only going to cover those with which I am most familiar.  That means, I’m not necessarily presenting the “best” research; just those that I know of and that I think are reasonably respectable.
• The Ed. Tech. Action Network (ETAN) has done a decent job of summarizing some of the research.  You can find their page with lots of links here.
• Finally, I’m not terribly proud of my work in this area.  I know that advocates of ed. tech. say that we MUST show positive student achievement effects to move the policy agenda forward.  But, for me, student achievement, especially as typically measured in these studies, is not even close to the most important outcome we need to be considering when evaluating the impact of technology in education.  I’m much more interested in outcomes such as student engagement and student learning (as distinct from student achievement).

Anyway, onward…

*The study that’s getting the most attention and that is politically loaded is one that is still being undertaken.  Mathematica, Inc., along with SRI, two of the major independent research firms in the country, have been contracted to conduct The National Study of the Effectiveness of Educational Technology Interventions.  The project’s website contains all the information you need to know about the study, including the first report which was issued last year.  So far, after one year, according to the press release issued at the time of the release of the report, “On average, after one year, products did not increase or decrease test scores by amounts that were statistically different from zero.”  So, no link between tech. and achievement.  But, that was after one year. The next report should be out soon.

*I’ve been pleased with the amount of attention the West Virginia study I co-lead in 1998-99 has received over the last decade.  The report from that study documents fairly significant positive relationships between the use of computers and student achievement.

*Harold Wenglinsky has done some significant work in this field.  His first major study, conducted in 1998 while he was at ETS, demonstrated that under the right conditions, the use of computers in schools was positively related to math achievement.  More specifically, “higher mathematics scores were related to adequate access to computer technology (hardware, software, and overall infrastructure) in conjunction with teachers trained in technology use and the use of computers to learn new, higher-order concepts.”

*The USEiT (Use, Support, and Effect of Instuctional Technology) study, stands out to me for its high quality and for the quality of the many reports that have been disseminated from that one study.  Take a look, particularly, at Reports 10 and 13.  Of the many findings coming from that study, the researchers discovered that “students who reported greater frequency of technology use at school to edit papers were likely to have higher total English/language arts test scores and higher writing scores on the Massachusetts Comprehensive Assessment System (MCAS) than students who did not.”  I still use some of the scales that the USEiT researchers developed for my own work.

*Finally, and more recently, Missouri’s eMints program has been well-documented and thoroughly studied.  There’s an entire page of research reports, including the most recent analysis of student achievement.  eMints has been consistently positively associated with student achievement.

There’s more; gobs more.  If you cross-reference the works I’ve pointed to, you’ll be well on your way to collecting a critical mass of the work that’s been done in this area.

One of my posts from almost four months ago has been resurrected by comments from Tina K. and Amir.  In that post, I suggested I’d do some more digging.  So, I dug.

Some background…these are NAEP data with tables and statistics generated by the NAEP Data Explorer.  The NDE is an awesome (free!) tool for analyzing NAEP data.  It’s kinda amazing to me that more folks, including the media, haven’t picked up on this tool to do some really quick and easy data analysis.  Anyway, to satisfy the inquiries of Amir and particularly Tina, I analyzed 8th grade data from the 2007 NAEP administration.  The 8th grade assessment includes the best “type of computer” use data; i.e. we can break down computer use by some specific applications.  That’s what I did.  The math achievement results disaggregated by response category follow:

[NOTE: click on images to enlarge them]

So, quite clearly, the same results appear as with the 4th grade data in my earlier post.  The group of students who never or hardly ever use computers score significantly higher than the other groups, across all applications.  Again, I don’t know anything about those students demographically.  But, still…

Going one step further, I ran a regression analysis with four of the independent variables (i.e. the “types” of uses).  The NDE would only allow me to use four; it’s a statistical/psychometric thing…don’t ask.  So, I took out word processing and drawing as those seemed likely the most remotely associated with math achievement.  The results are as follows (again, click on the image to enlarge):

Make sense? Yeah, I know, unlikely. Unless you are well versed and regularly practiced in regression analysis, there’s no reason that would make any sense to you. So, let me try to summarize some key results:

• Of all the variance in math achievement, differences in these four types of computer use for math account for 16%. That’s not that high; not terrible, but it’s safe to say that, overall, computer use for math does not explain much of why kids differ on their math scores.

The independent variables are “contrast coded” which is the right way to do this analysis. But, it limits what we can say. That being said,…

• The average score for a student who never or hardly ever uses computers in any of those ways is 291.
• Students who use the Internet for math once every few weeks score a bit higher than the previously mentioned student (i.e. never or ever uses in any of the ways).  That is, by simply adding Internet use for math once every few weeks adds a little bit to the average score of the non-computer using student.
• Same story for using graphing programs for charts.
• Adding Internet use once every few weeks AND graphing programs once ever few weeks has a cumulative positive effect on the non-computer using student (again, though, VERY small positive effect).
• The more frequently kids use math programs to drill on math facts, the lower they score.

So, there you have it.  I’ll likely play around a bit more with the NDE to see what else I find with respect to other subjects and other uses of computers.  Fun times!

This is what I get for playing around with the NAEP Data Explorer: 

So, the use percentages are a bit depressing.  But, it gets worse…much worse…

This is 4th graders, 2007, and, yes…the score for the group of students who report daily or almost daily use of computers at school for math is (statistically significantly) lower than all of the other groups.

The smart thing may have been to tuck these findings in my back pocket and walk away.  But, I’d bet that if we controlled for a bunch of demographic variables and even a baseline math score, those differences would disappear.  I’ll have to dig a little further…

[THIS IS THE SECOND ENTRY IN MY WEEKLY SERIES]

I’m not alone in my concerns over the achievement gap and educational equity more generally.  I have, however, felt for a while now that among the many articles and reports I’ve read on these matters, the most convincing is this one.  Schools, Achievement, and Inequality: A Seasonal Perspective, written by Alexander, Entwistle & Olsen in 2001, tells a compelling empirical tale of the cumulative effects of two phenomena:  differences in school readiness and summer learning loss.  As the simple chart below shows (the data are made up), by the time low income students reach school age, they trail their higher income counterparts with respect to student achievement.  Over the course of subsequent school years, the schools serve all students equally well (i.e. the slopes of the achievement lines are equivalent from Fall to Spring).  Then, over the summer, higher income students demonstrate slight achievement gains (not nearly as much as during the school year) while low income students make no gains (nor do they necessarily suffer learning loss).  The result, over time, is that the achievement gap between low- and high-income students expands over time, though that widening can be attributed mostly to “out-of-school” factors.

So, the policy implications are fairly clear: if we are serious about narrowing the achievement gap, we need to fund universal pre-K programs, and seriously consider either funding summer programs for low-income students or, more radically, think about year-round schooling.  Schooling according to the agrarian calendar has run its course?

 [click for larger image]