Does class size play a part in SAT scores or is there something more to the story? In this segment of Visualize This, take a gander at this data from the National Center for Education Statistics (NCES).
Get your entry in by November 24, 2009 at 11:59pm EST. One entry per person. The two best entries will win a copy of The Visual Miscellaneum. Find more details here.
Here are instructions on how to post images to the forums in case you're new here.
Hi Nathan,
Do you want us to use only SAT scores and teacher/student ratios? Or can we use other data on that site?
You can bring in any data that you like as long as you focus on this segment's secret ingredient i.e. class size and quality of education :)
Okay great. Which table did you use for class size?
Ah I forgot about that. You can find pupil/teacher ratio by state here:
After reading a number of the comments on the previous post asking for a scatterplot and talking about the effect from the percent of students taking the test, I decided to make a scatterplot that showed that effect:
From this, you can see that the percent of students taking the test does have a significant impact: when only a few students take the test, there tends to be a selection effect: the students who take the test tend to be students who do well on the test. But that if you look at the two groups individually (those with a high percentage taking the test, in yellow and those with a low percentage taking the test, in blue), you can see a relationship due to class size as well.
The screen shot doesn't really give the full sense, though. You can get to the interactive version here (you can also find the data in .csv format there). In that version, you can change the x axis, y axis, and color scale to any of the variables of interest. In addition to the average SAT scores, class size, and percent of students taking the test, the average state expenditure per student is also included as a possibility. You can also turn on the display of lines for the national averages, turn on a color scale, or click on individual points to get full details for that state.
All of this was written in javascript, using Flot. Also, let me say thanks to Nathan for the interesting original post and this followup!
Nice work, lotze. You get a gold star for being first.
Nice. Can we use any NCES data in conjunction with tabn144?
yes, you may.
I'm new to this infographics stuff, so be kind!
Given that the most sensible thing seems to be plotting teacher/student ratio (decreasing) against average SAT score by State (increasing) I thought a bar chart pointing downwards with a line graph superimposed might do the business.
I still may have a tinker with it - I presume I can change the image up until the deadline? :-)
Doug
I just posted about the visualizations on the Grockit blog, which covers how we're looking at various education issues, particularly the use of technology to improve education.
I've also updated my visualization to include ACT data and longitude/latitude, which means that you can create (among other things) a map showing which states tend to use SAT versus ACT.
Lotze, nice tool; flot looks versatile. What technologies are other folks using to create their graphs?
Nathan, your use of line graphs is useful when demonstrating change in rank over time, but I don't see the value in using this over a scatterplot if the intent is to demonstrate correlation.
I have limited myself to the original SAT information to save time looking for the perfect data set. My report addresses the same issue Lotze identifies (i.e. participation rates must be accounted for before comparing SAT scores) by attempting to find subgroups where participation rates are no longer a factor and then ranking within those groups.
The effect of class size is a hot topic; I am not sure if there is a proven link to student performance. Exploring this relationship at the level of the state is likely to result in misleading findings at best. Consider the schools in hypothetical state A (student-to-teacher ratio, mean SAT scores) {(50:1, 500) (40:1, 400) (30:1, 300)} versus those in state B {(30:1, 550), (25:1, 500), (20:1, 450)}. At the level of the state we find A having a mean student-to-teacher ratio of 40:1 and mean SAT score of 400, and for state B, 25:1 and 500. Based the state level we would assume that decreasing the student-to-teacher ratio improves SAT results, but within each of the states the exact opposite trend is evident.
The following is my submission. I attempted to achieve a very content-driven and high data-density visualization that allows for sound, albeit limited comparisons and conclusions to be reached. The details on how the report was produced is available at http://pushindatalikeweight.com/ .
Here is our submit:
https://monentreprise5.bimeapp.com/players/dashboard/E02DC7212573C55BB6CD05C0A48E03E31
Nice job from other participants!
We tried our new scatterplot/bubble graph with your dataset. More colors are coming soon :)
Thanks for putting all this together!
I was intrigued by the original graphic - in particular I was surprised there wasn't a stronger relationship between student/teacher ratios and SAT scores. So I decided to do a little digging (thanks for the interesting insight!)
After seeing how big the difference in SAT participation rates were from state to state, I settled instead on graduation rate as my measure of "success".
Since the lack of correlation in the original example was what surprised me, I thought I'd look at a few other items - focussing on the strength of correlation as the pertinent measure.
I put together a quick (interactive) graphic here:
I was interested in how things held up over time as well, so I also focussed on that aspect.
To be really useful, it would be interesting to add more metrics (race, gender?) as well as add a longer history, but this abbreviated version made for an interesting exercise.
Hi, here's my submission (best viewed at original size):
and here's a bigger version of one of the charts:
I've added on a few years of earlier data to see what happens within each state.
Hi - Fun concept with a 'visualize This'-competition. I made a visualization of the data too.
I think the very strong and obvious connection between level of participation and scores can't be overlooked, and so I have used this as my base - leaving the possible connection between size of classes and SAT-score more open to interpretation.
I have only used data for 2006/2007, where we have the pupil/teacher ratio data available.
I love the site, but I haven't been inspired to post until now, since I actually know a thing or two about education policy research in the US. I'm obviously a bit behind the curve in terms of creating elegant visualizations (hopefully someday!), and I've stuck to a series of scatterplots from stata via inkscape. I also wanted to highlight the NAEP data as a resource for making legitimate achievement comparisons across time and space (within the US).
Of course a more direct answer to the question "do we need more teachers?" is provided by actual experiences with reducing class sizes. The Tennessee STAR project, for instance, was a randomized class size reduction study that found positive effects on student performance in the early grades. However, the state-wide class size reduction initiative in California in the 1990s led to little improvement, largely because there weren't enough facilities and qualified teachers to support the policy. So the effect of class sizes is complicated, and even though there's strong causal evidence that smaller classes are better, there might be better ways to spend money in schools (hiring more teachers is very costly).
Anyhow, my entry sticks to state-level associations, and is [hopefully] linked below. The text will probably only be legible in the full-sized version.
(and since it seems that this post may be a couple of hours past the deadline, I suppose I'm hoping for leniency on that front-- it's still the 24th in Hawaii?...)
Cheers,
paul
thanks all for your entries. i'll be announcing the winners soon.
This is tangential to the competition, but I purchased your FlowingPrints education series for a charter school I visited in NY. Wrote the visit up at http://notestoself.posterous.com/kipp-to-infinity-and-beyond. Who knows? Maybe some kids will be inspired by the infographics!
:)
You must log in to post.
