QTRS Dec 19, 2024
Graphs and interesting content for the curious
If you enjoyed this content, please press the like button above and share it. This really helps spread the word about Briefed by Data. Of course, if you are not already a subscriber, please consider becoming one.
As we are in the holiday season, I’ll ask you to consider becoming a paid subscriber to Briefed by Data. Consider it a gift or tip to a favorite writer. I’d like to keep Briefed by Data free, but at the same time it is a fair amount of work.
As I see it...
As I was giving one of my final exams this past week, I was thinking about its value. Anectodal, it seems that fewer final exams are given in favor of take-home exams, projects, or papers. Even rarer is the cumulative final that students spend over 2 hours on, such as mine. There are numerous rationales given for this change, such as students don’t recall as much information when taking exams as compared to projects and papers, as well as tests are stressful and create anxiety.
As I see it, a point that is overlooked is that exams push students to focus longer and harder and that there may be cognitive benefits to this regardless of the content. I think an apt analogy is to sports. As a rower, I worked as hard as I could in practices, but it is really hard, if not impossible, to work as hard as you do when you race. Racing itself, win or lose, makes one better. I think this is true for testing and that there is cumulative effect.
As my four sons were going through elementary school, one thing that fell out of favor was timed multiplication tests. The argument was that math isn’t a speed test. It is ok to go slower and be more accurate. On one hand, this makes sense, but on the other, I think the timed multiplication pushes the brain to improve. At the same time, knowing facts quickly makes one more able to do more complex math problems. We see this at the college level in that not knowing basic arithmetic, let alone being able to do it quickly, hinders a student’s ability to do more complex problems. Am I wrong or right? Let me know in the comments.
Let’s get to some data.
Increased variance in TIMSS scores is unique to the U.S.
Speaking of education, the AEI article US Students Best Other Nations—in Achievement Gap Growth (12/11/2024) has a set of great interactive graphs. The main point being that the variation of scores from the 10th percentile to the 90th percentile has grown in the U.S. and that this is unique to the U.S. The two graphs below are TIMSS 4th and 8th grade math scores for the U.S. There is a drop-down menu to choose other countries.
In reading this graph, 2011 is, in essence, time 0. Scores for other years are relative to 2011, which, for the first graph, is the best year except for the 90th percentile score.
For 4th grades, in the graph above, the difference between the 10th percentile score and 90th percentile score in 1994 was about 20 points. In 2023, it was about 60 points. The 10th and 25th percentile scores are lower in 2023 than in 1995, while the 50th percentile score is about the same.
This is a good example where if someone were to just report the median, 50th percentile, then we could say that scores were up until COVID hit and dropped back to where we were in 1994, largely due to COVID. That story, technically correct, would miss a lot. Variation in scores started increasing before COVID and had no impact on the 90th percentile score, with only a small decrease for the 75th percentile.
Why has the variation increased? I’ll put some of the blame on smart phones. I think they are generally bad for all of us, but those that have more cognitive ability will control them better than those with less. As to the impact of COVID, or not on the top quarter, there is likely a wealth effect here. Parents with more money have the resources to help their children learn.
All this seems plausible, but may be BS. The problem here is that this increased variation is unique to the U.S. What makes the U.S. different? Go to the article and check. I don’t have an answer at the moment.
The graph below is U.S. 8th grade scores and shows similar patterns as above.
Nudging college students
While I’m on education, here are the findings from the paper Conditions Under Which College Students Can be Responsive to Text-based Nudging (Dec 2024). Note: GSU is Georgia State University, and I see them as one of the leaders in using data and technology to improve retention and graduation rates.
To preview our findings, the chatbot outreach most consistently affected student behavior when it addressed “acute” administrative processes. In other words, it was most effective for resolving issues that were serious, time-sensitive, or both, such as managing registration holds and outstanding balances, filing the FAFSA, and registering for next-semester courses. In contrast, communication encouraging students to utilize supplemental supports—meeting with advisors, attending coursespecific supplemental instruction, and other less obviously acute issues—yielded little to no effect. Outreach also did not influence overall indicators of academic success, including credit hours attempted / earned, term GPA, or next-semester persistence. Despite the lack of impact on overall measures of student persistence and success, GSU judged the impacts on navigation of administrative processes to be sufficiently compelling that at the end of each experiment they scaled the administrative chatbot to all of their enrolled undergraduates on both campuses the following academic year.
I get the desire for colleges to help students stay in college, and if chatbots help them get paper work done, then use it. At the same time, I perplexed about the need for this. Is attending college more complicated today than in the past? Mabye, as adding more administrators usually doesn’t simplify things. If not, why can’t students manage these things today? In particular, what is up with students not registering for classes for the next semester?
Nudges are good, but maybe we should put some effort into addressing the root causes of why students are having problems dealing with these time-sensitive issues. I’m assuming future employers aren’t going to use bots to remind employees when work needs to be completed.
D vs. R differences—infedelity
The first graph here from Which Men Cheat Least? Republican Husbands, Especially “Religious Wife Guys” (12/12/2024). During the six years between when the data was collected, Republicans had essentially no change in their views that extramarital sex is “always wrong.” Democrats, on the other hand, started close to Republicans but dropped 23 points to where barely half believe inequality is always wrong. I sure home they are clear with their future wives about this. Now, this doesn’t mean they cheat more but…
they do, which is the second graph. The problem with this graph is that it doesn’t separate the time periods as the first one does. This is just another example of how the world views of the two parties are diverging.
U.S. sets new natural gas consumption record
From the eia article, U.S. natural gas consumption reaches new highs driven by the electric power sector (12/13/2024).
U.S. natural gas consumption grew by 1% to reach a new annual high of 89.4 billion cubic feet per day (Bcf/d) in 2023, according to our Natural Gas Annual, and continued growing in the first nine months of 2024. The 1% increase in natural gas consumption in 2023 was driven by a 6.7% (2.2 Bcf/d) increase in consumption in the electric power sector, the largest natural gas consuming sector. U.S. consumption of natural gas for power generation averaged 35.4 Bcf/d, or 40% of U.S. natural gas consumed in 2023.
We just want more energy, and solar and wind aren’t (yet?) up to the task of meeting demand even as we do get more electricity from them. As I’ve noted, data centers are a big driver of electricity demand. Which leads us to
Data center update
Switch to procure up to 12GW of nuclear power from Oklo reactors by 2044 (12/18/2024).
Small nuclear reactor developer and nuclear fuel recycling company Oklo has signed a non-binding Master Power Agreement with US data center developer Switch to supply up to 12GW of power through 2044.
Under the agreement, Oklo will develop, construct, and operate an undisclosed number of its Aurora Powerhouse nuclear reactors and provide power to Switch through a series of long-term Power Purchase Agreements (PPAs).
ExxonMobil plots natural gas power plant to exclusively power data centers (12/12/2024).
US multinational oil and gas major ExxonMobil has revealed plans for a new 1.5GW natural gas-fired power plant dedicated to powering data centers.
The project is in the early stages of development. It would mark the first time Exxon has built a power plant that does not supply energy to its own operations.
What we are seeing is that data centers need so much electricity that they are setting up their own power generation instead of using the grid. That’s amazing. Are there other examples of industries that have their own reactors or natural gas systems to generate electricity?
The spinning CD
A classic holiday song from the Dropkick Murphys. You’ll probably find it either offensive or funny.
Please share and like
Sharing and liking posts attracts new readers and boosts algorithm performance. I appreciate everything you do to support Briefed by Data.
Comments
Please let me know if you believe I expressed something incorrectly or misinterpreted the data. I'd rather know the truth and understand the world than be correct. I welcome comments and disagreement. We should all be forced to express our opinions and change our minds, but we should also know how to respectfully disagree and move on. Send me article ideas, feedback, or other thoughts at briefedbydata@substack.com.
Bio
I am a tenured mathematics professor at Ithaca College (PhD Math: Stochastic Processes, MS Applied Statistics, MS Math, BS Math, BS Exercise Science), and I consider myself an accidental academic (opinions are my own). I'm a gardener, drummer, rower, runner, inline skater, 46er, and R user. I’ve written the textbooks “R for College Mathematics and Statistics” and “Applied Calculus with R.” I welcome any collaborations. I welcome any collaborations.
I think that learning happens in many ways, through repetition, through insight, through reflection, and so on. Even if one objects to the contrived environment of a timed multiplication test, the act of preparing for that test forces a student to be in a position to learn. In my opinion, there is intrinsic value in the timed test of cumulative knowledge, and the preparations provide a secondary motivation for these.