5 ms·

This. If you're not "teaching to the test", what are you teaching, exactly?

karaterobot 3y ago
It'd be a less flexible approach, with more lessons oriented around test problems than the underlying concepts they're meant to represent? So, an equivalent would be learning leetcode programming puzzles to pass an interview, vs. learning to be a well-rounded programmer.
detaro 3y ago
Related skills that are useful outside the limited form the standardized test takes. To take a practical example, if you wanted to get a group of kids to get best high marks in my high-school CS final exams, it'd been very effective to have them drill a a few standard sorting algorithms and a pile of formal definitions, so they are certain to be able to perfectly reproduce them from memory when prompted. But I'm fairly sure the few extra weeks of messing around with graph algorithms, despite that not being material relevant for the exams, probably were more useful for everyone in the class despite someone typoing a heapsort or not remembering how UTF-8 bit encoding works under pressure and loosing some points on that.
aylmao 3y ago
There's a clue later on in the article, when it describes reading for fun almost every day is also lower than previous years. The goal of school is not for students to learn what The Pearl by Hemingway is about. The goal is for students to learn "to read"— ie, learn what to skim, what to pay attention to, what's being said between lines, how literary devices are used, how to piece different layers of ideas from a stream of words, etc. And importantly, to enjoy it at least to some degree, since it's a healthy habit Teaching the test in this context would be similar to teaching students to write some "top 20" algorithms, and recite their time complexities, knowing at least some of them will be on the test. It's definitely easier than teaching students the parts that make up a programming language, what big-O notation actually means, and the sort of generalized knowledge that will not only allow them to write and analyze _any_ program, but to see the beauty in programming and maybe even inspire a few to program for fun. This view of school as some sort of "training camp" for a test, is a terrible manifestation of focusing on the metric to the detriment of focusing on the goal.
- brycewray 3y ago
  Hemingway?
  - jewayne 3y ago
    Was it over when the Germans bombed Pearl Harbor?
    - brycewray 3y ago
      Germans? Forget it. He's rolling.
- jewayne 3y ago
  Answers like this worry me, because it implies you believe that general principles can be taught without teaching basic facts or learning specific examples. Correct me if I'm wrong, but I think it's generally agreed upon that it's easiest to learn a specific example first, then use that as a jumping off point to take on the more daunting task of learning general principles. Programming languages are a great example of this -- in an introductory programming class, you're going to learn to create programs in a specific language, using all the constructs necessary to do so. And after you have some experience and mastery with that language, you have something to compare and contrast other languages against. And it's that mastery that actually inspires (more than) a few to program for fun.
gms7777 3y ago
The problem is that there are plenty of valuable skills that are difficult to test in a standardized way. Most standardized test questions focus on Bloom levels 1-3 (remember, understand, apply) and struggle to test higher order skills (analyze, evaluate, create). Those questions are harder to write, usually take much longer for students to answer, and it's hard to format them such that they can be graded in a standardized manner (e.g., how do you test the skills ability to "create", to generate new ideas, to plan, to design, using a multiple choice question?) Placing a heavy emphasis on teaching to standardized tests means that teachers spend less time fostering higher order thinking skills.
- no_wizard 3y ago
  >how do you test the skills ability to "create", to generate new ideas, to plan, to design, using a multiple choice question? The real answer to this is "don't use multiple choice questions", but that would require significant investment in actually properly grading the outputs. One of course could say that then you're going to teach to criteria (IE, what are the evaluators looking for?) but that may or may not be a bad thing, depending on how its handled.
- scarmig 3y ago
  But you need to master the lower order skills before you can master the higher order skills. If someone is unable to evaluate a multi-digit sum, they'll be unable to devise creative solutions that use multi-digit sums.
  - gms7777 3y ago
    Of course, I'm not saying that the lower order skills are not important -- as you say, they're foundational. And I'm not even saying that it's a bad idea to test lower order skills. But they're not the only important thing, and "teaching to the test" tends to treat them like they are and prioritize them over critical thinking, synthesis, and creative skills.
xeromal 3y ago
I have a buddy who managed to pass his engineer PE without studying and is generally very good at all math and logic based problems without really having to "learn the test". What he excels at though is breaking down a problem and deriving the solution rather than pattern matching. It isn't a very good skill for a timed test, but it is a very good skill in real life in engineering and pretty much everything else. Teaching the test takes away the learning of this valuable skill to have.
- taeric 3y ago
  If they passed the PE Exam, they are good at testing. They may also be good at other things, but that is the measuring problem, in a nutshell. How do you build a measuring mechanism that measures what you care about, and not only what it is good at measuring? The best example of this is a car's speedometer. It doesn't actually measure how fast the car is going. Mainly because that is a surprisingly hard thing to do. Instead, it measures something that is easy to measure and converts to what we care about. Specifically, how fast are the tires rotating? Which is all to say, hard problem space. I have no doubt there are degenerate cases out there. My assertion is that is largely in the noise, at the moment. Happy (well, mayhap that isn't the correct word, but interested?) to be shown data that shows otherwise.

It'd be a less flexible approach, with more lessons oriented around test problems than the underlying concepts they're meant to represent? So, an equivalent would be learning leetcode programming puzzles to pass an interview, vs. learning to be a well-rounded programmer.

detaro 3y ago

Related skills that are useful outside the limited form the standardized test takes. To take a practical example, if you wanted to get a group of kids to get best high marks in my high-school CS final exams, it'd been very effective to have them drill a a few standard sorting algorithms and a pile of formal definitions, so they are certain to be able to perfectly reproduce them from memory when prompted. But I'm fairly sure the few extra weeks of messing around with graph algorithms, despite that not being material relevant for the exams, probably were more useful for everyone in the class despite someone typoing a heapsort or not remembering how UTF-8 bit encoding works under pressure and loosing some points on that.

aylmao 3y ago

There's a clue later on in the article, when it describes reading for fun almost every day is also lower than previous years. The goal of school is not for students to learn what The Pearl by Hemingway is about. The goal is for students to learn "to read"— ie, learn what to skim, what to pay attention to, what's being said between lines, how literary devices are used, how to piece different layers of ideas from a stream of words, etc. And importantly, to enjoy it at least to some degree, since it's a healthy habit Teaching the test in this context would be similar to teaching students to write some "top 20" algorithms, and recite their time complexities, knowing at least some of them will be on the test. It's definitely easier than teaching students the parts that make up a programming language, what big-O notation actually means, and the sort of generalized knowledge that will not only allow them to write and analyze _any_ program, but to see the beauty in programming and maybe even inspire a few to program for fun. This view of school as some sort of "training camp" for a test, is a terrible manifestation of focusing on the metric to the detriment of focusing on the goal.

brycewray 3y ago
Hemingway?
- jewayne 3y ago
  Was it over when the Germans bombed Pearl Harbor?
  - brycewray 3y ago
    Germans? Forget it. He's rolling.
jewayne 3y ago
Answers like this worry me, because it implies you believe that general principles can be taught without teaching basic facts or learning specific examples. Correct me if I'm wrong, but I think it's generally agreed upon that it's easiest to learn a specific example first, then use that as a jumping off point to take on the more daunting task of learning general principles. Programming languages are a great example of this -- in an introductory programming class, you're going to learn to create programs in a specific language, using all the constructs necessary to do so. And after you have some experience and mastery with that language, you have something to compare and contrast other languages against. And it's that mastery that actually inspires (more than) a few to program for fun.

gms7777 3y ago

The problem is that there are plenty of valuable skills that are difficult to test in a standardized way. Most standardized test questions focus on Bloom levels 1-3 (remember, understand, apply) and struggle to test higher order skills (analyze, evaluate, create). Those questions are harder to write, usually take much longer for students to answer, and it's hard to format them such that they can be graded in a standardized manner (e.g., how do you test the skills ability to "create", to generate new ideas, to plan, to design, using a multiple choice question?) Placing a heavy emphasis on teaching to standardized tests means that teachers spend less time fostering higher order thinking skills.

no_wizard 3y ago
>how do you test the skills ability to "create", to generate new ideas, to plan, to design, using a multiple choice question? The real answer to this is "don't use multiple choice questions", but that would require significant investment in actually properly grading the outputs. One of course could say that then you're going to teach to criteria (IE, what are the evaluators looking for?) but that may or may not be a bad thing, depending on how its handled.
scarmig 3y ago
But you need to master the lower order skills before you can master the higher order skills. If someone is unable to evaluate a multi-digit sum, they'll be unable to devise creative solutions that use multi-digit sums.
- gms7777 3y ago
  Of course, I'm not saying that the lower order skills are not important -- as you say, they're foundational. And I'm not even saying that it's a bad idea to test lower order skills. But they're not the only important thing, and "teaching to the test" tends to treat them like they are and prioritize them over critical thinking, synthesis, and creative skills.

xeromal 3y ago

I have a buddy who managed to pass his engineer PE without studying and is generally very good at all math and logic based problems without really having to "learn the test". What he excels at though is breaking down a problem and deriving the solution rather than pattern matching. It isn't a very good skill for a timed test, but it is a very good skill in real life in engineering and pretty much everything else. Teaching the test takes away the learning of this valuable skill to have.

taeric 3y ago
If they passed the PE Exam, they are good at testing. They may also be good at other things, but that is the measuring problem, in a nutshell. How do you build a measuring mechanism that measures what you care about, and not only what it is good at measuring? The best example of this is a car's speedometer. It doesn't actually measure how fast the car is going. Mainly because that is a surprisingly hard thing to do. Instead, it measures something that is easy to measure and converts to what we care about. Specifically, how fast are the tires rotating? Which is all to say, hard problem space. I have no doubt there are degenerate cases out there. My assertion is that is largely in the noise, at the moment. Happy (well, mayhap that isn't the correct word, but interested?) to be shown data that shows otherwise.