Comment by quadrifoliate | Hacker News Reader

Comment by quadrifoliate | Hacker News Reader

Parent Full thread

If you're a programmer (like many on this site are), I find it hard to see how you think state assessment tests can be objective or measure real learning for a student in any statistically meaningful way. They make whiteboard coding tests look extremely subjective, fair, and balanced in comparison.

I know people who have had to teach 3rd graders to give standardized tests, and reality is that a lot depends on the mood of the kid that day. Some kids just idly click away, and the teacher has to tell them to actually read the damn question before clicking on an answer.

A lot of the signal that those tests will give is just that kids weren't super attentive or feeling excited about giving the tests this year, for obvious pandemic reasons. Hard to be in a great mood when your parents are home all the time, everyone is out in masks, and all events and fun things are canceled or have restrictions.

But that doesn't seem to prevent people from pretending like there's some big scandal backed up by objective data about "learning shortfalls" and such.

Obviously tests are imperfect, but it's silly to dismiss them as useless. Standardized tests are the best tool we have to ensure our schools are providing all children quality educations regardless of economic or racial background or disabilities. Making education policy without measuring outcomes in a rigorous way is like doing science without running experiments. Children of priveleged families will always do fine. Their parents have the knowledge and bandwidth to take a proactive role in their children's educations, but we can't leave behind all the kids who aren't so lucky.

> Making education policy without measuring outcomes in a rigorous way is like doing science without running experiments.

It's worse to present an experiment as objective without properly measuring its error bars. As I have pointed out, there are a million different factors that can influence performance on tests. Claiming that you are "measuring outcomes in a rigorous way" from these tests without putting in the time and money to assess and control for all those factors makes for bad public policy. Worse, you'll end up providing incentives to focus only on test scores to the detriment of actual classroom learning.

As I have pointed out, there are a million different factors that can influence performance on tests.

For individuals, yes. But for the entire cohort not really - the results are being compared to previous years, so the error bars should be approximately the same. Children are, on average, the same as two, five or ten years ago, so the test outcomes should be about the same. The article suggests that all children are now doing worse on tests than previous years.

> Children are, on average, the same as two, five or ten years ago, so the test outcomes should be about the same.

The mental state of children today is not the same for those five or ten years ago, for a myriad of reasons; this should be obvious.

> The article suggests that all children are now doing worse on tests than previous years.

Yes. I am also sure they will do worse on a rubric for mental well-being. Unfortunately no one seems to have thought about measuring or improving that, since most of the standardized-test focused measures probably worsen it.

I read you to be dismissing tests outright. If you merely want to stipulate that the test scores are subjected to rigorous statistical analyses that "properly measure its error bars" we are in total agreement.

> If you merely want to stipulate that the test scores are subjected to rigorous statistical analyses that "properly measure its error bars" we are in total agreement.

I am saying that the system should fail open in their absence. "Students' achievement dropped 50 points!" makes for a really potent policy cudgel. "Students' learning dropped 50 points, +/- 40" is silly and meaningless. The former should never be available, unless the +/- is explicitly mentioned.

This is similar to prescription medicines. We have a lot of prescription medicines that do great things, which no one is debating. But without a proper evaluation process, you have zero access to them. Test assessments should be similar in my opinion.

> Standardized tests are the best tool we have to ensure our schools are providing all children quality educations regardless of economic or racial background or disabilities.

It's a work of fiction but I'd like to refer to "The Wire Season 4" and how they portray standardised test taking to get you a fresh perspective outside of the utopian thought about curricula.

Standardized tests work just fine to assess competency. White board interviews probably would to if they were designed and graded by assessment professionals. Those tests are designed to assess competency at a specific level, and they are a lot better at it than many give them credit for. You don't have to be in a great mood to be able to read, especially at a third grade level. These tests are far better than anyone wants to admit at showing how terrible our schools are in the U.S.

> You don't have to be in a great mood to be able to read, especially at a third grade level.

Massive [citation needed]. It seems reasonably obvious that mood will affect ability of reading, especially when you can't measure it directly; you're probably doing so via some questions that ask about comprehension. Have you not seen all the conversations about adults burning out at their jobs during the pandemic? Why do you expect children to be totally unaffected?

> These tests are far better than anyone wants to admit at showing how terrible our schools are in the U.S.

The science around this is extremely flimsy, and is mostly supplemented by random strident declarations like yours. Learning is a complex topic even in adults, and influenced by many environmental factors. It's staggering to me that people (usually without an education degree) can casually claim to know a lot about how it's happening in children, who learn at a far faster rate than adults.

Reply on news.ycombinator.com