Posts by Bart Janssen
Last ←Newer Page 1 2 3 4 5 Older→ First
-
Hard News: How to make more awesome use…, in reply to Russell Brown,
goodies for people who did that
Regular Colin photos
-
Hard News: Briefing, blaming, backing down, in reply to linger,
We should expect that the more specific factors, which have a more direct causal connection, will be more consistently associated with outcomes than class size per se. To that extent, the results are not “odd” at all.
But the only way you would get such a large difference between class size itself and some of the factors directly linked to class size is if decreasing class size had some negative component as well. Especially if multiple factors dependent on class size have a positive effect, as in this case. I agree there will be aspects related to class size that should have a higher impact but not to the degree seen in the analysis.
This suggests to me that the data is a merge of multiple studies where most studies measured class size but only a few studies measured some of the other factors. In order to present all those factors in one table the data would need to be processed. Without very careful data processing such differences could (and my guess in this case have) distort the findings.
But I really don't know. I'm not an expert in education nor is my stats good enough to tackle meta analysis. However I have seen numerous meta analyses fall apart under scrutiny after they were published making me very nervous of such approaches. Personally I'd prefer to treat the individual studies as independent. It takes more effort to assess the conclusions of independent studies and more time but in my experience it is well spent.
As the excerpt that Ross posted shows Hattie is not unaware of some of the issues indicating eg that the low impact of class size may be due to failure to alter teaching practices to take advantage of smaller classes.
-
Hard News: Briefing, blaming, backing down, in reply to James,
These numbers are being quoted as if they’re regression coefficients
What's worse is these numbers are being quoted as if they are real. I personally think the meta analysis is likely to be shown to be a crock of shite. There are some serious red flags in the list that suggest the analysis has created something odd.
-
Hard News: Briefing, blaming, backing down, in reply to Scott Chris,
we have no reliable method of assessing teacher competency
Are you sure? My guess is that such measures exist, they may not be simplistic enough for politicians or herald editors though.
Kind of question it might be worth asking a graduate with a degree in education.
-
Hard News: Briefing, blaming, backing down, in reply to Rob Stowell,
Rapid and relevant individual feedback (1) and direct instruction (2) … are both dependent, to some degree, on class size….
And that is the problem with such studies. For most of the factors listed there are strong interdependencies. Any statistician will tell you that the moment you have variables that depend on each other you have a statistical nightmare in the making.
That's why a huge amount of effort is spent in science to isolate a variable before you even do the experiment, let alone analyse the data.
But in the real world and particularly in experiments with humans it can get a tensy bit tricky to do proper controlled experiments. And for some reason it's even worse with immature humans. I don't know ... the life of a scientist is rough when the public won't let you deliberately harm some children for the benefit of science. How's a scientist going to get a proper evil reputation, I've even practiced the laugh ...
Hence meta analysis. Where data from many different studies with many different aims and many different controls and many different measures all get added together in one big dataset and then squished under the weight of ever more MATHS until something gets squeezed out. Sadly the history of meta analyses is that what gets squeezed out is usually less than savory. But we keep trying because if we could just get the math right then we could make better use of all that data.
Sarcasm aside. One thing that is really clear from education studies is that spending money on training teachers throughout their career (not just at the start) really makes a difference to students. OMG! See science is useful. It's also clear that a bunch of other things help too, many of which are easier to achieve in a smaller class size but do not specifically require a smaller class size.
What is interesting is that class size has such a small effect when many of the things that arise from small class sizes have a greater effect, eg one-on-one attention. What this strongly suggests is that the MATH used in this analysis might have been a bit buggered up. But, thank god, I'm not a statistician and so I don't have to try and figure out where they went wrong. I can just go to the individual studies themselves, or better yet take the advice of the experts in the field - ya know the ones with genuine degrees in education - and accept their conclusions.
Which are: Smaller class sizes are good for students; teacher training is good for students; both of these together are even better for students. Do both.
-
To return to the original point, why would anyone invest in improving NZ export industries? There simply are no compelling industries worth the investment. It's all very well to sound off about the overseas investors but they responding to the lack of genuine competitive advantage in our export industries.
Without a real advantage there is always another country more convenient. We have cheap water and cheap energy (if we bothered to generate it) at the moment that is translated into milk and not much else.
Creating an culture that allows people to innovate and develop novel industries in NZ and is also accepting of failure is the only way to create anything worth the investment. Until we commit to doing that we will remain small and indebted.
As an aside I think one of the problems is that the NZ stock market is still untrusted and some would argue untrustworthy.
-
Speaker: The Voyage: The Engine Room…, in reply to Martin Lindberg,
Umm, yes?
Really. I store bytes on a secure server. I then can manipulate those bytes to transfer the "value" to other people. Not sure I need a bank to do that really. I do need a secure network and secure servers.
Real physical money ... well typically I get $200 cash and put it in my wallet and it lasts 2-3 months.
-
Speaker: The Voyage: The Engine Room…, in reply to Stewart,
where services (& especially financial services – spinning gold from straw) are the primary wealth ‘generators’
You hear this a lot. And while I agree there is some truth in it I wonder if it isn't in part rationalisation for people getting paid large amounts for jobs that don't really contribute to the wealth (social and financial) of the nation.
-
Speaker: The Voyage: The Engine Room…, in reply to Angus Robertson,
New banks will arise
Interestingly they may not look like banks. A lot of the functions that banks perform now can be carried out by other methods now. Do you need a bank to send money overseas? Do you need a bank to finance a new venture? Do you need a bank to store your money? Do you need a bank to pay your staff?
I think one of the issues around the financial institutions is that as technology has advanced it has made them in many ways irrelevant.
-
Rocket scientists got jobs at investment banks
It's somewhat telling that we (the voting public) chose an investment banker as PM and not a rocket scientist. We value the opinions of bankers and economists much more than the opinions of engineers and scientists.
Why would anyone want to work in the engine room when all the glory is on the main deck?