The /newstuff Chronicles #73

[Enjay]

With an afternoon head, rather than a 2 am head, I'm seeing the difference between repetition and increased sampling. Haven't had a chance to check out the link, deep, but will do.

The reason I got hung up on it was we expect students to be able to answer exam questions like "Why did Jim repeat his experiment and take an average of the result?" by writing "He repeated the experiment to make the result more reliable". However students insist on writing "he repeated the experiment to make the result more accurate" which will not get the mark in the exam. Like I said, I just got hung up on that and couldn't see the wood for the trees.

So here you go folks, if you are sitting any science based exam in Scotland this May/June and you see a question like "Why did Jim repeat..."

If you want to throw away marks, say to make it more accurate.

If you want to get a mark, say to make his result more reliable.

[ReX]

"If you want to get a mark, say to make his result more reliable."

Ah, the power. The absolute POWER!

I'm with Nigel here. Accuracy depends on the method, instruments, calibration, etc. and not on the number of samples. You can make a reading (or set of readings) more accurate only by improving the means of making that measurement. Making more measurements, on the other hand, tries to ensure that errors in observation are reduced when considered over several measurements. This is an indicator of the reliability of the measurements, not their accuracy.

[Metabolist]

Well, it really depends on the equipment you use. The equipment I was referring to was the Atomic Absorption (AA) machine, which measures substance concentration in liquid samples by nebulising said liquid and burning it in a flame. The resultant spectrum (processed by the machine) indicates the concentration in milligrams.

However, it is an extremely temperamental and unreliable machine - it needs to be re-calibrated on every sample and even then, there's still a large factor for error. Hell, we even have to adjust the calibration readings depending on those from the control samples (which are of a fixed concentration) all the time and take a mean from them! As five samples are taken from each liquid sample (the obtaining of which I could explain but I'd be here all night if I did that) and a mean is drawn from that, it does, overall produce a more accurate result than one reading (which may be well off the mark). This is advanced, compared to the sort of thing you'd get asked about on a GCSE or probably even an A-level Chemistry paper, so to be honest, it's only really relevant in mitigating circumstances as this. The Work Experience lad we had in last July was completely flabberghasted by it - as was I when I started working there.

So, if anyone wants to argue with me about scientific matters now, then touché to you! ( ;-)

[LeaverAndrew]

NHFL will have different types of maps in it...Sure Ep 1 was all "base" like maps but Ep 2,3 won't b like that...And I will admit that Base type maps are easy for me to build... but If anyone saw Pir 2 map 25 You'll see the I can make a different types of maps...Even Pir map 17 is somewhat Different.

[deep]

I forget about 99% of the details, but remember 99% of the theory:) In this short format, I'd suggest to think "probability" when discussing statistics.

IOW, it's not the average that counts, but the probability that the average does in fact reflect the average of the population you sampled/studied. Enough of that:)

Here's a more techie link that goes into the areas I'm referring to (I know I'm nuts, but it was fun reading it again):

http://davidmlane.com/hyperstat/

You gotta love the terms that are used. How's your Chi Square today?

=========================

Just so we stay on topic:

Damn fine levels - but I play them ALL in ZDOOM! Hurray for the ports.

[Enjay]

Could have done with you at my work today. The entire biology department sat around busting our heads trying to remember all the good stuff about things like Chi squared and standard deviations that we had done at university. One of our senior students needed it for a research project they were doing.

[Guest Ultimate DooMer]

<smartarse>The AA machine doesn't give brilliantly accurate results, as the five samples often give different readings. There is also the question of whether or not the standards are made up correctly. I think the ICP (inductively coupled plasma) machine is better for giving accurate results. OK, so standards are still used, and it does take longer to use but the setup is better (as you can look at the spectra to see if they are giving the right readings) You don't even have to change lamps all the time.
And yes, taking repeat readings improves reliability, not accuracy (as an inaccurate machine will give the same inaccurate readings each time until fixed). </smartarse>

(still can't believe we are discussing chemistry and statistics on DooMworld....)

<doom>PHM levels....so much detail crammed into such a small space. Quite hard as well (esp. with -fast) but watch out for the visplane bug when entering the outdoor area in phm2.</doom>

[inferno_45]

Mind if I ask why you keep uploading Fragport and Operation Lighting? I doubt many folks are gonna want to re-download 3mb for just a few new textures and bug fixes.

[Guest Ultimate DooMer]

Because I don't want someone who hasn't downloaded them yet to get a version with bugs in it. (cos they make certain areas look bad) Also if anyone does a review of them, I don't want it to get slated for containing bugs.