Posts by David Hood
Last ←Newer Page 1 2 3 4 5 Older→ First
-
Taken from the walk along to Second Beach, Dunedin
-
I was in New Plymouth on a cloudless day, so set the camera to the fastest settings it could take a picture at to get a deliberately dark foreground and a full range of tone for the sky.
-
[Fixed it] ;-)
-
Upper Otago Harbour this morning. We were driving out to Orokanui Ecosanctuary for a day trip, along the Mount Cargill road, so had not realised the Harbour had some serious fog on it until we looked down to see it.
-
Sacha wrote "Nothing to stop people making up identifications, but I'm not sure why they would bother"
This census, the religion question will be really interesting in that respect: No Religion/ Not Stated is due to overtake Christian this census or next, but if people put Jedi they go in the "Not Counted" category (as I understand it). So if, as is quite possible, the Jedi total is bigger than the difference between Christian and No Religion/ Not Stated there will be all kinds of pointless speculation about the true intent of that category.
-
We went out to the St Clair Esplanade for our anniversary dinner the other evening. While I took this at high ISO and low light so the grain/focus isn't great, I did like the tones of colour.
-
Thanks! I've been a little worried that this might be a little nerdy, even for Public Address :-)
The nerdy end of the audience thanks you for the interesting information. I've actually just been doing an 8 week data analysis course, getting up to date on a few things I hadn't previous exposure to.
I'd say the appropriateness of the correction depends on the strength of the correlation, in as much as the estimate of actual population should be accompanied by an error range that gets smaller the more sure you are of the data. Where it becomes an issue of detail is when the error range is within the zone affecting the number of seats, but in general it is a reasonable way to figure out the true population.
The question for Stats NZ is the consistency- the reasons for when to use one figure and when to use the other. And as others have alluded, it probably falls into the judgments around numerical representations of situations requiring descent vs. situations requiring affiliation.
The orthogonal question is if seat numbers should be calculated on population (in which case the corrected data is closer to reality) or self-reporting (which would lower the threshold. But that is a matter of politics not statistics. Wanting to ponder this, I'm going to think for a few days then probably go off and simulate some census sets and see how the interplay of factors pans out.
-
I was talking to a university class the other day (among other things) about photo adjustment with computers, and how you need to be very careful about what is an acceptable level of manipulation to your audience.
I cited the disqualified winner of the 2012 National Geographic Photo Contest. -
3d printing looks like it is about to go into a long patent/ copyright (patents for the machines, copyright for designs) war period, which may add a few years to the spread rate of it.
-
I think, from memory, that Keith Ng demonstrated the power of Word at the MSD. I would add that everyone reading this has a tool on their computer (a web browser) that can provide unauthorized access to a website if that websites security is effectively non-existent.
I am thinking of those cases in other jurisdictions where people have noticed the web address contains something like ID=1001, wondered "what happens if I change the number to 1002, seen confidential information of other people, reported it as a problem with the website security, and wound up being charged with hacking/unauthorized access to a computer system.
Slightly off topic, but under such circumstances I think the model of "entry" is a very poor one (the metaphor that I am entering the server and looking around). I am making a request of the webserver, the webserver should be verifying who it is making the request to if the information is intended to be confidential. The fault to me lies with the people who set a system that provides intended confidential information to random public requests.
