Slashdot Mirror

A sample size of 34 is more than enough to establish statistical significance for the effect they measured. It is a common fallacy among non-statisticians to believe that a large sample size is needed for statistical significance, or that a larger sample size is needed for a larger population.

Here is the original paper.

There may be problems with this research, but "sample size" is not one of them.

It happens
http://cdn2.hubspot.net/hub/66...

Your Wikipedia page says you live in Ohio. Based on that the panels should pay for themselves in under a decade without any tax breaks. Do you have massive amounts of shade over your roof or something?

Yes, no, and yes.

Here's a map of solar availability in the United States:
http://cdn2.hubspot.net/hub/18...
I live in that green part, labelled "3.78 to 3.91 kW-hr/m^2/day". Not the absolutely worst part of America for solar availability (that would be the Northwest), but nowhere near the best. Compare this to the dark red part, labelled "6.65-6.78 kW-hr/m^2/day". So, right to start with, solar panels are going to produce about half as much power as they would produce in most of the southwest. Electricity costs in Ohio are about 11 cents per kilowatt hour, which is not incredibly low, but compared to say California, electricity would be 15 cents per kilowatt hour, or 18 cents per kilowatt hour in New York. (But the net-metering cost is about 6 cents per kilowatt hour, and our home electrical demand is low at noon most days, so net metering is probably the price to use).

Ten years is a very optimistic estimate for payback time. My quick calculation is that 11 years would be the rough payback time for the panels alone (purchased at a dollar a watt)-- but not the inverters and regulators, nor the installation costs, both of which will easily exceed a dollar a watt.

But, that might not be entirely a deal breaker if it were not for the main problem, which is that our lot is heavily shaded, and our house has neither a flat nor a south-facing roof.

Poor location for solar.

When you see a graph (from the first linked article) that shows 22% as THREE TIMES LARGER than 19% you know you are reading crap...

TFA: The results are intriguing—even a relatively simple neural network can be used to over-interpret an image, just like as children we enjoyed watching clouds and interpreting the random shapes. This network was trained mostly on images of animals, so naturally it tends to interpret shapes as animals.

Less intriguing: to consider that similar networks (especially once giving "recommendations" to unquestioning end users) might ascribe e.g. criminal propensity or lack of creditworthiness to the odd proverbial "innocent bystander" by over-amplifying distinctions they "think" to have learned.

The "Bad Blue sky" tank detector https://neil.fraser.name/writi... "might be apocryphal" (just like the Obstinate Lighthouse http://www.snopes.com/military... ;-)) but instructive nonetheless.

How does the FDA draw the line between 'must be approved' and 'not our problem' for devices that connect to a greater or lesser degree to other equipment?

I assume that the HDD was drawn in because it was unlucky enough to be an internal part; but with, say, something like this unit, which can burn CDs, transfer to USB devices, or connect to an ethernet network, do you need magic FDA CD-R blanks and flash drives, or is it just things considered 'integral' to the system(even if logically speaking they are no more or less important than something non-integral) that fall under certification requirements?

Depends on which screen you're talking about. For the primary flight displays, they can just be LCD screens connected to an in-panel computer. The FMS and similar stand-alone things are self-contained computers connected to a data bus. However, all of these components are housed in separate grounded metal cases with shielded wiring going to them, so it shouldn't be a problem in the first place.