Tag Archives:

## Percentage calculation problems or The shrunken bodies

Statista is always worth a look if you’re no statistics hater (and speak German). Today’s stats of the day about the question “How tall are you?” (» filtered by sex), and 22358 adult Germans had been asked.

The unfiltered overview shows 2.9% for the really big ones (to which I also belong, thanks to my 190 cm), rounded on top of the bars for clarity:

(190 cm and taller: 2.9%)

You can also enter a number to compare to – and the result is:

98.0% are smaller than 190 cm.
2.0% are like you taller than 190 cm.

Oops, did 0.9% of the people suddenly shrink? Or how else could this result be explained then? And why “are like you taller than 190 cm”?

If I enter 189 cm for testing purposes, I get: “96.8% are smaller than you. 3.2% are taller than you.” So nobody is 189 cm tall? Are 0.3% 189 cm tall and 2.9% taller, or 1.2% 189 cm and 2.0% taller? For 154 cm, the numbers “2.2%/97.8%” are reported, basically matching the bar graph, but here, too, with the words “smaller” and “taller” without mentioning the size of exactly 154 cm.

Well, apparently there’s room for improvement… but the title still says “BETA”. Let’s see if the error report that I sent them (they got a special link for that) will have any effect.

## Radio controlled clock wrong? – Tips

Since people keep coming to my blog looking for hints on how to deal with wrong radio controlled clocks and find only my annoyed post about a particular specimen with its stupid programming that lets is be wrong on the days of daylight savings time changes, I thought I’d put together a few hints for other kinds of radio clock problems:

• Don’t position it too far inside the building, closer to windows may be helpful (preferrably towards the transmitter, wherever that is for your area). This especially applies to buildings with reinforced concrete walls.
• Not too close to other electronic devices, especially those transmitting radio signals themselves. Also, the proximity to metal objects can cause interferences.
• Cheap receivers may have problems with weak batteries, so try fresh ones.
• And you shouldn’t forget to look at the often available symbol indicating whether the clock still receives a signal at all. For if it has to keep running off its own quartz, a cheap model might show visible differences after only a few days.

Further notes:

• Since the signal data usually offers only very little means of error detection (single parity bits for parts of the data; even 2 (or 4 or 6…) bits received wron cannot be detected), low reception can easily cause the clock to set the wrong time, especially if it’s programmed (too) simply and does not compare the signals over several minutes to avoid such consistencies. (The time signal is transmitted every minute, containing 59 bits with 1 bit per second.)
• Most clocks don’t check the signal all the time, but at most every hour or a few times per day – or just once, and hopefully not at 0:00 like the specimen I mentioned.

Hope this helps someone… If you got anything to add, note, or correct, feel free to do so in the comments.

Photo by Martin Schmid – Fotolia.com

## A radio-controlled clock is wrong…

Apart from the fact that in (not only) my opinion we could do without these clock changes – then, though, without the one in autumn, I prefer it when it gets dark later –:

If a radio-controlled clock is programmed so stupidly that it checks for the correct time only once per day at 00:00 (12:00 am) (even though this makes the programming process easier since that’s the situation after inserting new batteries) and thus is wrong almost the entire Sunday, this, I think, really doesn’t indicate a thoroughly working and sufficiently thinking manufacturer.

At least temperature and pressure-based forecast seemed to be correct… But, well, it was only TCM, the private brand of discounter Tchibo.

## Little calculation imprecision of 53%

Update: Im Excel-Team-Blog wurde eine kleine Erklärung veröffentlicht (englisch), es beträfe “nur” die Anzeige, nicht das zwischengespeicherte Ergebnis. Dumm also “nur” für die, die auch mal Werte anzeigen lassen und nicht nur damit rechnen…

Update 2: Ein Hotfix ist mittlerweile verfügbar.

Microsoft Excel 2007, a spreadsheet calculation program for which the correctness of calculations is not quite unimportant, messes up some multiplications that should result in 65,535, e.g.

850 * 77.1 = 100,000

How it gets the idea to turn a result that has the lowest 16 bits (integer representation) all set to 1 into 100,000 (decimal) – and only with some pairs of numbers, not all – I have no idea. Maybe they wanted to hide another bug or flaw this way… anyway, that’s become a complete flop. They want to release a patch as soon as possible…

Maybe it was just a test to see if people are actually using Excel 2007 already and not stick with older versions (which probably many companies do)…?

(via Golem.de)

Update: The Excel Team Blog published a little explanation, according to which “only” the display is affected, not the stored result. So it’s “only” bad for those who have results displayed and don’t just calculate…

Update 2: A hotfix is now available.

## Public Alpha [Update]

Not that I ever were an Apple fan – and not that I thought they’d really care much about Windows – but if you look at what they released as “Public Beta” of their browser “Safari” for Windows (with several Windows-atypical peculiarities, of course), you’ll surely won’t become a fan, and you only shake your head, seeing all these display problems it has.

And not just here:

Page123