Thursday Throughput 2/28

Michael Siegel

Michael Siegel is an astronomer living in Pennsylvania. He blogs at his own site, and has written a novel.

Related Post Roulette

10 Responses

  1. Kolohe says:

    ThTh9 – The most remarkable part of sampling to me is that once you get ‘good enough’ – i.e. your sample size is large enough for your total population, getting a larger sample size doesn’t help you that much. Plus, the amount of ‘good enough’ sample size goes up relatively much more slowly than number in the populations you are studying.

    I.e. for 30 thousand people, you need to get survey answers from about 2200 of them. But for 30 million people you only need to get 200 more, about 2400. (but the hard part is making sure the sample is indeed a random selection of the population, and not skewed to any one segment that tends to have a particular attribute.)Report

    • Oscar Gordon in reply to Kolohe says:

      And unless you’ve had a class on stats, the why of that can seem wrong, and cause one to mistrust the stats for the wrong reason.Report

    • Road Scholar in reply to Kolohe says:

      Yep. That’s why political polls so often have sample sizes of around 1000. Doesn’t much matter whether it’s a national, state, or local population.Report

  2. Michael Cain says:

    ThTh9: The ever-popular Monty Hall problem, “Should you switch doors?” Or my favorite bar bet:

    I’ll bet a beer that if we go around this crowd and ask 30 people their birth date, some two of them will have the same date (month and day).

    The odds in my favor are just about 70/30.Report

  3. Jaybird says:

    ThTh1 surprised me.

    I thought it was that Augustus lifted a couple of days.

    But I am left with more questions… WHY DID THEY ADD THEM TO THE FRONT?!?!?

    Seriously, we could have had Seventhmonth be the seventh month. Eighthmonth be the eighth. Ninthmonth the ninth. Tenthmonth the tenth! It could have been BEAUTIFUL.Report

    • Kolohe in reply to Jaybird says:

      It’s quite remarkable that the month naming convention survived 600 years of Republic, 500 years of Empire, breaking apart of Latin into medieval romance languages, the merger between Anglo Saxon Old English and Norman Old French, then another thousand years of English language evolution.Report

      • Road Scholar in reply to Kolohe says:

        What kills me are fundy X-tians that claim our calendar is somehow proof of the Biblical creation story because we observe a seven-day week. Ignoring that most (all?) the days are named after pagan gods.Report

      • Jaybird in reply to Kolohe says:

        If elected president, I will move January and February to the end of the year. The leap year day will be the last day of the year and it will be an official federal holiday.

        If the government is forcing an extra day on you, why shouldn’t it be a day off?Report

  4. Brandon Berg says:

    Since it’s 2019, there will be no leap year. So those of you still on the Julian Calendar will not be falling any further behind.

    This is technically correct, but seems to imply that if it were a leap year, those on the Julian calendar would be falling father behind, which is not correct. The Julian calendar only falls an extra day behind in years divisible by 100 but not by 400. The last time this happened was 119 years ago, and the next time will be in 81 years. Which is why in nearly 2000 years it’s only managed to fall behind by about two weeks.

    Edit: Really missing those HTML format buttons on mobile.Report