字幕列表 影片播放 列印英文字幕 This video was made possible by Brilliant. Learn complex topics simply for 20% off by being one of the first 200 to sign up at brilliant.org/Wendover. This is one block in one city in one state of the United States. The US Constitution's Article 1, Section 2 prescribes something that must happen on this block: “Representatives and direct Taxes shall be apportioned among the several States which may be included within this Union, according to their respective Numbers… The actual Enumeration shall be made within three Years after the first Meeting of the Congress of the United States, and within every subsequent Term of ten Years, in such Manner as they shall by Law direct.” What that means is that every ten years, the US government must determine how many people live on this block, in addition to how many people live on every other block in every other city in every other state and territory of the United States. They must determine the population of the country—a straightforward task, but one that thousands of US Census Bureau employees spend a decade preparing for, and a year and $15.6 billion executing. It turns out that counting more than 300 million people isn't exactly easy. Instead of looking at how the Census counts every one of those hundreds of millions, lets rather look at how they count the couple dozen people living on this one city block. The Bureau started preparing for the 2020 Census on November 1st, 2011. They have a staff of about 4,000 who plan and coordinate each Census, along the their other programs, and then, every ten years, when it comes time to actually conduct the Census, they employ more than half a million other individuals temporarily. The 4,000 full-time employees, though, spend much of the decade forming their operational plan, the first draft of which was done in late 2015. They then spent the next three years refining that. A significant element of this plan includes deciding where the Census' field offices will be located. You see, the Census divides the country up into six regions which are headquartered with permanent offices in Los Angeles, Denver, Chicago, Atlanta, Philadelphia, and New York City. Then, within those regions, the Census opens up 248 temporary field offices to conduct local operations. These field offices are distributed based on density so Colorado, for example, only has four while New York, on the other hand, has 21. Each of these offices has to hire hundreds or thousands of enumerators to go out into the streets and conduct the Census. Finding half a million temporary workers is, of course, incredibly difficult and it's even more difficult for the 2020 census given that the country currently has the lowest unemployment rate in recent history. In order to attract those half million people they do have to account for the drastically different wage conditions across the country. Enumerators in Mesa County, Colorado, where our block is located, for example, are paid $16.00 per hour which, for this area, is a pretty decent hourly wage—especially for a low-skilled job requiring few qualifications. Of course, if you offered that pay in New York City, on the other hand, it would be far less attractive so the pay there is upped to between $25.00 and $28.00 an hour. Despite the decent pay, though, the Census still always has enormous difficulty filling all their positions so they spend years teeing up all these workers. The first of the temporary hires for the 2020 Census happened in mid-2018. That's because, before the Census can go and ask every household on this block how many people live there, they have to figure out where each of these households is. In the past ten years, new houses could have been built, structures could have been subdivided, and plenty more could have changed so this is a crucial step for accuracy. In addition, if a homeless person is living in this alley, if someone's living in a motorhome in someone's driveway, or if there are any other abnormal residences, this too needs to be counted. Previously, the Census would have sent someone to walk this block, in addition to every other block in the country, and mark down where each residence is, which clearly took an enormous amount of person-hours. With the 2020 Census, though, they're shifting technique to use more and more satellite imagery to identify residences. Of course, that's not perfect. For example, it's hard to tell if this is one residence or two, so for any areas that they're not sure about they'll still send people to check in person. At the conclusion of this process, at least in theory, they should have a list of every single place where people live in the US. Also in 2018, the Census Bureau had to submit to Congress the questions they planned to ask the American public. The Census is incredibly simple—after all, it has to be if they expect everyone to answer it. It asks four background questions, five about the primary resident, then seven about every other resident of a given household. Despite this, the Bureau spends an enormous amount of time debating how to ask these questions. Question selection is so complex that they published a 98-page document explaining how they selected the 2020 questions and their wording. For example, in question three, for the non-primary residents, they added specification on whether their relation to the primary resident was as a same-sex or opposite-sex couple. In the past there was the option to mark the relation as a spouse or unmarried partner, but there was no distinction between same-sex or opposite-sex. They did, however, calculate what percentage of couples were same-sex based on how they responded to the question asking if they were male or female. There was an issue with that, though. Say there's a married couple living in this house consisting of one man and one woman. If the woman accidentally marks her sex as male and also marked her relationship as a spouse, they would be counted as a same-sex married couple. Given that the overall proportion of Census-surveyed couples who are same-sex is small, these mistakes could significantly skew the data. With the redesign of this question, it acts as a failsafe so that even if the woman in this house marks her sex as male, they'll still be counted as an opposite-sex couple. These are the tiny details the Bureau has to think about for each and every question and change. They have to think about how to best structure them to get the best possible accuracy. Once they have their questions, one of the final things the Census needs to figure out is how they're actually going to ask for responses. Our block is located in Grand Junction, Colorado, which is a small city, so it, along with most of the country, simply receives a mailed invitation instructing the recipient to complete the census online. This works well for most of this relatively urban area, but not for all of it. For example, residences in this small section of Grand Junction will receive both an invitation to complete the Census online and a paper Census form. That's because it's home to a number of assisted living facilities and therefore a number of older individuals who are far less likely to respond online. About 22% of the country is classified by the Census as areas where individuals are less likely to respond online and therefore receive paper Census forms as well. There are, though, areas where both of those options won't work. Nearby Eagle County, for example, has a population where one-third are of hispanic origin and plenty of that population has limited or no English-language skills. Therefore, residences here receive a bilingual invitation to complete the Census online. There's then one more category, used in areas like Costilla county where there is a large Spanish population and limited internet access, where bilingual paper Census forms are sent out. The Census classifies the entirety of the country into one of these four categories in order to have the best possible chance of getting an answer. Once offices are opened, employees are hired, residences are logged, questions are written, and letters are planned, it's finally time for the Census to start in earnest. The first Census responses cannot happen until 2020 starts, and all responses must be in by the time 2020 finishes. They have exactly 366 days to count every single person in the US, and deliver that count to the US Congress. Our block in Grand Junction is far from the first place to be counted though. That happens in Toksook Bay, Alaska. Census workers arrive there in late-January, 2020. Toksook Bay, along with 200 or so other rural Alaskan towns, are exceptions in the overall timeline of the Census count as, in mid-Winter, when the water and ground are frozen, one can get around relatively easily on snowmobile whereas, later on, when most of the country's Census counting is done, it's very tough to travel when the ice and snow are melting. Down in the lower 48, the seasons are far less of a concern so in mid-March, 95% of American households receive a mailed invitation to fill in the Census online. That includes our block, but if the earlier investigation found someone living in a mobile home in someone's driveway there, they would likely get their invitation hand-delivered. Mobile homes, residences in areas with recent natural disasters, or households that use PO boxes don't get mail delivered to their door so that's when the Census uses this hand-delivery technique, which is used in about 4% of cases. A small other proportion—less than 1%—is visited and surveyed in person from the first instance, mostly in the most remote areas of Alaska and Maine, in addition to on some American Indian reservations. On our block, like most others, if households don't respond to the first invitation, they'll get a reminder letter, then a reminder postcard, then a reminder letter with a paper questionnaire, then one more reminder postcard before the Census gives up on the household responding by internet, mail, or phone. Typically, about a third of the houses on this block will not have responded on time, even though it's a legal requirement. Non-respondent residences on our block will get a visit from an enumerator, but they might not be the first to get visited. In-person visits are prioritized strategically. For example, knowing that college students will start move out from their dorms and student housing starting in mid-May, they send enumerators to non-respondents at college campuses first. Each of our houses will get three visits, then if they don't respond on the fourth, they'll also try to ask neighbors if they know the details of the residents of the given household. They'll try both the household and the neighbors again for the fifth and sixth try, but then, if the household doesn't respond after five mailings, six visits, and three knocks on the neighbor's doors, then the Census finally decides that the residence must be uninhabited. Of course, in addition to counting everyone in every household, they also have to count people who don't live in a household. For example, to count homeless people, the Census will set up at shelters, soup kitchens, food vans, in addition to just going out into the streets. Another challenge is counting people who live and work on American ships who don't have a permanent residence. The Census runs a special program for this—the Maritime and Military Vessel Enumeration—where ships are sent Census kits including customized forms with different questions targeted at seafarers. These are all then sent back to the Census to be included in the count, and a variety of techniques are used to assure that nobody is double-counted. Throughout the second half of 2020, in-person visits will finish up, special operations will conclude, and the Census will get into a huge number crunching operation to figure out, with as much accuracy as possible, what the final count is. They not only have to give that total number, but they also have to break it down by state, by county, by town, by race, by age, by relationship status, and by much, much more. It takes some time, but there's a very, very specific deadline—December 31st, 2020. The US Congress needs to know the count by then in order to re-allocate seats in the House of Representatives proportionally to the population of the states. In the 2010 Census, 308,745,538 people were counted as living in the US. In 2020, the number is expected to be around 330 million but, to know for sure, we'll just have to wait for December 31st, 2020. One of the big jobs at the Census Bureau is as a math statistician and for that, according to their job listing, you need to be adept at things like number theory, statistics, probability, algebra, and calculus, all of which you can learn or brush up on with Brilliant. Each of their courses for each of those subjects is interactive, wonderfully designed, and helps you learn complex topics like calculus not through rote memorization, but rather through breaking down the concepts into small, intuitive chunks and then building it all back together. This is how you can truly understand calculus or number theory or statistics, rather than just learning the process to solve problems. If math isn't your thing, they also have dozens of amazing other courses on subjects like scientific thinking, quantum mechanics, computer science, and plenty more. To support Wendover Productions and learn more about Brilliant, head over to brilliant.org/Wendover and, the first 200 to sign up there will also get 20% off.