I wrote: Total federal employees 1.8 (domestic & overseas)
Should be: not "1.8" but "1.8 million". Because US capitalist society is a competitive society based on business and government secrets, rivalry and lack of nationwide coordination or planning in most fields (i.e. the "closed-off society" model where you need homeland security), there is also a proliferation of statistical surveys and data sources using different categories to quantify the same type of phenomena - and not one integrated system, based on nationwide standard classifications, permitting a "data warehouse" that permits all analyses that anybody might like to do. There is a certain amount of overlap in classifications, between different surveys, but also divergences in the survey instruments used. The advantage of that reality is that you can get different angles on the same subject, such that one survey might correct the limits of another. Say that, for example you want to measure a flow of funds collected by the government, and redistributed by the government to citizens, let's say, some kind of social assistance benefit. First all, a worker might authorise the employer to deduct that money from his wage and pay it, and receives a confirmation that it has been deducted in his paycheck. On his tax form, he will declare what he has paid. The employer pays the money from an account. Then the federal government collects it as income. The government places the money in an account. The government charges administration costs. Then the account managers allocate the money to a specific budget. That budget has again got an account. Then the funds are authorised to be spent. Then the federal government transfers funds to the state government. Then it's federal expenditure and state income. The state government then puts the money into an account, and allocates it to a budget, charging administration costs. The money is then authorised to be spent for the designated purpose. Then it is authorised to be spent on a specific person who is eligible to receive it. Then it is transferred to a private back account. The state records an expenditure, and the bank account records an income. The bank charges an administration cost. Then somebody from a household records that that money as an income. Then th person uplifts the money. Then that income is again declared for tax purposes. Finally, the money is spent again to buy a designated good or service priced by the supplier. The supplier receives the money, and places it in an account. The economist then calls this "the market" (it doesn't really matter whether we're taling about a corporation or a government, the principle of the thing is the same). Clearly then, as a statistician, you could attempt to survey that financial flow at dozens of different points, using administrative, personal, household, financial, bank and commercial records and accounting statements. In addition, you could also use data on other associated financial flows, to estimate the magnitude of this flow, because you know it must be within a certain range, within certain limits. If you started out with 1,000 dollars, then you cannot end up with a million dollars or 10 dollars in most cases, unless there is some very creative accounting occurring, a bit of "magic". Or, you could utilise surveys already collected previously for estimation purposes. This insight is sometimes used in the investigation of frauds and corruption: by tracing a financial flow through and cross-referencing, we can discover whether money suddenly "disappears" or "mysteriously appears" out of nowhere. The problem with a system which relies on buying and selling for its morality is that the number of transactions is so great that it is almost impossible to verify all transactions. To secure a money-price morality is impossible, all you can really to is trust in God and hope for the best. It's not that the information is not there, it is rather the quantity of information and the fact that money in circulation can be used in transactions which escape accounting scrutiny. The disadvantage of many different types of information records and statistical surveys on the same phenomenon is, that it is frequently difficult to get believable data, you really have to compare different data sets on the same topic with respect to concepts and methods, in order to assess the accuracy, validity and relevance of statistical descriptions. For that purpose, you need to know a lot about the available information, and cross-check different observations of the same thing. This is not necessarily an argument about data quality, but an argument about the validity and relevance of statistical descriptions. A statistician may conscientiously seek to obtain the best measurements possible, but because of the conceptualisation of his measurement units, he may still miss something. He requires good qualitative knowledge, before he can obtain good quantitative knowledge. Obviously, any discussion about "the division of work" is highly contentious, quite apart from measurement issues; this description could conceptually be output-based, occupationally based, employment-based, income-based, institutionally based or personal characteristic (gender, ethnic, cultural affiliation etc.) based, and then we could group individuals in numerous different ways non-arbitrarily on the basis of their real social and organisational relationships, institutional structures, social practices and so on. It is also possible for instance, that (1) far more children under 16 have a regular paid job than I have estimated, or (2) that I underestimate the number of fulltime housewives without paid work outside the home, or (3) that I underestimate the magnitude of administrative work. I don't claim to make a complete or exact analysis, I am merely providing indicative figures to illustrate the nature and scope of the topic and try to be as objective as I can, without being able to assess all the intricacies of American social relations from behind my PC with the resources that I have available. As regards religious organisations, I found another estimate (group 813110) which suggests there are about 164,000 religious organisations in the US employing 1.5 million people altogether (managers, employers, employees, self-employed, unpaid workers and so on). But in an occupational classification, which I used in my brief analysis, the number of workers would be a lot less, because their "main occupational function" would in many cases not be to propagate/organise religious worship or activity or minister to religious needs, but rather provide a service of some kind that is classified in another way occupationally according to the type of service it is. Hence, different conceptual systems are required, where factors treated as constants in one survey are treated as variables in another, or so that dependent variables here are independent variables there, or that the grouping of observations is carried out using different theoretical assumptions or concepts. Jurriaan