The following is a message just posted on the State Data Center listserv. Chuck Purvis, MTC ****************************************************************************************************** To All Leads, Coordinating, and Affiliate Agencies: Below are two questions that were posed regarding the County-to-County Worker Flow files. Phillip Salopek of POP division has responded to them. Question: Does anybody know why a zero value would show up in the commuting patterns file? For example, in the "live in?work in" file, the tables tell me that of the workers who live in Choctaw County, none work in Baldwin County. If nobody who lives in Choctaw County works in Baldwin County, why does it appear in the file at all? This is a minor point, but there has to be an explanation for why the files were constructed that way, and I am curious. Answer: There are some flows of size zero that show up in the Census 2000 County-to-County Worker Flow files. They should not be there. If they had been noticed prior to the release of the file they would have been deleted. These flows result from the fact that there are records in the source files for which the person weight is zero. Although this is relatively rare, it does happen and is not a mistake. One explanation is that weights may be set to zero in cases where an area was oversampled. There may also be other reasons for weights to be zero. At any rate the zero flow records should be ignored. Question: I'm surprised about the lack of buzz on the SDC Listserve in regard to the commuting (county to county) numbers. In NH at least, the numbers are positively wild. We have more people commuting to Ariz, Calif and other western states than we do commuting to some NH counties! I know some airline pilots do often commute very long distances, but NH does not have that many pilots in total. Is there a rationale for continental commuting distances? I'm not talking about 2 people to CA and 4 to AZ. I'm talking about 5 to 40 people "working" in more than a dozen western and mid-western states. Answer: The Census 2000 County-to-County Worker Flow files contain data derived from the long form question, "At what location did this person work LAST WEEK?" If the person worked at more than one location they are instructed to print where they worked most last week. Thus, these data are tagged to a particular reference week. People are not being asked their usual workplace location. What these data show is that during any given week, work arrangements are highly variable. It is likely that most of these long distance commutes reflect normal business travel. A person in another city on a business trip for 3 or more days during the reference week would show up in the data as one of these long distance commuters. Some of the phenomenon may also be due to workers in the military, as well as people who may be temporarily working far away, perhaps staying there temporarily or during the week and returning home on weekends. While there may not be much concern over the number of residents of a county who seem to be away from their normal work location any given week, these numbers can be quite valuable for the receiving county. It may be very useful to know that there can be a significant number of people working in the area who are not reflected in other data sources because they don't live in the immediate vicinity. If you have any further questions, please contact Phil at: Phillip.A.Salopek(a)Census.gov. Thanks, Renée Jefferson-Copeland Customer Liaison Office Census Bureau