I have a researcher who was looking to use NHS commuting flows table, "Commuting Flow - Census Subdivisions: Sex (3) for the Employed Labour Force Aged 15 Years and Over Having a Usual Place of Work, for Census Subdivisions, Flows Greater than or Equal to 20" (99-012-X2011032) for one of his classes this year, but he's not sure he can trust the information he's seeing.
The CSD he brought to my attention is the village of Gagetown (1304005). First off I'll tell you that it has a population of 698 and is suppressed in the NHS profile site. Now to the juicy stuff. When my researcher uses the above mentioned commuting flow table and looks at Gagetown VL as POW (GNR 36.1) six place of residence CSDs appear. It is at this stage that I have the question: how do we interpret these numbers (see table below)?
POW (Gagetown VL)
Lincoln P 60
Burton P 95
Oromocto T 410
New Maryland VL 20
Fredericton CY 55
Surrey CY 20
The obvious interpretation is that there are roughly  people who commute to the village of Gagetown for their usual place of work (plus any numbers which didn't make the 20ppl/CSD threshold); however, that doesn't make a lot of sense when you think about the reality of the situation. First, as I mentioned above, there are about 700 people who live in the village and presumably some of them work there, too. Then when you look at the patterns above it seems even more suspicious. My researcher and I wonder if 5th Canadian Division Support Base Gagetown, formerly and still locally known as CFB Gagetown, got confused with the village of Gagetown. Are we just misinterpreting the numbers or does this seem possible?
In the NHS questionnaire, question 46 asks "At what address did this person usually work most of the time?" [their emphasis]. There are instruction to use the city or town rather than the metropolitan area of which it is part, and the options were "City, town, village, township, municipality or Indian reserve." There is also a note indicating that if the place of work is different than the address of the employer (example school teacher works at a school rather than at a school board), one is supposed to indicate the address where one actually works. However, there is some question as to whether or not CFB Gagetown is in Oromocto, where I think most of the administrative offices are, or in Burton Parish, where a good part of the training area actually takes place. I would be willing to bet that a good number of people who answered the question could have easily entered 123 Sesame Street, Gagetown, NB. What I am not willing to bet is that there were between 15 and 25 people from Surrey BC who worked in the village of Gagetown, or even that there were  from Oromocto and  from Fredericton who commuted to Gagetown VL.
I know that random rounding can affect numbers, but this seems a little more complicated that that. Has anyone else noticed these kinds of issues? Hopefully we're just misunderstanding the table. What my researcher is really worried about is that he can't trust a number of the other NHS tables he's had occasion to question.
Luckily Gagetown P doesn't seem to have been mixed up in this schmozzle.
We reported your observation to subject matter for more information. Here is the response we received from Subject Matter.
We have examined further data related to individuals working in the village of Gagetown. Most respondent do work on CFB Gagetown and not in the village of Gagetown. This error has been flagged for future processing and coding tools such as Reference files will be corrected.
On another note, data users must be aware that the Place of residence is the location where respondents are enumerated and do not necessarily represent where respondents are located at the time of the survey. Therefore some commuting flows will be unusual such as a Place of Residence in British-Columbia and a Place of Work in New-Brunswick.