While working with the H056001 variable (Median Contract Rent [dollars]) at the census tract level from Summary File 3 of the 2000 Census, I noticed there are "jam values" in the data of "2000+" and "100-". It is unclear to me whether a jam value of "100-" represents 100 or less or 99 or less. I do see a single tract with a value of 100 in the data, so I expect the jam value of "100-" to represent values of 99 or less. However, there are also values of 0 in the data. Should these values of 0 be treated as missing data/suppressed data? I haven't been able to find clear documentation for these 0 values and what they represent. I am leaning towards the idea that they should be treated as missing given the Universe for this variable is Specified renter-occupied housing units paying cash rent (i.e. a median of $0 for cash rent is not realistic).
I also looked at the b25058_001 variable (Median Contract Rent [dollars]) at the census tract level from the 2008-2012 5-Year ACS and the 2013-2017 5-Year ACS. In these data there are also jam values of "2000+" and "100-". There are no values of 0, but there are values of "-". I'm assuming a value of "-" means the estimate was unavailable or suppressed?
I submitted this question to the ACSO users support email and received confirmation that for the median contract rent variable in the Census 2000 a value of 0 is used to represent an insufficient number of cases while in the ACS a value of "-" is used to represent an insufficient number of cases.
It appears this is something that users should also be mindful of when using the median home value (h085001) and median household income data (p053001) from Census 2000 as well (i.e. those variables are bottom-coded at "10,000-" and "2,500-" respectively, but there are also values of 0 in the data). Those values of 0 do not represent $0, but rather represent an insufficient number of cases.
That does seem pretty sloppy to put a value that can be mistaken so easily.