How do I get total population counts for (1) Kansas City, Missouri (2) Toledo, Ohio, and (3) Atlanta, Georgia?

When I only use the city variable, I get no counts of black males for these 3 locations. Do I have to use the statefip and city variable for these cities?

Due to restrictions that aim to preserve respondent confidentiality, the CITY variable does not identify all US cities. As is detailed on the codes tab of the CITY variable, depending on the sample you are using, each of these cities are not identifiable. For samples in 1990, 2000, and the ACS, the most detailed geographic information available in the public use microdata files is the PUMA (public use microdata area). Cities are identifiable if the majority of each PUMA’s population resided in a given city’s geographic boundary. More detail about the identification of the CITY variable in IPUMS USA is available here.

An alternative source of information for population counts is the IPUMS NHGIS project. This project provides access to previously aggregated tables of US Census data at very low levels of geographic identification. Because the data is not microdata, but is already aggregated, these data are not subject to confidentiality restrictions. From what you have written above, it sounds like what you are looking for should be available here.

As @JeffBloem pointed out, there are serious restrictions on geography due to confidentiality reasons.

You should be getting counts out of other sources. At the moment, you can (still) get them at American FactFinder (although that is going away some time this or next year, and a new platform will be in place of it):

  1. Go to https://factfinder.census.gov/faces/nav/jsf/pages/searchresults.xhtml
  2. Select Geographies -> Places and select your four cities (which is a bit cumbersome, you need to select state and then a particular city in that state)
  3. Select Race - AFF should hint you at DP05 table (Demographic Profile, as far as I understand the Census mnemonics for tables).

Voila.