I think it's important to remember that for genealogy, the historical records we use were created for another purpose. Understanding that purpose is key to getting the most out of the information inside our sources.
You can learn more about the 1940 Census by looking at research guides like the offerings at The US Census Bureau or the US National Archives, or by reading the reports at IPUMS-USA. Articles like Claire Kluskens' "Who Talked to the Census Taker?", while focusing on earlier census years, can also provide insight.
Have you read the enumerator instructions? You can download a PDF from the Census Bureau or read them online at IPUMSUSA.
Have you researched the enumerator for that neighborhood? Do you see irregularities in the other entries for that ED? US Newspapers are full of gossip, and articles can give clues about whether enumerators were of sound character, or may have issues that would interfere with their attention to detail.
Most important of all -- examine your own hidden assumptions. Everyone makes assumptions without being aware of it. Part of being a good researcher is dragging those ideas out into the light of day. Without being aware of it, we might assume:
- that a record set is complete
- that a record set is accurate
- that the enumerator always followed the instructions
- that the informant understood the question that was asked
- that we understood the role of the enumerator
Today's Genealogy Tip of the Day In What Capacity Does Their Name Appear? is relevant to an enumerator as well as notary public.
Sometimes the only way to set your mind at ease about unexpected things in a record is to researched thoroughly the people involved in the records, starting from scratch, with no preconceptions. I've seen several examples in my own census research where it quickly became evident that these "surprise" people belonged with families who are enumerated on the same sheet or on nearby pages.
In an obituary, I found a list of bearers at a funeral who were listed as nephews of the deceased, but my prior research had already shown that they were the deceased's grandsons. If I had come across that obituary first, trying to prove that they were nephews would have sent me off on a long, fruitless line of research.
Because of these experiences, and others, my standard procedure is to assume that the person may not have the relation to the head of household which is stated in the census. I ask "How can I find out more about people named [name] born in [location] around [calculated birth year]?" and go from there, starting with finding out how many same-name candidates there might be for that question. I do this in addition to all the good suggestions that were made in the previous answer.