Our policy is to not publish the names of living people. We shouldn't illustrate questions with links to un-redacted images. We don't have the permission of the living people on the image, so why should we expose their names indirectly by linking to it? I'd rather see someone clip out a part of the page and redact the names, not unlike what a presenter might do for a webinar where DNA matches are being shown on screen, and the emails and kit numbers and names are blurred out.
See Guidelines for Sharing Information with Others Recommended by the National Genealogical Society, a PDF download, for one example of recommendations for good practice on how to share information with others.
We really need to see the Census record (and surrounding pages) the poster is asking about in order to give an informed opinion on her/his question. It's possible to give general advice without it, but it does make a difference.
I don't think we need to see the image in order to answer this question. The question asks:
Can you offer any insights, or suggestions of sources I could check?
There are two ways to answer a question like this: a "little picture" solution and a "big picture" solution.
Our own personal insight into how often a census household doesn't compare to one we would expect is limited -- unless we are conducting a One-Name study or a One-Place Study, we've only looked at a limited number of pages for any given census. We don't have the in-depth knowledge of someone who has done population studies, or studied the community extensively, or has done the statistics to determine whether or not the census has under-counted the population.
Taking the "little picture" view, I can offer examples from my own research in other census years where people have been enumerated in ways that make Ancestry assign them to the wrong household. I can talk about all the clues I found that made me step back and take a second and third look at the census pages to see what was going on -- but that isn't a satisfactory answer to the question about whether errors in the census are common.
From a "little picture" view, I could advise someone to do newspaper research to see what they could find out about the enumerator for that district. In England, I found several long articles about how the Bishop had chastised the local vicar in one of my study places for not doing his job. It was a 'lucky dip' -- those articles directly answered my question about why I was seeing irregularities in the baptism register.
For a "big picture" view, a researcher can look for population studies that addressed under-counts and other irregularities in the census.
In some of the 'what's going on here' censuses, I discovered that the 'extra' person actually belonged to a nearby household.
In my opinion -- unlike the other genealogy groups you refer to in the comments, it's not our job to solve the specific problem that a person asks in the question. It's our mission to show how to solve that kind of problem, using the example in the question as a case study. We need to think like a teacher and a genealogy presenter as well as being problem-solvers.
So in this case, if I can get the time to find the examples from my own research, I could use as examples in my answer the case where an aunt was mistakenly put with the wrong family because the enumerator wrote the people down out of order, or the nephew who belonged with the family on the next page. I could link to examples in the answer which met our site guidelines of not talking about people known to be living.
The issue for the site as a whole is -- how much will my answer be improved by showing links to the specific examples, instead of just telling you about them? Will it be more help if I snipped out the examples, and included the images? Or would a "big picture" approach be of more use to answer this question?
Yes, we're all human and curious, so we like to see what people are looknig at. But if you aren't familiar enough with a record set to answer a question without looking at the specific image mentioned in the question, it's okay to pass on a question and not answer it.
See How to decode / read US Veterans Bureau Form 7202? as an example of a question where the specific person's information was blurred out. I was able to identify the source and answer the question without seeing the specific person's name on the card.