The census pages are littered with words crossed out; words added, and various markings that often obscure the enumerator's valuable words. We have to make sense of them.
The alterations fall into two groups. Those that seem to have been made at an early stage, perhaps by the enumerator or the registrar. Typical of these are where the enumerator has described someone as a pauper. This was not required to be entered, but it often is. It is then crossed out; either by the enumerator or the registrar, who was supposed to check the returns before they were attested to and sent off to London. With these alterations, the data should be entered with a note stating that it was crossed out. If you can get this into the occupation field, well and good, if not, enter it as a note. Where records (including whole households) are crossed out, enter them with a note. Even if they should appear elsewhere; they may not - and what is lost?
The second group of alterations appear to have been made much later; probably by the clerks in London. Typical of these are those found in the place of birth field. They have inserted a county where only a place name is given. Sometimes it is obvious they are right, other times you wonder how they could possibly have known. We stretch the "as is" concept here, and you can enter the county they suggest if it looks right to you. When they don't know the answer, they often write ALS; I have no idea what this means. Sometimes, when the age is given as N K, someone has entered an age in round numbers next to the age column. This is best entered as a note as we have no idea who entered it and why. One of the reasons for the census was to work out the size of the agricultural and other industrial sectors. By 1891 there was quite a complex system of classification; they can often be seen in the occupation field; we ignore them.
Transcribers & checkers have to make a judgement on what is to be included. If information looks good but is crossed out, enter it with a note. If it looks like a later addition, ignore it except where it is covered above.
Finally, there are the markings that add nothing to the data. Typical of these are the annoyingly thick strokes through ages; presumably made as the clerks grouped the population by age. We just have to work through these.