On the 1881 census she is 26, (birth year 1855)
No. The birth year would be either 1854 or 1855
In 1891 she is 35, (birth year 1856)
1855 or 1856
In 1901 she is 48, (birth year 1853)
1852 of 1853
In 1911 she is 57, (birth year 1854)
1853 or 1854
I have her marriage certificate to a Charles Wright they were married in 1876 and she states her age as 21, (birth year 1855)
1854 or 1855
and on her death certificate in 1920 she is 65, (birth year 1855)
1854 or 1855
Perhaps you are correct in assuming she was born before 1855
I am assuming nothing. Merely pointing out that any age is always the age a person was (or believed themselves to be, or claimed to be) on a specific day, and that they may not have actually had their birthday yet in that year.
The census is usually taken about a quarter of the way into the calendar year, about the end of March. So anyone whose birthday is later than this will not have had their birthday in that year. So 'calculating' a year of birth by subtracting the stated age from the year of the census gives a wrong result in three out of four cases.
Apart from the 1891 census, it does look as if she is likely to have been born before the critical date, 1 January 1855, when statutory registration began.
On some of the census her Mothers name is Jane on another census her Mother's name is Ann. I personally think Jane is correct.
Have you viewed the originals of all the census to make sure that you have personally trancribed all the details? Sometimes transcribers make mistakes, but then so do enumerators and the people who provided the information in the first place.
Or could it perhaps be that her mother Jane died and her father married an Ann?
If you could find a birth record for Margaret would her Mother's name be recorded?
Normally, yes, baptism records in Scotland in the 1850s do include the mother's name and maiden surname. They may or may not state the date of birth.