Hi Jacqui,
I agree with Hester. It's acceptable to have missing data as a result of
plausible range checks, and the age is clearly wrong (although depending
on what your study is for, you could suggest an automated check of the
age field on entry as a future improvement).
If your whole study is focused on age and it would mean excluding them
completely from all analyses, you could consider a substitution, and
running it with and without them in - I'd probably be inclined to go
with the median (or even the mode) rather than the mean as your 54 year
old may skew the mean (assuming 54 is old for a student, rather than the
age at which students can afford the fees). But if it's only one
participant it's hardly worth the effort.
Brian
On 20/Apr/2011 12:12, Jacqui A Hart wrote:
> Hi all
>
> This may seem like a really silly question, but I don't know the best way to deal with it.
>
> I ran an online survey via the university intranet (accessible only by students). One respondent recorded their age as 10 years. Clearly this was an input error (to the best of my knowledge, we do not have any students as young as 10 :0). The range is between 18yrs and 54yrs. Should I change the item and input the mean?
>
> Thanks in advance
> All the best
> Jacqui
|