Multivariable Testing and Acuracy

How to pull out

[thecleaner]

If you are running 10 test at once on a website..

For example

test 1 - two headlines
test 2 two ps
test 3 two openings
etc etc ...(10 test total)

and you notice that one of your testing variables is failing horribly...(ie a new headline) Would it be ok to pull that variable in the middle of a large test and leave the other 9 running?

Or will this pull all the data out of wack?

Would you get more accurate results if you pull all 10 then start the test over?

[Receptional]

Depends on the software you are running I would guess.

Whilst the data would be out of wack, I can't see how the maths would go against you if you dropped one mid stream as long as you KNOW that the data is conclusive. You would - I assume - have slightly out of wack data, but only in proportion to the other nine variables.

Infact... I'd guess that the data would be OK on the other nine - but the one you took out would not be statistically reliable result.

Mind you... I haven't tried it!