PLease, help, t-test or something else? And how :)

#1
HI!

I have a huge problem. Those are my results from experiments: animals which ate water *i.e. control group) has these value for amount of enzymes:
1857
6218
5114
6972
1820
4237
Animals which ate food (ie test group has these values:
1859
2277
2260
3847
1632
17155

I checked for outliers and 17155 i excluded.

So, my mail concern is which t test would be appropriate?

I expect that enzymes are increased if animals ate food, but if I remove outlier (17155 in test group, 6972 in control group is very close, but not outlier yet ) then mean is lower in treated than in control group. By two tailed t test there is no difference, but by one tailed there is difference, and from that point it is absurd because it means that test group has LOWER amount, and common sense says it should be higher.
The same thing is happening no matter whether I use Welch correction or not.
Please, any advice, help or suggestion?

Thanks a lot.
 

Dason

Ambassador to the humans
#2
Is there any way to learn more about the outlier?

Also - have you considered the possibility that your common sense is just completely wrong?
 
#3
outlier was checked by grubbs test (and that animal wasnt quite happy during experiment).

and I am considering that my common sense might be completely wrong, but for me is seems that that is not the case in this situation. :)