[RsR] anova

[Iasonas Lamprianou]

Dear friends, 

I have run an ANOVA using a linear variable as dependent and a nominal variable with five groups as independent (factor). The F test is statistically significant F(4, 431)=2.54, p=0.036.However, the multiple comparisons have shown no difference between any two groups (no matter whether I used sheffe, tukeys, lsd etc). What is the interpretation? Do I need a robust anova here?

Thanks
Jason