Relatively simple questions...

I'm analyzing survey data from four different groups.

Group 1, n=249
Group 2, n=216
Group 3, n=147
Group 4, n=33

I ran an ANOVA using a some scaled dependent variables (respondents answer question using a 1 -5 Likert scale). When I run the ANOVA and perform a Tukey test I get the message "Uses Harmonic Mean Sample Size = 87.429." and also, "The group sizes are unequal." The Harmonic Mean of the group sizes is used." I know I'm setting myself up for Type I error, but does this mean the whole analysis is shot? Can I do something to compensate for the low sample sizes in Groups 3 and 4?