There might not be enough data to estimate all the coefficients in the model.

How many coefficients does your model have?

What is your sample size?

What degrees of freedom of the model are reported?

Hi, it was my pleasure to answer your question, however the step to answer your question would be:

- What kind of variable you are using? (continuous variable, dichotomous ...)

- Is your dependent and independent variable parametric or non-parametric?

- Have you tested the normality of the variables?

- Why not apply a linear regression test instead of averages comparison?

I need more details to answer your question, because if your dependent variable has normal distribution you can use a prediction model showing the relationship of dependence between variable "y" and independent variables.

Best wishes.