Would it be correct to use two sample t - test to compare means, if populations of samples differ alot, for example, one sample contains 20-40 elements and second approx. 1000 ?