On the statistical significance of possible variations in the solar neutrino flux