A correlation of 15% is significant in any statistical population (of sufficient size which the studies had). When there is zero (or nearly zero) correlation for left-handed homosexual men with older brothers and zero (or nearly zero) correlation for handedness for homosexual women regardless of age, sex or quantity of siblings, I view the right-handed-homosexual-men-with-older-brothers rate of 15% to be a strong correlation. Far lesser correlations are justified by state insurance departments for setting differential rates for insurance premiums, or even among meteorologists for predicting the amount of snow that will fall in a winter storm, for example.

It means there's something worth investigating, but we don't know enough to know where to start.