I'm a libra. I love my sign. I also see Scorpio women to stand out of a crowd, like their sign is obvious enough to make me believe astrological sun signs can reveal parts of the personality, but most studies do not show a correlation. One study did, but it was biased as the entire sample came from people requesting an astrological chart of some sort. Anyway, the studies have yet to prove a correlation, but they also have yet to disprove it. I think the main problem was the approach in most studies. When I took a research methods course, my group did this as a study. But we could only survey classmates as we weren't trying to get (is it IRB?) approval...the board that approves the study to be endorsed by your college for you to sample other people. Anyway, the stats didn't prove it, but I think it was either the person who handled the stats portion might have made a mistake in the math, or it just wasn't a large enough sample size. Why? I'll tell you.
I surveyed first before deciding which signs to go with. But either way, about half the people in my class were earth signs. The other half were other signs (fire, air, and water). I also asked them to take a personality test... Lowry’s True Colors personality assessment. The way that worked was personalities were divided into colors. Green was the thinking color similar to the air. Blue was feeling color similar to the water. Orange was the adventure color similar to the fire. Gold was the organizing color similar to earth. Most other studies I read had used Myers Briggs. Anyway, I had people rank their colors in personality with the scoring. So you weren't just all green. You'd be majority green, then blue , etc. So, the people that were earth signs pretty much all had gold as their dominant personality color. The people who weren't earth signs pretty much had other colors, which if I remember correctly seemed to match their horoscope as well. Anyway, the actual results looked like it was going to prove it, but when the t-tests were done, I guess it wasn't enough. I still wonder how that study would perform with a huge sample.