Depends on the campaign. In 5e, for various campaigns, we've utilized
The Matrix,
Point Buy, modified Point Buy (change an 8 to a 6 for +1 to any other stat), and 4d6 drop lowest.
RE: rolling
I've had players (well, a player) get really upset about rolling two low-ish numbers and two average ones. I've had others slightly disappointed they didn't have at least one low stat. I've had players embrace a few low stats as, to them, it makes for a more interesting character. In 7+ years of DMing and playing 5e (250ish sessions?) I've yet to see this stat modifier "power imbalance" in game play that some are lamenting. The only time, at our table, that power imbalance seemed to be a problem was due to a distinct playstyle difference which really had nothing to do with stats and everything to do with making sure the players and DM are all a match for the table.
So, I ponder, at what point does this imbalance, caused by the variability of rolling for stats, take shape? When a character doesn't have any positive modifiers? When they have at least one stat that is 10% worse than any other character? 15% worse? 20% worse? When they have only one stat that is +2 or higher while others have 2 or more such stats? When a character has stats worse than a commoner NPC (i.e. all below "average")? Somewhere else?
In a cooperative game where presumably everyone is out to have a good time, create an exciting story, and share the spotlight in the course of doing so, I'm not sure where the line of imbalance exists, or if one truly exists at all. I'm sure folks are concerned with characters being "effective", but I'm of the mindset that any character, regardless of stats, can contribute in a meaningful way. Others' experiences (and/or opinions) of course may vary - I'm most interested in hearing of experiences where low stats were a tangible problem at the table that took away from fun and story for an entire group.