Left turns: a follow-up

In “Green lights and left turns”, I talked about the throughput benefits of having left-turn arrows come at the end of the green-light cycle, rather than at the beginning. The following week, Joseph, at Corpus Callosum, posted an entry that advocated taking a hard line on traffic violations, in the interest of safety. Think for a minute, Joseph said, “about what it means, that we are willing to trade the lives of about ~37,000 people, per year, for greater convenience in getting to McDonald’s.”

In a comment to that entry, D. C. Sessions also mentions the left-turn arrow point:

Cheap example: left turn after (rather than before) green is proven to save lives and improve traffic flow. All it costs is a programming change, but it’s only in use in a few cities.

Noting that I talked about traffic flow, but not safety, I asked D. C. for pointers to safety information. He suggested I look into the city of Scottsdale, Arizona, and so I did. I found this on Scottsdale’s Traffic Signal Questions page:

Why does Scottsdale have lagging left turn arrows?, while other Valley cities have leading left turn arrows?

The City of Tucson has had lagging left turn arrows since 1984. Scottsdale tested lagging arrows in 1988, and discovered that this signal operation resulted in less delay and fewer accidents in our community. They were implemented citywide in 1989.

That statement is backed up by this study from 2003, and a follow-up in 2007. In the first, “A study of accidents with lead versus lag left-turn phasing”, co-authored by Scottsdale’s former traffic engineering director, they compare statistics from the cities of Mesa, which uses leading arrows, and Scottsdale, which uses trailing (lagging) arrows. Citing another report, they give some overall national statistics:

• 45 percent of all intersection accidents in the United States involve left-turning vehicles.
• 83 percent of intersections with left-turn phasing used lead phasing.
• 11 percent used lag phasing.

They further note that “Traffic engineers have differing and, sometimes, strong opinions on the merits of each phasing. Each type has valid advantages and disadvantages.” And they say that their study is only considering safety with respect to left-turn head-on (LTHO) collisions. They picked Mesa and Scottsdale because the cities are similar with respect to traffic, and both had good traffic and accident recording.

Unfortunately, the data tables aren’t included in the papers that are posted to the web (they appear to have been created with optical character recognition). But the text tells us that there was no significant difference in the safety of the two types of phasing, considering only the left-turn head-on collisions.

So we turn to the 2007 follow-up by the former Scottsdale engineer.

The 2007 study has the advantage of eight years of collision data, and the ability to do more analysis on the data they had — see the first page of the paper for the details of that.

The results of this study show:

• In general, the LTHO collision rates for the lagging left-turn arrow operation were statistically significantly less than the LTHO collision rates for the leading left-turn arrow operation.
• The percentage of all intersection collisions involving left-turn vehicles and opposing through vehicles was smaller in the city with lagging left-turn arrow operation than in the city with leading left-turn arrow operation.
• The intersection collision rates did not statistically significantly differ for intersection collisions that did not involve a left-turn vehicle.

The author speculates that a reason for the lower accident rate is that drivers making left turns at lagging-arrow intersections are more willing to wait through the green light, knowing that they have an arrow coming. They’re less likely, he suggests, to risk using an insufficient gap in traffic to try to make the turn. Of course, this assumes familiarity with the mechanism — learning, on behalf of the drivers — and so requires some time for the data to settle down and show a benefit.

Further — and I find this particularly interesting — cars that continue turning after their arrow goes away are intruding on opposing traffic very differently in the two situations. At leading-arrow intersections, they’re abruptly turning across directly oncoming traffic, while at lagging-arrow intersections, they’re turning across traffic on the cross-street. The drivers on the cross street can more easily see that there’s a car crossing the path, and can more reliably avoid a collision with it.

I’d actually be interested in seeing a panel discussion where traffic engineers debate the pluses and minuses of the two systems. But it seems to me that the lagging arrow is sometimes better (and never worse) for traffic flow, and appears to have safety benefits. So as I said in my earlier post, it looks like we should switch.