In my Scientific American column this month I wrote about the first study to compare driver distraction while voice texting (using Siri on the iPhone or Vlingo on Android) with texting manually. (You can download the study (pdf).

To my surprise and alarm the study found no difference in safety. Whether you text by speaking or by looking at the screen and typing, your response time is twice as long as it would be if you were not interacting with your gadget. Weirder yet, the study found “eye gazes to the forward roadway also significantly decreased compared to baseline, no matter which texting method was used.”

How could that be possible? Isn’t the whole point of voice texting that you never have tolook away from the road?
I thought, for sure, that the answer involved error corrections. Today’s cell phone voice-to-text software makes plenty of mistranscriptions; surely, I thought, the participants’ time looking away from the road was the time they spent correcting errors.

But then I read this: “Participants were instructed to send the text message generated by the voice-to-text application, whether it correctly detected their speech or not.”

Now I was really confused. I wrote to the author of the study, Christine Yager, associate transportation researcher at Texas A&M Transportation Institute, who was kind enough to reply: “I don’t have a data-driven explanation for why that may have been the result but, anecdotally, many subjects said something to the effect of, ‘If I were using this in real life, I would check to make sure it’s accurate before sending.’ Perhaps the participants instinctively glanced at the phone to check its accuracy (even though we instructed participants to send the message whether it correctly transcribed their voice or not).”

Meanwhile, I read the full study and discovered another few facts that might shed light on the results.

  • The test driving was done was at an abandoned airport, on its runways; the prescribed route was mostly driving long straightaways. There were no stoplights, stop signs or other cars. Speed was capped at 48 kilometers per hour.
  • To determine reaction-time impairment researchers programmed a dashboard light to come on periodically; participants were asked to press a button on the turn-signal lever as soon as they noticed. That’s not quite the same thing as reacting to real-world driving events.
  • As the researchers noted, “Vlingo was incapable of reading a message that the participant received, thereby making it a blended manual-automated condition. Initiating Siri only took one button press, whereas Vlingo required at least three.” In other words, which voice-to-text app people use should also make a difference.

That, for sure, is partly why people took their eyes off the road—when using Android, they had to. But is it fair to pollute Siri’s results by averaging it with the data from Android testing? And both work differently than the new Moto-X phone I had written about, which doesn't require any physical interaction with the handset to activate the voice commands.

  • Here’s the kicker: About half of the participants’ own phones don’t have voice-to-text features—and of the remaining half, 80 percent said that they rarely or never use voice-to-text!

In other words, they were voice-texting newbies when they participated in the study. Of course they took their eyes off the road; they were simultaneously texting, driving and learning how to use the phones!

The authors admitted as much: “A driver more accustomed to using those applications might have executed the task more efficiently. This, again, is an area requiring further research.”


Even so, there’s enough data in this study to satisfy me of its conclusion: Texting behind the wheel, no matter how you do it, is a bad business. It can wait.