The Problem With Radio Buttons

I made a mistake.

Google Voice - Inbox

I clicked the X in the lower right corner of the screenshot above. Just as I was about to click it I thought, “That&srquo;s an odd place for a Delete button…”

Only, of course, it wasn’t a Delete button. It’s the button you click when you want to tell the good people at Google Voice that their transcription wasn’t helpful.

I know why I clicked it. I was looking for a way to delete a voice mail and that was the first thing I saw that matched my mental model, so to speak. Back to my mistake. It wasn’t really a user error. Well, maybe it was if I could easily recover from it, but I can’t.

There’s no way to deselect a radio button. That’s just not how they function. I know. You know this already. So do most smart people like you and me. And yet…

Radio buttons are very handy, but I don’t think they are useful for a binary choice. At least, not Yes/No.

Better Yes/No


Yes. Makes sense. I’m a great designer. You are, too. This isn’t an earth-shattering design change. But it does allow for a user to recover from a mistake. You can deselect a checkbox.

So consider this a friendly reminder: Radio buttons are for choosing between mutually exclusive… um, choices. But for the most part, they are better suited when the user has to choose between 3 or more things.


  1. Adam Connor says:

    By not being able to recover, I’ll assume you mean that you didn’t want to check the “check” box, as you didn’t necessarily want to indicate to Google that the transcription was helpful either.

    If that’s the case then isn’t the issue more that you were provided a binary choice when what you really needed for a third option – an “N/A” or “undecided” type option?

    I do personally prefer your solution, but I have to wonder what Google’s intent is here in terms of collecting data. If they were to use the design you laid out, they would not be able to infer that just because the box is unchecked the transcription was unhelpful. So if they are using this question for monitoring quality of the transcriptions, it wouldn’t really be useful to them.

    Of course then one also has to consider how often users would actually answer a question like this. Presenting an “optional” data collection tool like this w/ 3 possible choices on something like this may be overwhelmingly ignored and thus lend itself more to clutter for users rather than useful data collection and user engagement.

  2. Matthew Oliphant says:

    Hey Adam,

    I definitely take your point. A third option would be another way to go, though that adds clutter to the screen. Though that might lead on to use a drop-down. It would be good to collect whether the transcript was unhelpful, but then Google has to figure out what unhelpful means. And now I wonder if it should read, “Did the transcript match the voice mail well enough for you to garner meaning from it.”

    But yes, I didn’t want to check the Yes as I really didn’t have an opinion one way or the other about this message.

    Especially since there’s no transcript.

  3. Matthew Oliphant says:

    Forgot to add…

    I think people who use Google Voice wouldn’t see this as an optional feature to be ignored. Since Goggle sends you a text message of the transcript, and said transcript can be quite off, I think this is something many users would want to give feedback on in order to get Google to improve.

    And frankly, the best way to improve it would be to have the user transcribe the message to how they hear it and let Google’s magical faerie engines compare the diffs.

Leave a Comment