Comments on: Beta Has Launched! http://misterpoll.wordpress.com/2007/10/01/beta-has-launched/ Web site news, hot topics, and other fascinating tidbits. Mon, 07 Jul 2008 02:37:19 +0000 http://wordpress.org/?v=MU By: Ed http://misterpoll.wordpress.com/2007/10/01/beta-has-launched/#comment-3483 Ed Fri, 26 Oct 2007 20:03:38 +0000 http://misterpoll.wordpress.com/2007/10/01/beta-has-launched/#comment-3483 OUTSTANDING!!! ;-) OUTSTANDING!!! ;-)

]]> By: misterpoll http://misterpoll.wordpress.com/2007/10/01/beta-has-launched/#comment-3370 misterpoll Tue, 23 Oct 2007 19:36:36 +0000 http://misterpoll.wordpress.com/2007/10/01/beta-has-launched/#comment-3370 Hi Ed, Let me start with an example, because the terminology of "votes" vs "voters" can be too confusing without it. Let's say I create a simple multi-select question: What ice cream flavor do you like? You may pick more than one. [ ] Chocolate [ ] Vanilla Now, let's say that you and I both vote, and we each select both flavors. The old system considers that 2 voters and 4 votes. It always showed the results as a percentage of votes, so that the total is always 100%. It would show: Chocolate: 50% (2 out of 4 votes) Vanilla: 50% (2 out of 4 votes) Over the years, we have actually had a lot of complaints about this method. It seems that most people wanted to see the results expressed as a percentage of voters. I even had an extended email dialog with a statistics professor once, where he told me it was statistically invalid to show the results as a percentage of votes. I still don't know that I agree with that, but in the end, we did decide to change how multi-select results are shown. In the new system, the same results would be shown this way: Chocolate: 100% (2 out of 2 voters) Vanilla: 100% (2 out of 2 voters) We thought about letting people flip between the two methods but finally decided that would add more confusion than benefit. Of course now we'll probably have people complaining that they want it the other way. Luckily, those people will be able to use the new download feature and slice up the data any way they please. Hi Ed,

Let me start with an example, because the terminology of “votes” vs “voters” can be too confusing without it. Let’s say I create a simple multi-select question:

What ice cream flavor do you like? You may pick more than one.

[ ] Chocolate
[ ] Vanilla

Now, let’s say that you and I both vote, and we each select both flavors. The old system considers that 2 voters and 4 votes. It always showed the results as a percentage of votes, so that the total is always 100%. It would show:

Chocolate: 50% (2 out of 4 votes)
Vanilla: 50% (2 out of 4 votes)

Over the years, we have actually had a lot of complaints about this method. It seems that most people wanted to see the results expressed as a percentage of voters. I even had an extended email dialog with a statistics professor once, where he told me it was statistically invalid to show the results as a percentage of votes. I still don’t know that I agree with that, but in the end, we did decide to change how multi-select results are shown. In the new system, the same results would be shown this way:

Chocolate: 100% (2 out of 2 voters)
Vanilla: 100% (2 out of 2 voters)

We thought about letting people flip between the two methods but finally decided that would add more confusion than benefit. Of course now we’ll probably have people complaining that they want it the other way. Luckily, those people will be able to use the new download feature and slice up the data any way they please.

]]> By: ed http://misterpoll.wordpress.com/2007/10/01/beta-has-launched/#comment-3301 ed Sun, 21 Oct 2007 19:43:28 +0000 http://misterpoll.wordpress.com/2007/10/01/beta-has-launched/#comment-3301 Great improvement! Question/suggestion: is it possible to show results of multiple-selction questions differently with results shown as percentage of number of voters as well as percentage of number of votes? Example: if there are 5 voters on a question and each votes for 3 responses, there are 15 votes. If all 5 have selection that is the same that is shown as 33% of votes, which it is, but it would be nice to also see that 100% of voters selected that. Keep up the good work. Great improvement! Question/suggestion: is it possible to show results of multiple-selction questions differently with results shown as percentage of number of voters as well as percentage of number of votes? Example: if there are 5 voters on a question and each votes for 3 responses, there are 15 votes. If all 5 have selection that is the same that is shown as 33% of votes, which it is, but it would be nice to also see that 100% of voters selected that.

Keep up the good work.

]]> By: misterpoll http://misterpoll.wordpress.com/2007/10/01/beta-has-launched/#comment-2925 misterpoll Fri, 05 Oct 2007 05:50:02 +0000 http://misterpoll.wordpress.com/2007/10/01/beta-has-launched/#comment-2925 Tom: Initially, the individual voting data will only be available to the poll author. We toyed with the idea of opening this up to voters, but we decided that many poll authors would be protective of this data. If there is enough interest, we may make it an option for poll authors to publish more detailed data to the public in some way. For now, voters will be limited to the aggregate voting numbers they're used to seeing. Tom: Initially, the individual voting data will only be available to the poll author. We toyed with the idea of opening this up to voters, but we decided that many poll authors would be protective of this data. If there is enough interest, we may make it an option for poll authors to publish more detailed data to the public in some way. For now, voters will be limited to the aggregate voting numbers they’re used to seeing.

]]> By: Tom http://misterpoll.wordpress.com/2007/10/01/beta-has-launched/#comment-2924 Tom Fri, 05 Oct 2007 05:28:32 +0000 http://misterpoll.wordpress.com/2007/10/01/beta-has-launched/#comment-2924 Will it be possible for voters to also see individual voting data, or only the original author of the poll? Will it be possible for voters to also see individual voting data, or only the original author of the poll?

]]> By: misterpoll http://misterpoll.wordpress.com/2007/10/01/beta-has-launched/#comment-2923 misterpoll Thu, 04 Oct 2007 23:27:20 +0000 http://misterpoll.wordpress.com/2007/10/01/beta-has-launched/#comment-2923 Phil, interesting suggestion about swapping the search / new poll boxes on the front page. I did worry about people mistaking the new poll box for a search when that page was designed because they look so similar. It didn't occur to me that we might could minimize that by swapping their positions. I'll experiment with it a bit. Thanks! Phil, interesting suggestion about swapping the search / new poll boxes on the front page. I did worry about people mistaking the new poll box for a search when that page was designed because they look so similar. It didn’t occur to me that we might could minimize that by swapping their positions. I’ll experiment with it a bit. Thanks!

]]> By: misterpoll http://misterpoll.wordpress.com/2007/10/01/beta-has-launched/#comment-2922 misterpoll Thu, 04 Oct 2007 23:23:49 +0000 http://misterpoll.wordpress.com/2007/10/01/beta-has-launched/#comment-2922 Thanks, Phil! You won't lose your previous results. They will be saved, but in a separate historical set of results. If you login to your account on the beta site and click "View your polls", you should see a complete list of your polls. Each poll will have a set of "launches" listed for it. (A launch is an independent set of results, so you could launch a poll multiple times, say once for family and once for your office, and keep the results separate.) Converted polls from the old site will have a launch called "A Fresh Start", which is where votes from the new site will go. If the poll had any results from the old site, it will also have a launch called "Converted Results". This is where your old results are. Why are they separate? The old site and the new site track results in vastly different and incompatible ways. The old site only tracked the total number of votes each choice received. The new site actually saves each individual vote, and you'll be able to get at that data when we finalize the download format. We simply couldn't convert the old data into the new format without inventing all of those details. So, we isolated the old results into a special launch type to preserve the data for you, then started a fresh new launch to continue taking votes in the new system. It does mean that when the new site launches, from the voter's perspective, all polls will start over at zero. It's a bit awkward, but necessary to make sure we don't taint the new system right out of the gate. Fortunately, it's only temporary since things will start to look better as people begin to vote again. I hope that made sense. I'll try to work this into our FAQ in a more compact form. :-) Thanks, Phil!

You won’t lose your previous results. They will be saved, but in a separate historical set of results. If you login to your account on the beta site and click “View your polls”, you should see a complete list of your polls. Each poll will have a set of “launches” listed for it. (A launch is an independent set of results, so you could launch a poll multiple times, say once for family and once for your office, and keep the results separate.) Converted polls from the old site will have a launch called “A Fresh Start”, which is where votes from the new site will go. If the poll had any results from the old site, it will also have a launch called “Converted Results”. This is where your old results are.

Why are they separate? The old site and the new site track results in vastly different and incompatible ways. The old site only tracked the total number of votes each choice received. The new site actually saves each individual vote, and you’ll be able to get at that data when we finalize the download format. We simply couldn’t convert the old data into the new format without inventing all of those details. So, we isolated the old results into a special launch type to preserve the data for you, then started a fresh new launch to continue taking votes in the new system.

It does mean that when the new site launches, from the voter’s perspective, all polls will start over at zero. It’s a bit awkward, but necessary to make sure we don’t taint the new system right out of the gate. Fortunately, it’s only temporary since things will start to look better as people begin to vote again.

I hope that made sense. I’ll try to work this into our FAQ in a more compact form. :-)

]]> By: Phil http://misterpoll.wordpress.com/2007/10/01/beta-has-launched/#comment-2921 Phil Thu, 04 Oct 2007 18:33:17 +0000 http://misterpoll.wordpress.com/2007/10/01/beta-has-launched/#comment-2921 Oh, one suggestion: Switch the "Make a Poll" and "Search for a Poll" boxes. (I'm guessing) that people search for polls more than create them? I know that I've filled out the Make a Poll box a few times when I meant to be searching. My eyes go to the center of the screen automatically, so I just assume that the box I see is for searching. Just an idea... Oh, one suggestion: Switch the “Make a Poll” and “Search for a Poll” boxes. (I’m guessing) that people search for polls more than create them? I know that I’ve filled out the Make a Poll box a few times when I meant to be searching. My eyes go to the center of the screen automatically, so I just assume that the box I see is for searching.

Just an idea…

]]> By: Phil http://misterpoll.wordpress.com/2007/10/01/beta-has-launched/#comment-2920 Phil Thu, 04 Oct 2007 18:31:04 +0000 http://misterpoll.wordpress.com/2007/10/01/beta-has-launched/#comment-2920 It looks sweet, great work! One thing I noticed...the polls on the beta site don't have the data from the old site...is this just an oversight, or will we lose all our previous results when the new site is released? Keep up the great work! It looks sweet, great work! One thing I noticed…the polls on the beta site don’t have the data from the old site…is this just an oversight, or will we lose all our previous results when the new site is released?

Keep up the great work!

]]>