Facebook Cookbook - Jay Goldman [13]
Figure 2-1. Scrabulous in SocialMedia’s Appsaholic web app and Facebook app
Figure 2-2. Scrabulous in Deft Labs’ AppHound Facebook app
Test-Driving Ideas with Facebook Polls
Problem
I think I have an amazing idea for a Facebook app or a feature, but I want to test it out before I go and build the whole thing.
Solution
Facebook Polls offers a cost-effective way to get some real-time feedback on an idea before you invest a lot of time and effort. You can find out more about Polls at http://www.facebook.com/business/?polls, and you can jump straight into creating a new Poll at https://secure.facebook.com/add_poll.php (Figure 2-3).
Figure 2-3. Create a New Poll, step 1
Polls consist of a short question and up to five multiple-choice options. You can filter your audience by one of Interests, Location, Age, or Sex in order to shrink your audience and get an answer faster, or you can just leave it open to anyone.
The second step (Figure 2-4) is all about the Benjamins: paying for your Poll.
Figure 2-4. Create a New Poll, step 2
The top of this screen gives a preview, the middle lets you configure how much you want to spend, and the bottom (omitted in the screenshot) gathers your credit card information. You’ll note in the middle section that you pay a $1 insertion fee and can then select either $0.25 or $0.50 per response (with estimated runtimes of 24 or 48 hours, respectively), as well as a maximum number of responses. You may never get to that number if people don’t feel like participating, so the maximum cost indicated here really is just an estimate until you see how the Poll plays out. You’ll see the $1 fee hit your card immediately and the rest of the fee at the conclusion of the Poll.
You’ll be given a URL for the dashboard of your Poll, where you can watch the results roll in as they happen. Facebook provides a demo poll (http://tinyurl.com/4r3m84), which shows you a “What’s your favorite soft drink?” example that was answered by 148 users.
Discussion
In researching this book, I spoke to many accomplished developers who told me that using Polls was almost like a secret weapon in their success (which I humbly reveal to you with their permission): they had all started down a certain road and ended up on a completely different but very popular route thanks to the results of a Poll.
If you haven’t done many statistics courses before (or if they were more slanted toward the math side of stats and less to their application in the social sciences), you might not be familiar with the ideas of reliability and validity. Both are critical to building any kind of survey if you’re going to rely on the answers to make important decisions:
Reliability
This is the constancy of your measurement. A survey is considered reliable if it returns the same result each time it runs (i.e., a person’s score is reasonably the same each time she takes the test). This can be thought of as the “repeatability” of your survey. You can verify reliability by running the same Poll at different times with the same audience to see if you get the same result.
Validity
Did your Poll actually measure what you set out to quantify? It’s important that you’re measuring an actual causal relationship, rather than a coincidence in the data. It’s harder to construct an invalid Poll when you’re asking only one question, but the goal is to make sure that the question you’re asking is actually a direct outcome of the answers you’ve offered. People often behave very differently from