A helpful process to follow when designing surveys—and even forms—is to start by specifying domains, which are topics or areas of knowledge that are gathered in the survey.
- [Instructor] Now we are going to start the chapter on survey curation with this video on specifying domains. So, I'm going to get serious for a moment here, because I really want you to understand this. The number one mistake I see people make with designing surveys is not specifying domains. If you skip this step, your survey is often doomed. No matter how many people filled it out, no matter how expensive it was to do. Okay, now that I have sufficiently scared you into specifying domains, I'm going to tell you what domains are. The word domain means an area, like Queen's Domain is around her castle. But in surveys domains mean the same things as dimensions do in data warehousing. They mean general topics or concepts that we want to make sure we measure in our survey. I'm talking very high-level, big picture concepts. On a health survey one domain might be a particular health condition like asthma, or a particular health behavior like exercising. In market research I often see the domain purchase intention, meaning level of intending to purchase the product or service expressed by the respondent. Also, I see brand awareness a lot. When I did market research data collection we'd ask, "What brands of cellular service providers "have you heard of?" And then you you'd say, "A bunch." And we'd say, "Have you ever heard of company X?" That's brand awareness. I emphasize that domains are at a very high level. When I start working with customers often they want to jump right into the weeds of writing questions. I have to hold them back. Let's say we want to ask people about asthma on a health survey. It's not that simple. What exactly do we want to know? Are we concerned about when they were diagnosed, like childhood asthma versus adult asthma? Do we want to measure their risk factors for asthma? So we start fleshing out the asthma domain into subdomains. Again, this is brainstorming. Be free, be creative, don't edit yourself at this stage. And here's the market research version of us fleshing out domains and subdomains. Let's look at purchase intention. So they say they want to purchase it. When, within the next six months? And how sure are they that they want to purchase it? How strong is their intention? Again, be free, be creative. At this stage let all your ideas come out. I emphasize this freedom and creativity because later we will be confined to the relatively small real estate of a survey. So we will want our few questions on our survey to be very packed full of measurements of things we want to measure. This brainstorming really helps us focus on things we want to measure, and grouping them into domains and subdomains helps organize our survey. And make sure we get at least one question in it to measure each domain. In other words, we brainstorm and write down our domains and subdomains at the beginning before we design the questions in the survey to set priorities. Then later we can leave the wordsmithing of perfect questions to the end. This is a great process that can help you produce a great survey in the end, and I'm going to show you how to curate this. Here is the example scenario. My company does public health campaigns. Kind of like marketing campaigns, but I'm trying to get you to be healthy instead of buy something. It's called social promotion. So, I was helping my colleague do one on Twitter. We had a Twitter account tweeting out information about risk factors for type two diabetes, and asking people to consider if they had them. If they thought they had them we told them to go get screened at the clinic for diabetes. So why did we do the project in the first place? Obviously, we wanted people to get screened for diabetes. If you know your disease status you can do things to keep yourself healthy, so it's a good thing. That's why when we decided to do a survey, we decided to do an online anonymous survey of Twitter users following the account to see if they actually got screened for type two diabetes during the social promotion period. And we figured if we were asking that question, we might as well ask a few more questions to evaluate our social promotion. Let's go look at your exercise file for this video, a Word document called Domains for Twitter Survey. Okay, here is the exercise file Domains for Twitter Survey. Easily you can see our main domains, qualifications, screened, demographics, engagement, and satisfaction. In the meeting where we made this we started with no words, and then we realized we needed a list of qualifications. So, I started the qualifications list and wrote some subdomains. While we were working on that we remembered engagement and satisfaction with tweets had been shown in our research to improve likelihood of action, so we added those domains. If you read the subdomains you'll see they are kind of a junk drawer or a grab bag of ideas. But this exercise was really helpful to us. You will see that as we proceed through this chapter, and come to see what the final version of these domains look like.
- How curation files function as part of data management
- Back-end curation
- Front-end curation
- Steps for dashboard design
- Designing surveys
- Creating warehouse, analytic, and application flow diagrams
- Text-based curation files