In this module, we're going to talk a little bit about different survey modes. If you remember from our earlier definitions, survey modes are basically the technology through which you deliver your survey. So, we're going to talk about in this module are different modes for delivering surveys and some basic differences in these modes and how to think about how do you design your survey and what technology you're going to deliver it in. In the olden days, we had very few options for how he delivered surveys. The census, for instance, in the United States was often collected in door-to-door. In fact, the gold standard for survey collection for many, many decades in the United States was to send a person to a house and collect data from other people in a face-to-face situation. For a number of decades as well, phone surveys were also a high-quality form of survey method. There's reasons why each of these both became such important forms of surveys but also why people have moved away from them for especially more everyday survey use in the past few years. So, we think about all the different survey modes, this is not an exhaustive list by any stretch of the imagination but you could still do face-to-face surveys. You can do phone surveys, you could do paper surveys through mail, you could do email surveys, web surveys, mobile surveys as well as SMS and text surveys. All of these are options for you to deliver basically your instrument to a set of people. Each of these modes have different implications for design. How you design a survey for a paper form is vastly different than how you design a survey for the web, and we're going to talk about some of those differences but first, let's talk about how do we think about and characterize those differences more broadly. So, let's think about dimensions of the survey modes. What are the two major scale ways to think about how we differentiate between these different modes of delivery. What I want you to think about are two main ones. One is interviewer presence. Whether there's an interviewer there for the survey being delivered or not as well as instrument features. What features does the actual technology enable you to use when you're delivering the questions to the person? So, let's first talk about interviewer presence. Interviewer presence has a lot of effect on your surveys. Whether there's an actual professional person there who's delivering the survey can affect, for instance, how the flow of the survey goes. A professional interviewer is going to keep the survey moving at the clip that they maintain. It's also harder to say no to a person than it is to say no to an email ask or a web form. So, as you can imagine somebody showing up at your door, it's harder to say I'm not going to participate than it is if you get a form in the mail. It's also harder to say no to a person who calls you at your home than it is to close a web tab where there's a survey embedded in it. The final thing to think about interviewer presence, and I'm going to dive into these a little bit more deeply, is interviewers have social desirability and acquiescence effects on how the interview happens. I'm going to talk about each of those separately. So, social desirability is a common human characteristic where we want basically to look acceptable to other people. In survey work that means that when interviewer asks a question, the respondent will shape their response to see more desirable or acceptable to that interviewer. Even if the interviewer is a stranger, people don't want to look non-normative or they don't want to look weird. So, good example from the literature. The question how often do you drive a car after drinking alcoholic beverages, frequently, occasionally, seldom, never. As you can imagine, this is a question that could be if you answer frequently to this is going to have implications for how a person might view you. In studies of this, 63 percent of people said never to an interviewer when a person was present versus 52 percent when there was no person present. You can see there there's a nontrivial response to how people want to be seen when there's another person asking the question. Acquiescence is similar to social desirability and that it shapes how people respond to questions when another person is present. In most cultures, people want to be agreeable. So, an example of how this works out in survey responses. The question individuals are more to blame than social conditions for crime and lawlessness in this country. To that question on an agree to disagree scale, 60 percent of people agree with that statement. In the same survey, social conditions are more to blame than individuals for crime and lawlessness in this country. Fifty seven percent of people agreed to that statement. Now, these are mutually exclusive statements. These are positions that you would think would have opposite scores than what you see, but people just want to agree with a statement that a surveyor says. That means they have to be very careful when a human is present to see how acquiescence of your respondents shapes whether they're actually agreeing with a concept you're trying to study, or whether they're just trying to seem like a reasonable human to a surveyor who's asking them questions. The other large scale dimension of survey modes that we should be thinking about are instrument features. Instrument features are basically the characteristics of the mode that might shape how a respondent experiences the survey. Let's talk about some of these in more depth. One of the largest instant features to consider, and I think the one that shapes most other design decisions is cost. Face-to-face interviews and phone interviews are prohibitively expensive because you have to of course pay people to do the work. Mailing something to somebody's office or home can also be very expensive from postage, much less so than actually paying a person to sit at a phone and keep calling people but still can be prohibitively expensive especially when you start to add total cost of a survey for things like incentives or people on your side of the staff doing analysis. So, cost of these modes are I think a huge factor in how people think about design. Another factor that people often think about is, what I'll call here respondent burden or basically how hard does the person have to work to answer your survey. The classic example is paper surveys can place huge cognitive burden on respondents. They allow for more control of response rates because you know how many surveys you've sent out and how many you get back. But if you've ever done a paper survey and you see any skip patterns like if your answer to question seven is one or four go to page 36, those types of cognitive loads are too high for people and it causes them to drop out of surveys pretty quickly. So, respondent burden can be costs not just to you as a researcher but what's the cost to the respondent. Phone surveys also of course carry respondent burden because often we're interrupting a person during a key time, maybe they're at dinner or maybe they're not interruptible and the cost to the respondent at that point is of course their time and not a convenient moment in their life. Another feature of instruments that we should think about is the flexibility of that instrument. So, some modes, for instance, web surveys or computer-assisted surveys allow for a skip patterns that are fueled by algorithms within the survey software. So, instead of having the burden on the respondent to figure out the skip pattern, the computer figures it out and the respondent just has to answer the questions. The problem with flexibility opt-in means that you lose control of the process, however. So, even though the computer can control the skip patterns, it also means that you don't have a person there to make sure that the respondent is answering all the questions, or that they're not distracted while answering questions, or that they don't have their three-year-old answer the questions. There's a whole set of things that you lose control over when you add flexibility into your design process for respondents. Another important instrument feature to consider is adaptability. More recently, we've had tools like the web surveys and tech surveys that can be triggered by events in a respondent's day, meaning that they adapt to the context of the respondent. So, an example that we saw earlier was a survey that's triggered by a respondent leaving a website or hitting a certain page on a website. All of these contextual adaptive surveys really allow us to collect more targeted, more meaningful data from respondents as they experience new things throughout their day. Mixed mode surveys are basically where you take some of these different modes and you mix them up in order to maximize the strengths of the different modes. So, for instance, a print screener, a screener being a small survey that asks about a few questions about an experience to try to figure out the group that you're really interested in. A print screener can be sent out to people at low cost and that will then incentivize people to participate in a longer online survey. You can also send an email out to people about an upcoming phone survey. These types of mixed mode surveys can often be a balance between the cost of a particular mode of survey and the adaptability or flexibility of another mode of survey. So, in summary, the mode of the survey that you choose is going to really shape how you design your survey instrument and your survey overall. It's also going to shape the data that you get back from your survey. It's really important to consider interview presence both the good and the bad and instrument features when deciding on a survey mode. Interviewers can help keep a survey track, they can help promote people to respond and to respond honestly, but at the same time the social desirability and acquiescence effects mean that you're going to have some shaping of responses just because there's another person in the room. Another final point to consider is that mixed modes can be in a very effective way to balance the benefits of different types of survey modes. So, adding different characteristics of these techniques can help the cost on one side but also the flexibility and adaptability on the other.
