Welcome to Module 4 in our course Framework for Data Collection and Analysis. In this 4th module, we will describe to you six example surveys to which we provide additional material on the course website. And we are highlighting in our description of these example surveys a few features of the total survey, total survey quality framework that you saw in the previous module. We encourage you to dive into these surveys, and to look for similar surveys in your own countries. And see how they differ in these key elements. With that, you will broaden your horizon on the different opportunities to collect such data and the difference methods and different qualities across countries. Here's an overview, over the eight example surveys that we'll provide, talk to you about. I will walk you through each single one of them in one of the video snippets. Obviously there's only a very brief set of material that we can provide here in the video lectures. But they are all interesting in as much as they provide key indicators that serve policy needs, researcher needs, and often also needs of private sector companies that want to know about the state of a particular country. Or even larger entities like the European Union that is interested in data across different European countries. So all of those types of surveys collecting policy and social science indicators. The first one in this series of surveys that we'll discuss is the National Crime Victimization Survey. The purpose for this survey is a concern that the police records underestimate crime prevalence. Here is an interesting case, right? We have crime statistics from the police records and those would be a good example for administrative data that can be used in analysis. And in fact, in the US, there's a huge endeavor going on right now. They're trying to, you know, link these different administrative databases together so that a good picture is available for the crime and justice system. However, if a crime is not reported to the police, you have under coverage of crimes, and there is no record of them. So, you have maybe, without any missing values and without any data entry errors, administrative records on each of the crimes reported. Nothing is there on the underreported, or actually, not reported crime for that matter. That's one purpose. Another purpose is that countries see value in perceived victimization so even, so not everything is a crime. There is a perception of being victimized and a perception of being fearful of potential victimization. That is an interesting social indicator and has led to many, many countries collecting a survey that is somewhat similar to the crime and victimization survey that we see here. We have in this one in the United States. Measurement of both households, and person victimization through self-report. So there's usually an informant for a household that reports about the entire household, and then the individuals in the household. Some of the surveys also have this concept of being the household sample, and then either every member in the household is surveyed or a selection happens at that household screening stage. With the Crime and Victimization Survey, it is possible to report and analyze rates of victimization across a variety of subgroups that are determined through the survey variables. Right now, the NCVS is looking at another redesign. There has been a lot of interesting methodological studies going on in that survey. So if you are interested looking through those old methods reports, we'll come back to some of them in the question and design course. Then there are the NCVS Key Design Features for you to look at on the perspective of a methodologist. We have target population, in this case U.S. household members, 12 years and older. It is a multi-staged area probability sample, with about 16,000 persons per year in 90,000 households. It's called a rotating panel survey. That means they're seven interviews done six months apart. And then a new person / household is used. Okay? So there's a string of like a mini-panel that these respondents are in a couple of years. And then other people will be interviewed. The surveys interview administered, the first wave is in person and wave 2-7 are preferably done over the phone with the computer assisted telephone interviewing. What's interesting here, the first survey, these types of surveys are often done as a bounding interview, so that you record everything that happened up to this point and then later on you can always reference expectancy, was it reported up to that point. Again, something that we will discuss in the question and design course. Speaking of question and design, here is the questionnaire for the National Crime and Victimization Survey, and a link on the bottom that gives you the full set. This is kind of an old little screenshot here from it. Also, you see some results posted here. This is a difference in the rate of victimization, per thousand persons. Age 12 or older, for a series of years here, 1994 through 2011. These other, not design features, but design issues. As I said, purpose of this is to overcome issues of underreporting, relative to the police records. But of course a survey has errors too, as we discussed. So here we have usually an underreporting of minor and repetitive victimizations. This has to do with memory issues. Memory and recall will be a whole segment in the course on questionnaire design. Also an issue with reference period, another piece relevant for good questionnaire design. But then also we have nonresponse at the person-level. There's certain segments of the population that are typically hard to find or hard to get to an interview. Older people are more likely to participate in the survey. Younger people less likely. In part because they are less at home. Here in the Crime and Victimization Rate Survey it matters a lot that people most likely to be victimized and people most likely to be victims, actually, might be the hardest to contact. And therefore maybe not part of the survey itself, or refused to participate.