unit 5 crj test

Matching up Sampling, Fixing Survey Issues, and Research CJ3675 Unit 5

In Unit 5, we will be exploring survey design, matching types of survey to research, and discussing different survey methodology. Surveys provide great information and can be use as a data-gathering resource. Depending on the survey type, many surveys are one-time data-collection methods, unless it is part of a longitudinal study. Additionally, surveys are more often than not fairly short. Surveys can be used to gather information on a participant's perception of different issues such as the police. Furthermore, self-report surveys ask participants to report victimization, offending, or both, as well as deviant behaviors.

Selecting a Survey Method

Surveys can be very useful tools in terms of data collection. Right now, there are a few different types of surveys. First, there is the traditional in-person survey questionnaire, telephone surveys, online surveys, as well as mail surveys. Each survey type has different pros and cons that will be covered in your reading and resources for this week. When designing a survey, it is important to first ask a few questions, who is my target population? What type of technology will they have access to? What age range is the intended survey population? Is the survey local or do we need a nationally representative sample?

As discussed last week, we should check out what the consortium has available to see if a preexisting database can be utilized for the intended research. If not, then we can move on to designing a survey. At first glance, survey design appears to be cut and dry. Unfortunately, this is not true. A lot of work and effort is put into survey design and researchers who have been designing surveys for decades still run into instrumentation issues. Prior to submitting the required Institutional Review Board (IRB) paperwork for permission to engage in research we need to figure out what methods will be used, and we will need to include a copy of the survey for review by the board.

What type of setting are we looking to collect data at or in? Are we trying to get a local representative and maybe we want to go sit somewhere in the local mall or a popular local establishment. If we are targeting younger generations that are married to their technology we may want to consider an Internet survey. Do we need to have a sample the represents the continental United States of America and have a lot of time available but financial resources are limited? We may want to consider doing a telephone survey. Do we have a lot of free time, a good financial resource and need a nationally representative sample? We may want to consider a mail survey. Depending on the population, the time line, the amount of free time or assistance available from volunteers, and the amount of financial resources dictates the type of survey to be used.

Survey Design

Once we have settled on what type of survey data collection, we need to decide what we would like to measure, what type of questions are needed for those measurement, and the overall design of the survey. Over the course of this class we have discussed two main types of variables. The independent variable, or predictor variable, and the dependent variable, the outcome variable; we run different analyses to see if any relationships exists between the variables and how the independent variable impacts the dependent variable. In previous units we also discussed operationalization. Operationalization takes a concept and changes it into a variable by answering how is it being measured. In terms of survey design we can, for example, think of theoretical concepts and try to find questions to measure them. If we wanted to test an opportunity model on cybervictimization, we would need to start by defining what we mean theoretically.

Theoretically speaking, an opportunity model is the combination of two popular theories; Routine Activities Theory and Lifestyles Theory. Under Routine Activities Theory, we know that lack of capable guardian, presence of a motivated offender, and a suitable target that exist spatio-temporally increase the chance of predatory crime. This model speaks to how we get to lack of guardianship, and highlights that this component is the most important, and that motivated offenders are always present, however, it does not explore target suitability. Lifestyles Theory addresses that issue and states that people who engage in certain types of behaviors have an increased chance of victimization. Collectively, the two theories (Routine Activities Theory and Lifestyles Theory) are known as L-RAT or Opportunity Theory. At its conception, the theory only applied to traditional in-real-life environments but the theory has since been tested and found applicable to the digital environment as well.

In designing this survey, we would need to operationalize the different components of the theory. A popular approach is to review previous literature and find out what questions have been used before to measure these concepts. For instance, in reviewing the literature, we will find that opportunity factors are broken down into frequency, risk, and self-guardianship. Since we are talking about a digital environment frequency would refer to or be operationalized by questions that asks how much time a participant spends on the Internet, the number of social media accounts they have, how much time they spend browsing and posting on social media, and perhaps the amount of time they spend on their cellular phone. Risk could be operationalized in a digital environment as the type of websites participants' use, how willing they are to engage in a conversation with someone they have never met in real life, and how much personal information they are willing to give out over the Internet. Self-guardianship can be defined as privacy settings or willingness to block individuals who harass or threaten them, and could be operationalized by asking the participants what type of privacy settings they use, if their profile is public or private, and if they block individuals who threaten or harass them.

Once we have established our concepts and have operationalized them, we can start to determine which questions can be dropped because another thing we want to be mindful of with survey research and using survey questionnaires is how long it will take a participant to complete. If we are surveying college students we want to be mindful that we are asking professors to donate class time and surveys should be limited to ten to fifteen minutes. We also need to be mindful of the participants, if a potential participant hears it will take thirty minutes or opens up a thirty page survey, they may be less inclined to participant, which is well within their ethical rights. Remember, all participation is voluntary, responses need to be anonymous and if they cannot be kept anonymous they must be confidential, and a participant can skip any question or stop participating at any time. Once the design is set, it is time to start reviewing and drilling down. If you remember back to the previous discussion, one of the threats to internal validity is instrumentation. We cannot change the questionnaire once it has been given out because it will impact our research and study. A way to mitigate this threat is to use focus groups of similar participants from a different pool that will not be surveyed and to handle any survey issues.

Survey Issues

In constructing a survey it is important that we are mindful of the accompanying issues. The most popular issues are using bias/leading questions, double-barreled questions, providing preset answers that are not mutually exclusive (overlapping answers) or exhaustive (do not include an answer for everyone). There are ways to identify these issues and fix them. The first way is using a focus group. A focus group is a great way to pre-test survey questionnaires on a group that is similar but not from the same sampling pool. We do not want to use a focus group from our sampling pool because it would cause other threats to internal validity such as test/re-test. We want to use a population that is similar. For instance, if we wanted to survey Kean University students about their National Football League fandom, we could gather focus group participants from Jersey City State University because both groups are college students but we are not biasing our selection pool.

Let's discuss survey issues using the NFL fandom example. It is fascinating to meet different people and figure out where their fandom lies and how it came to be. For instance, sometimes you will find a person who was born and raised in New Jersey and their entire family is from New Jersey and are a fan of the Miami Dolphins football team. Did this happen because they were born during the time leading up into the Dan Marino Era? Do they really like the colors orange and green? Do they have a peer group who cheers for the Dolphins? In order to find out we want to engage in survey research to find out. One of the questions we will need to ask is what NFL team they cheer for or consider "their" team. One of the issues we have discussed is the categories in regards to the answers are not exhaustive, meaning not all of the options are listed. For this question, at first look we may believe we only need thirty-two answers because currently, there are thirty-two teams in the NFL. However, we would cause an issue to occur. Did we provide an option for someone who is not a NFL football fan? How about the anomalies that exist who root for multiple teams (which we could then debate if they are a true fan)? We did not. We would need to include at minimum thirty-four responses.

Now let us arguably say for purposes of this example that we included multiple teams in the answer sets. We will then have caused a different issue. We would have created an issue because the categories will not be mutually exclusive meaning that participants would be able to select more than one answer (unless the question is built for this type of answer, this practice should be avoided). This creates an issue because then the participant would fall into two categories and their answers would need to be excluded or the participant may not answer the question because of confusion. It is best to allow questions to only have one answer in each answer selection.

Additionally, focus groups help to identify and avoid bias/leading questions, questions that ask participants about things that would not be in their average knowledge base, as well as double-barreled questions. Double-barreled question and answers provide multiple question or multiple answers that may be unrelated. For instance, using our NFL example, if we asked are you a football fan and did you go to the game on Sunday? We are asking participants two questions in one question they may be a fan but did not go to the game. In order to fix this issue we would break this down into two separate questions. Furthermore, if the question asks if they purchased food from a vendor while attending the game, and we wrote answer a) yes, beer and hotdogs, it is plausible that someone purchased a hot dog independent of a beer and vice versa. We would make sure that the answers were separated.

(CSLO 5, CSLO 6, CSLO 7))

References

Cohen, L. & Felson, M. (1979). Social change and crime rate trends: A routine activity approach. American Sociological Review. 44(4). pp. 588-608.

Hayton, A. C. (2017). Understanding factors that impact cyberbullying offending and victimization. (Unpublished doctoral dissertation). University of New Haven, West Haven, C.T.

Hindelang, M., Gottfredson, M., & Garofalo, J. (1978). Victims of personal crime: An empirical foundation for a theory of personal victimization. Ballinger Publishing Company: Cambridge, Mass.