Assignment 4: Methodology Section
JoseRivera123
script3.docxSlide 1: The Research Process
A research design is an overall plan of study that researchers will use when developing a project and collecting data. Research designs include four steps: selecting a research topic, reviewing the current literature and considering theory, formulating a research question, and then developing a research design based off of the question you are asking and the method you want to use in order to collect data.
The first step in the research process is selecting a research topic. Topic selection may be influenced by various external factors as discussed in previous modules including academic motivations, policy motivations, political motivations, and personal interests. In criminology, often times our research questions relate to crime, deviance, or victimization. When doing research, you should always strive to choose a topic that interests you so that you will not get tired of conducting a research project. Additionally, the topic should be practice, so that you can achieve some sort of success while researching.
Once you have chosen a general topic, the second step in the research process is conducting a literature review which includes both past studies and books that have been written on your chosen topic, as well as theoretical explanations that have been used to explain phenomena within your chosen topic. When reading prior literature, it is always important to consider not just the content, but also the methodology that past researchers have used when developing projects on your chosen topic. This is important for two reasons: one, you may be able to find what methodology is best suited for researching your topic, and two, you may be able to use a different methodology than what was used in the past to make your project unique. Conducing a literature review can take a lot of time and energy, as you are expected to read as much literature as you can, if not all of the past literature. While you may find a gap in the literature that provides you with a unique research question, you may also have to reroute your chosen topic based on past research. Throughout the research process, you must always be flexible when considering your topic, questions, and how you may need to evolve your own project or interests in order to develop something new and meaningful.
Formulating a research question can happen both before or after the literature review. It may also evolve throughout the research process based on what you learn in your literature review. Essentially, a scientific research question is one that is answerable through systematic collection and analysis of verifiable data. Research questions should contribute to conversations and investigations that are occurring in the social science field of your choosing; in this case, criminology. A good research question should be focused and feasible in terms of managing time and resources.
After you have completed these three steps in the research process, you must develop your research design. The research design should detail how you will go about gathering and analyzing data to answer your research question.
Slide 2: Designing Research
Most research questions, especially in quantitative research, are searching to establish or explain a relationship between two or more variables. Before starting to identify these relationships however, researchers must define the variables that they are using. First, a variable is a measured concept that may vary across cases or across time, but is essentially some characteristic or concept of identification you are attempting to measure. Variables can include concepts such as age, gender, and income as well as more abstract characteristics such as self-control, anger, or attachment. After identifying what variables will be important in your research, you must consider the units of analysis.
Units of analysis are the entities such as people, nations and artifacts that are being studied, compared in terms of variables. Social scientists, including criminologists, study a variety of different units of analysis, including individual people and groups such as families or communities. If a researcher was studying how whether affects people’s moods, their unit of analysis would be individuals. Alternatively, if a researcher was considering whether larger organizations have more bureaucratic characteristics than smaller organizations, the unit of analysis would be organizations. Before you start conducting your research, you should be able to describe the unit of analysis for your research question.
After determining your unit of analysis, your should start identifying your variables. As discussed earlier, an independent variable is one that a researcher will manipulate with the belief that it will influence or cause a change in a different variable. The variable expected to change based on the influence of the independent variable is the dependent variable. You must always be able to distinguish between your independent and dependent variables in order to conduct a research project.
There are additional types of variables that may be included in your projects including extraneous variables, antecedent variables, intervening variables, and control variables. Extraneous variables are those that are not part of a hypothesized relationships and they can be described as antecedent or intervening. Antecedent variables are those that occur before the influence of the independent variable on the dependent variable that may influence their relationship. If I am considering how the amount of water influences plant growth, an antecedent variable may be a bug infestation that occurred prior to the watering that affected the relationship between watering and plant growth.
An intervening variable is one that might influence the effect of the independent variable on the dependent variable. Or, in other terms, it is an effect of the independent variable that causes the change in the dependent variable, meaning that the independent variable does not directly cause the dependent variable. Consider the statistic that hotter weather increases homicides. When just considering these two variables, the relationship seems to be direct, however, there may be an intervening variable mediating the relationship, such as anger. Hot temperatures tend to increase anger, and then greater levels of anger may lead to assaults and homicides. Thus, the intervening variable of anger, which is caused by the high temperatures is causing the homicides rather then the temperature itself.
Lastly, control variables are those that are held constant to prevent variation during analysis. Holding variables constant allows researchers to rule out variables that are not of immediate interest but that may also explain part of the relationship being investigated. Some variables commonly controlled for include gender, race, income, and other demographic characteristics as well as variables identified as influencing factors in past research.
As stated earlier, often times research questions are looking to define relationships between two or more variables. A causal relationship is one in which a theorized change in one variable directly produces an change in a second variable. Alternatively, spurious relationships are when one variable seems to affect a secondary variable, when in reality the change is caused by an antecedent or intervening variable. Researchers must consider the relationships produced in their analysis carefully, as statistical significance (or the likelihood that the result of a study, such as a relationship between two variables, could have occurred) may occur in both spurious and casual relationships.
Slide 3: Conceptualization & Operationalization
Conceptualization is the development and clarification of concepts and can includes various types of definitions such as conceptual definitions and operationalization’s. A conceptual definition is a verbal definition of a concept derived from theory which directs the search for measures. Operationalization is the process of identifying empirical indicators and the procedures for applying them to measure a concept.
Conceptualizations can emerge based on characteristics and definitions developed in past research. For example, the definition of social capital has been refined and developed based on past research considering what social capital is and how it is gathered. Conceptualization allows researchers to refine and elaborate on the theoretical foundations of their research and provide a basis for linking theory to data. Conceptualization of variables may differ if a researcher is doing deductive or inductive research. In deductive research, conceptualization includes translating portions of an abstract theory into specific variables to be tested, whereas in inductive research conceptualization is an important part of the process used to make sense of related observations and is often developed as part of the analysis rather than prior to the analysis.
Operationalization is when researchers identify ways of observing variation in order to connect concepts to empirical observations. Essentially, operationalization is defining the method that a researcher will use to measure a concept. Operationalization begins by defining and specifying dimensions of a concept and then determining methods of measuring these concepts. Often more than one method of measurement is used in order to ensure that a concept is being defined and measured correctly. These definitions of measurements are called operational definitions.
Slide 4: Levels of Measurement
Levels of measurement tell us what numbers mean when we compare people or other units into categories. There are four levels of measurement used in research: nominal, ordinal, interval, and ratio.
Nominal measurement is a system that classifies information into two or more categories that are non-numerical. Examples of variables that are usually nominal include: race, religion, or gender. Each of these variables have differing, non-numerical categories that participants in a research project can choose. In order to be nominal, variables must be exhaustive and mutually exclusive. By being exhaustive, the measurement requires that a measure includes all possible values or categories that can be classified. To be mutually exclusive, the measurement requires that each case be placed in one and only one category.
Ordinal measurement is when numbers indicate the rank order o cases on some variable where the numbers assigned indicate only the order of categories. Survey questions that have the categories never, sometimes, always are examples of ordinal ranking.
Interval measurement is when a variable has the same qualities as ordinal level variables such as ranking, but there is also an equal distance or interval between the assigned numerical values. Some examples of interval level variables include temperature, distance measurements such as miles, feet, or inches. However, you cannot calculate mathematical differences in interval level variables because there is no true zero because the zero point does not signify the absence of the power. For example, even if it is zero degrees out, that does not mean there is no temperature.
Ratio level measurements include variables with numerical values and fixed zeros which makes it possible to mathematically interpret the variable. For example, when considering income, you can divide one into the toher to form a ratio that signifies their comparison to one another, such as $20,000 being half of $40,000.
Slide 5: Validity and Reliability
Measurement validity is the goodness of fit between an operational definition of a concept and the actual value of a concept. Otherwise stated, it considers the question of whether or not we are actually measuring what we think we are measuring. Reliability is similar to validity, but measures the consistency of an operational definition, or does our measurement always measure the same way. Of these two concepts, validity is more critical, as reliability can be high even if validity is low if a researcher is successfully measuring a concept incorrectly the same way multiple times.
There are multiple forms of validity and reliability assessment. Reliability assessments include test-retest reliability, internal consistency, and inter-rater reliability. Test-retest reliability is a method of establishing reliability which involves testing the same persons or units on two separate occasions such as administering a survey on two separate days. However, test-retest reliability does have limitations, including needing to test participants or units twice which takes more time, and the possibility that a person might just re-report their original answers. Internal consistency avoids these limitations by measuring the consistency of scores across all items of a composite scale or measure. If you have multiple questions re-affirming the same concept, then answers that correspond with one another on the same scale would thus have high reliability. Inter-rater reliability is the extent to which different observes or coders get the same results when analyzing data separately. The greater the consistency the greater the reliability.
Validity assessments include face validity, content validity, convergent validity, and construct validity. Face validity is an assessment where a researcher uses superficial and subjective assessment of whether or not your study or test measures what it is supposed to measure. However, because it is dependent on researcher interpretation, it can be inherently biased. Content validity is measured using the knowledge of experts who are familiar with the construct being measured. Similarly to face validity, this assessment can still be biased depending on the objectivity of the expert. Convergent validity is based on the extent to which independent measures of the same concept are related to one another. Convergent validation is enhanced by using multiple alternative measures and by using measures based on different operational methods. Lastly, construct validity is an assessment based on an accumulation of research evidence indicating that a measure is related to other variables as theoretically expected which must often be accumulated across studies.
Slide 6: General Sampling Concepts
The four most basic concepts of sampling include the target population, population, sampling frame, and sampling unit. A target population includes the entire group you want to generalize your research to. The population are members of a target population from which your sample is actually selected from. Sampling frames are lists of members of a population from which a sample is selected. Lastly, a sample unit is any single unit sampled from the population.
The goal of sampling is to use a sample of elements of the population to learn about the entire population. There are two types of generalizability, sample generalizability is the ability to generalize from a subset (sample) of a larger population, and cross-population generalizability is the ability to generalize from findings about one group, population, or setting to other groups, populations, or setting. To generalize to populations, a sample must be as representative as possible, or it has characteristics similar to the population. A non-representative sample may contain characteristics which are over or under represented. A measurement of non-representative samples is sampling error, or the difference between the characteristics of a sample population from which it was drawn. Essentially, the representativeness of a sample can be undermined by nonresponse or bias.
Slide 7: Probability Sampling
Probability sampling is a method of sampling that allows researchers to know in advance how likely it is that any element of a population will be selected for a sample, thus making populations statistically representative and generalizable to a whole population. To develop a probability sampling method, researchers first need to define a target population, develop a sampling frame, and lastly, calculate the coverage error, or the error that occurs when the sampling frame does not match the population. The four major types of probability sampling includes sampling random sampling, systematic random sampling, stratified random sampling, and cluster sampling.
Slide 8: Simple and Systematic Random Sampling
Simple random sampling is a probability sampling design in which every case and every possible combination of cases has an equal chance of being included. Simple random sampling requires a complete list of the population, thus making it difficult to apply simple random sampling frames.
Systematic random sampling starts with a researcher determining the number of their sample in reference to a population. After dividing the sample size from the population size, a researcher gains the sampling interval. From the list of persons within a population, a researcher takes every nth element from the list.
Both simple and systematic sampling creates essentially the same sample, however both also require the population to be known and a list of persons in said population.
Slide 9: Stratified Random Sampling and Cluster Sampling
When using stratified random sampling, a researcher must distinguish all elements in a population according to their value on some characteristic. These characteristics form strata, or levels of groups based on some given characteristic. There are two types of stratified random sampling: proportionate and disproportionate stratified sampling. Proportionate stratified sampling allows researchers to create strata based on characters that are the same proportion of the whole population form which to select a random sample. Disproportionate stratified sampling is when a sample is taken from equal proportions of an entire population.
Cluster sampling is used when a researcher doesn’t have a sampling frame with a definite list of elements, or when it is too expensive to cover the sampling frame such as hidden populations or large geographical areas. When using cluster sampling, you may select inmates from clusters of prisons or students from clusters of schools.
Slide 10: Nonprobability Sampling
Nonprobability sampling is a method of sampling whereby each member of a population has unequal probability of selection. This method is used when a probability sample cannot be obtained or when a topic of the study includes rare or hard to access populations. There are four types of nonprobability sampling methods: availability sampling, quota sampling, purposive sampling, and snowball sampling.
Slide 11: The Four Types of Nonprobability Sampling
Availability sampling is used by selecting any and all units that are available for a researcher to access. However, due to the bias of this sample, it can be difficult to implement. Quota sampling is intended to overcome this flaw by creating samples that include quotas of units or participants who have certain characteristics represented in a population.
Purposive sampling is when each sample element is selected for a purpose, potentially due to the participants unique position in the population. Which may involve studying the entire population of some limited group or a sub-set of populations. Lastly, snowball sampling is used by getting referrals of further participants from a known member of a population. Those referrals then give the researcher more names, thus increasing the sample size.
Slide 12: Module Wrap-Up
After reading the texts and listening to the lecture prepared for this module, you should be confident in your ability in completing the learning objectives from the unit.
Your second assignment will be due at the end of this module, and will consist of identifying and describing the variables you plan to use in your final project using the characteristics discussed in this module including operationalization and defining units of analysis. Make sure to check blackboard or the syllabus for further guidelines regarding this assignment and do not hesitate to post questions in to the interactive discussion board for feedback from your classmates or professor.
