Below are three hypotheses, please use them to answer the two questions. The answer should relate to the chapter I attached. The chapter is only for you to understand what are data source, data setting, and data sample.  Hypothesis 1. Brushing teeth twice a day will prevent cavities.  Hypothesis 2. If you get at least 6 hours of sleep, you will do better on tests than if you get less sleep. Hypothesis 3. Witnessing more offenders involved in cyberbullying should lead to higher perceptions of hurt towards the victim of cyberbullying. 1. Propose a study to test the hypothesis including an explanation of the data source, data setting, and data sample 2. Provide your reasoning for your proposed study based on the concepts learned in class and in Chapter 4. For example, are there any threats to your study? 6 attachments Slide 1 of 6 attachment_1 attachment_1 attachment_2 attachment_2 attachment_3 attachment_3 attachment_4 attachment_4 attachment_5 attachment_5 attachment_6 attachment_6 Evidence for Causal and Associative Research Arguments 59 research. You must then test these claims by collecting and analyzing data as evidence for the claims, the pro- cess we will turn to next. identified as a negative correlation. So, for example, if McNallie and Hall had found that as perceived use of relationship maintenance behaviors increased, relation- ship satisfaction had declined, then they would have found a negative correlation. You will learn more about associative relationships in Chapter 9, “Analyzing and Interpreting Quantitative Data.” At this point, it is im- portant to stress that associative claims only satisfy one condition, association or covariation, for causality. Tests that McNallie and Hall used do not test for time order nor did the design of their study systematically rule out other potentially competing explanations. Evidence for Causal and Associative Research Arguments Quantitative social science researchers frequently use the term data to mean evidence because of the emphasis on quantifying communication variables under study. In Chapter 5. “Measuring and Designing Quantitative Social Science Research,” we will describe in detail how variables are converted to quantities. When we explore the methodologies of experimental and survey research and content analysis, you will have the opportunity to review many examples of the ways in which data are col- lected and analyzed as tests of claims. To understand how data are used as evidence, we must first understand the sources of data, the settings in which data are collected, and the strategies researchers use to collect the data. Independent and Dependent Variables Earlier we posed several hypothetical questions about the various causes, such as age or sex, that might affect your media choices. In the development of a causal model for media consumption, Hmielowski, Holbert, and Lee (2011) were interested in explaining what makes people consume political satire. They used past research to identify causes or predictors based on audience charac- teristics when viewing political satire. The current study expanded this list to include a new predictor, affinity for political humor. The authors conducted a telephone survey on a random sample generated professionally to test the effects of their set of predictors on viewing Jon Stewart’s The Daily Show and The Colbert Report When you are building a causal claim, such as the one in this study, the causes or predictors that will change or influence other communication responses are called independent variables; in this study, the independent variables were classified into five groups: demographics (age, sex, etc.), political orientations, media exposure, need for humor, and affinity for political humor as the new variable (Hmielowski et al., 2011, pp. 103-104). The researchers assumed that these independent variables would predict viewership of these two television shows. The dependent variable is the communication phenom- enon presumed to be affected by the predictors; this case, the authors measured political television satire ex- posure by time spent viewing The Daily Show and The Colbert Report. When the claim you construct predicts that a change in the dependent variable is preceded and influenced by a change in the independent variable, you are describing the basis for a causal claim. In the cycle of inquiry, we have explained that con- structing causal claims is only the first step in building a deductive argument for quantitative social science Data Sources By data sources, we mean the points from which data originate; the data are evidence collected to test a specific research claim. Recall that in quantitative social science research, claims are tested through empirical observation; researchers prefer to observe communication empiri- cally by examining the messages themselves or by asking participants to tell them about their communication be havior. There are four possible data sources: (a) existing texts; (b) self-reports of communicative behaviors, beliefs, and/or characteristics; (c) other reports of communicative behaviors, beliefs, and/or characteristics; and (d) direct observations of communicative behaviors. Let’s briefly differentiate each data source before we move on to con- sider the settings in which these sources can be collected and the means used to capture each source so that it can be used systematically in a research project. Texts You probably think of texts as written or spoken words, as we do! However, when you think of textual data for communication research, you must also include texts that are symbolic, performed, and purely visual or pictorial (Gow, 1996; A. Phillips, 2012). In the age of social media, 60 Chapter 4 Making Arguments for Association and Causality texts can include written posts, visual memes, signifiers or characteristics, are the third source of communi- such as hashtags, symbols for liking or following posts by cation data. In quantitative social science research, users, etc. For example, in a study of nonprofit advocacy other-report data are used when a researcher wants to organizations, Auger (2013) used posts from Facebook, compare self-perceptions of a communicative act with Twitter, and YouTube to analyze message characteris- how respondents perceive others’ perceptions of the tics and functions. And earlier, we described Hefner and same act, situation, or event. Wilson’s (2013) study of romantic comedy films. Virtually As an example of other report data in interper- any message in any form can serve as a text, so it is easy to sonal communication research, romantic partners are see that the list of potential texts is endless. In Chapter 8, often asked to indicate how they think their partners “Content Analysis,” you will explore how texts like these will feel and respond. Knobloch and her colleagues message examples are the primary data source for content (Knobloch, Sharabi, Delaney, & Suranne, 2016) inves- analyses as a quantitative social science methodology. tigated the role of self, partner, and relationship uncer- tainty in topic avoidance. Participants in this online Self-Reports survey used self-report in responding with their own A second possible source of data for communication beliefs about and feelings of uncertainty; they used research is to ask people to self-report, or to disclose other-report when they estimated how uncertain their their own behaviors, beliefs, or characteristics related partners felt about the relationship. Examples of other- to . If you want to know what people relationship. (2) how important your relationship is report items included “(1) your partner’s view of your communication. feel and how they think, self-report data should be your sto data source. In quantitative social science research, your partner, (3) how your partner feels about your re- you will collect self-report data in surveys and various lationship, and (4) your partner’s goals for the future of your relationship” (Knobloch et al., 2016, p. 33) scales used in experimental research. For example, in a Other-report data are susceptible to the same errors survey study of time management, Panek (2014) asked students to self-report the amount of leisure media use of self-report data. Ways to minimize these errors are discussed more fully when we discuss self- and other they engaged in and the guilt they experienced while report data in survey research. using it. In another study , Kelley, Su, and Britigan (2016) conducted a countywide survey in Nebraska to explore Observations clients’access to health information and discovered that In traditional quantitative social science research, such at-risk groups had less access to health information. as experimental and survey research and content analy- Self-reporting is the most frequent kind of data source in survey research. As you will learn, survey research is sis, direct observations of people in interaction as the a very common methodology in quantitative social sci- primary source of data are not common. As you will learn ence research. Part 3, these observations are commonplace in Because of the value of precision, self-report data methodologies such as discourse analysis and are the are subject to some standard biases or sources of error. primary data source in the bridge methodology, con For example, humans tend to overestimate their own versational analysis. In traditional quantitative social science research, observations may occur in observing positive qualities and behaviors and underestimate the amount of time people were engaged in a certain their negative qualities and behaviors. Depending on activity or counting frequencies of behaviors. Nonver- the research topic, their relationship with the researcher, bal behaviors are often captured in type and duration and many other factors, people may not report all their measures. Sometimes, observations include physiologi thoughts and feelings. Reported memories can be in- cal complete, inaccurate, and so on. You will learn much responses. For example, Pauley, Floyd, and Hesse more about how to deal with bias in self-report data col- (2015) found that affection interaction provided an lection and analysis in survey research. emotional buffer against increases in negative physi- ological responses, such as blood pressure and cortisol levels and heart rate, to stress. In another study, Floyd Other Reports and Denes (2015) collected saliva samples to study the Other-reports, collected by asking people to report relationship between individuals’ genotypes and at- their perceptions of another person’s behavior, beliefs, tachment patterns. Evidence for Causal and Associative Research Arguments 61 of artifacts controlling what of new ics result in large Health Data Settings: When Settings Count Data collection settings are the locations where com- municative texts, self-reports, and other reports are found. There are generally three settings in quanti- tative social science: laboratory, field, and archival sources. In Chapter 6, “Experimental Research,” you will see that a major form of control in construct- ing causal arguments is control over the setting. In laboratory settings, researchers select and control the environment to test claims about communication variables; they induce the variables of interest to occur, what behaviors occur, when, for how long, and with whom Communication research laboratories are often rooms in academic buildings that have been set up to resemble the settings where communication usually happens (e.g., classrooms, living rooms, workplaces, or medical waiting areas). Sometimes they contain one- way mirrors so that behaviors can be observed without participants’ knowledge. Labs may be equipped with audio or video-recording equipment that is less obtru- sive than simply placing a tape recorder or camera on a e in the of a room. Laboratory settings are somewhat artificial, no matter how cleverly you set up the space data collection equipment from the research partici- pants. For this reason, the rigor provided by a laboratory setting often comes at a cost: Research findings obtained in the laboratory may be less generalizable to people op- in nonresearch settings than findings obtained in fields Ad settings. Because of the lack of control and difficulties with access, far fewer quantitative social science research studies place in field settings, where communica- tion occurs in its usual and customary fashion. Here the tradeoff relinquishing experimental control in favor of fstudying naturally occurring communication behavior. New technologies are making it more possible for communication researchers to capture communica- tion as it medical settings (e.g., doctor’s office, hospital Clinic). Even in a field setting, the value of your analysis will be reduced if the participants changed their behaviors or edited their messages because you were present relative to the ways they interacted when no data collection was happening. We will return to the question of setting ef- fects in future chapters in Part 2 when considering ways to account for and reduce potential sources of error. The setting as a physical location is irrelevant in two instances: (a) when researchers use an archive of textual data to explore a research claim and (b) when research- ers employ a modality, which permits them to survey participants at a distance, such as using telephone or online surveys. An archive is a preexisting collection or other textual evidence. Examples of ar- chives include sources of media texts of all kinds, local and federal government records, legal and policy docu- ments, and so on; that is, virtually any public domain in which texts or messages are collected and stored is an archival source. These sources have radically changed in the past decade. Advances in technology, the advent anay companies that collect, analyze, and make available data from the archives of governmental agencies, such as the U.S. Bureau of Statistics, the National Center for Statistics, and the Centers for Disease Control and Prevention. Once such company, Cambridge Ana- lytica, caused considerable trouble for Facebook when their employees mined the archives of millions of Face- book users for political campaigning purposes. In this example, the setting occurred in millions of Facebook user profiles rather than in a distinctive type of physical location. Similar advances have been happening in survey technologies as well , making it possible for for phone re- searchers to tap landlines and mobile phones with com- plex algorithms. Online modalities mak it possible to send out surveys by email or web or virtually from any- where a survey link can be embedded. Survey respon- the survey on any laptop, smartphone, desktop where the link can be accessed. In survey research, the focus is shifting from where (setting) data are collected to how (modality) data are We will have more to say about data analytics and survey modalities later. A final consideration in our discussion of data from the quantitative social science paradigm per spective focuses on data collection strategies or methods. mask your dents can can take tablet, or Litudies take collected. comes with Occurs Selecting Data Samples: Preference for Random Sampling When researchers study communication from a quan. titative social science perspective, they will focus on selecting groups of people or messages to target for their research. The term population refers to an entire 62 Chapter 4 Making Arguments for Association and Causality set of people or messages that have been selected based on a common characteristic, such as voters, teenagers, women, social media posts, or television audiences or some other shared group characteristic. Some popu- lations are quite large, such as all U.S. citizens, and others are much smaller, such as students who live in one residence hall on your campus. When populations are relatively small, it is easy for researchers to study the whole set or group. However, most texts and people we wish to study come from very large populations with millions of members, such as romantic couples, tweets and Twitter users, cancer patients, or nonprofit organizations. Because it is often not feasible to study all mem- bers of a population, a smaller subset must be selected. This subset is called a sample and sampling is the term applied to the process of selecting members from a population for a sample. In quantitative social science research, there is a great concern for extracting samples from the population in ways that ensure the sample will represent the larger population. Suppose that you were interested in surveying U.S. households in your area to see which mayoral candidate people intended to vote for in an upcoming election. The population may include several tens of thousands of households. Because you cannot survey every house- hold before the election, you must select a smaller but representative sample that will allow you to make ac- curate predictions. Ideally, samples are drawn from a sampling frame, a list of all members in the popula- tion. If the list you had was incomplete, for example, it does not list all the registered voters in our example, then your resulting sample would also be incomplete. This kind of a problem is called a coverage error, a term we will explore more fully in Chapter 7, “Survey Research. Once you have identified the population for your sampling, you must choose who will be in your sample. One clear way to remove any researcher biases from this process is to use random or probability sam- pling method in which selection is based on providing every member of the population with an equal chance of making it into your sample. Reducing or removing this selection process will certainly strengthen the ac- curacy of your study’s results because it increases the chances you will obtain a representative sample, one in which the sample characteristics are good estimates of the population from which it has been drawn. Through representativeness, you also have greater potential for extending the power of your study’s findings to popula tion members who were not tested. Random sampling also permits you to estimate specific sampling errors precisely; it is why random sampling is by far the pre- ferred method of sampling strategies for quantitative social science researchers. Random Selection Methods Using sample data to represent a population is funda- mental for quantitative social science research. In this section, we will introduce you to three random selection methods: (a) simple random sampling, (b) systematic sampling with a random start, and (c) stratified sam- pling. These methods, along with several nonrandom selection methods common to quantitative social sci- ence research, are listed in Table 4.3. Chief among random selection methods is simple random sampling, in which each person or message in the population has an equal chance of being selected for inclusion in a study. To use this method, you will select a subset of the population randomly from your sampling frame. You will assign a number to each member of population and then use a random number generator to identify a random sample from your population. You can find many free random number generators online. Some technologies do this for you; for example, when we discuss survey research, we will explain the process of random-digit dialing (RDD), which generates small to very large random samples of phone numbers for both landlines and cell phones. The second random selection method is systematic sampling with a random start. To apply this method, you would select the first element by chance and then Table 4.3 Types of Data Sampling Methods in Quantitative Social Science Random sampling methods Simple random sampling Systematic sampling with a random start Stratified sampling Nonrandom sampling methods Convenience (or volunteer) sampling Purposive sampling Snowball or network sampling Quota sampling Evidence for Causal and Associative Research Arguments 63 tion of select the remaining elements systematically from the total sampling frame. For example, if you are studying the population of students at your university and your sampling frame contains 10,000 currently enrolled stu- dents from which you want a sample of 1,000 students, you could select the first student randomly by generating a random number between 1 and 1,000 from a random number generator; if the number generator lands on 506, you would begin with that person’s number listed on your sampling frame. You could then select every 10th student from the registrar’s list until you reached 1,000 students. Simple random sampling and systematic sampling yield virtually identical samples, so most researchers opt for systematic sampling as the easier of these two methods (Babbie, 2001). It is the easier method because you only need to select one random number to begin the process. However, a . a potential problem with systematic sam- pling is periodicity, a recurring pattern or arrangement that exists naturally in the sampling frame. For exam- ple, imagine that you want to obtain a random sample of rooms in a large dormitory on campus. The rooms might be arranged in such a way that sampling every 10th dorm room would lead to the selection of only rooms near the stairwells (adapted from Babbie, 2001). If your research is about the effect of noise on studying in the dorms, then a pattern like that could lead you to select a biased sample, not a representative one. Our third random selection method, stratified sampling, is more refined and complex than either simple random sampling or systematic sampling with random start. Stratified sampling organizes a popula- tion into subsets of similar elements; we can then select elements from each subset using systematic or simple methods, we must first obtain an accurate sampling frame. But in this case, we know that some groups in the population share common characteristics. For example, you may know that two thirds of your university’s stu- dent population is female and one third is male. You can stratify that population by separating the sampling frame into females and males and then randomly select- ing sample from each group, using the approximate proportions of the sexes that represent your university’s population. Large polling companies such as Gallup fre- quently stratify voters based on geographic location to increase sample representativeness in terms of ethnicity, social class, urban and rural differences, types of occu- pations, and so forth. Generally, if you are measuring several variables in a very large population, you are more likely to achieve representative sampling if you stratify the population than if you use simple random or sys- tematic sampling (Babbie, 2001). But stratified sampling is usually much costlier that the other two methods and therefore used more infrequently. No matter which random selection method you use, you must be concerned about sample representa- tiveness. If your sample adequately represents its parent population, then the results of your study will be gener- to all other members of that population. We call this ability to generalize e findings to a parent population external validity, which we will discuss in the last sec- of this chapter. But for now, let’s contrast random selection methods with several nonrandom selection methods that are used commonly in quantitative social science research. Nonrandom Selection Methods Nonrandom selection or nonprobability sampling occurs when we select people or texts in ways that do not ensure that the resulting data sample represents its population. Quantitative social science research- ers have two basic reasons for choosing nonrandom selection methods. First, the constraints of setting or research question may make random selec- tion methods untenable or unethical, even when your purpose is to explain and predict communication in a larger population (Fink & Gantz, 1996; Stake, 1998). In such cases, you can use quota sampling, defined later in this section, as an alternative to random selec- tion methods. Second, you may want to represent a population but be unable to use one of the random selection methods because you lack the required time or money. In that case, use of nonrandom selection methods will weaken your ability to claim that your sampled results apply to some larger population. You will need to acknowledge the possibility that selection bias is a limitation in your research report You will find all four nonrandom sampling meth ods listed in Table 4.2 in quantitative social science re- search, but we will confine our discussion to two of the most commonly used, convenience and quota sampling. You will learn about purposive and snowball sampling in Chapter 11, “Making Arguments for Multiple Plausible Realities.” The first nonrandom selection method is called convenience sampling. Convenience samples are composed of whatever data are readily accessible to the researcher. Convenience samples do not provide good your data Purchase answer to see full attachment Explanation & Answer: 5 Questions Tags: independent variable Dependent Variable Data Source Data setting test the hypothesis User generated content is uploaded by users for the purposes of learning and should be used following Studypool’s honor code & terms of service.