Download Communication Research: Methods, Types, and Ethics and more Study notes Communication in PDF only on Docsity!
Errors in ordinary inquiry Innacurate observations, overgeneralization, selective observation, illogical reasoning, (gamblers fallacy) Theory deals with the logical aspects of scientific inquiry; data collection is the observational & data analysis brings them together. A scientific approach to communication is empirical- communication researchers base their knowledge on observations not personal values. Social regularities represent probabilistic patterns, general pattern needs not be 100%. Aggregate collective actions and situations of many individuals Communication studies is a field of research on the production and uses of symbols (both linguistic & nonverbal) in concrete social & cultural contexts to enable the dynamics of systems, society, and culture. PRODUCTION, USES, & DYNAMICS/ 3 interests: message production, transmission & meaning making 2) systematically examine the content of communication messages, 3) functions & effects of messages. CH Triangulation- use of several different research methods to test the same finding Types of Research papers Proprietary report- why its being studied mainly for who the research is for Working reports- presentation of findings with a request for comments Short research note- why the note is justified by the findings & what the findings are Article in a scholarly journal*- Book chapter-usually summarize a body of work Book- of research reporting The Quantitative research report Title- 10-12 words used for abstracting & indexing Byline- name & affiliation Abstract- less than 120 word summary includes: problem investigated, characteristics of the participants, methods employed, major findings & major implications Introduction- 1) opening to introduce the problem & its signifigance 2) the purpose of the study 3) background information in lit review form 4) end with the research questions/hypotheses that guided the study Method- 1) particpants of the study 2) materials 3) procedures what was executed in the stud Results- answers to the post questions and hypotheses. Present data, describe all aspects of analysis, tables charts and figures. Discussion- what has been discovered & what future research could do, conclusions & limitations References Communication Purpose Exploration- provide a familiarity with that topic. 1) satisfy curiosity/understanding 2) test the feasibility before a more extensive study 3) develop methods for a later study Description- observes & then describes what was observed Casual/Functional Explanation- explain things. Casual explanation addresses why and functional addresses how Understanding (reason-based) Units of analysis-what or who you can study (unlimited)
Individuals, groups, organizations, social artifacts (product of social beings or their behavior). Some social artifacts are physical traces. Erosion traces are physical (ex: floor tiles) Accretion traces (finger prints & smudges). Some are archival records, communication messages, or social interactions. The Ecological Fallacy –assuming about a group or set of systems. Or drawing conclusion about a group based on the individual. Research question- inquires about a phenomenon. Hypotheses ask about it. TIME IN RESEARCH DESIGN Cross-sectional studies- study a phenomenon in a cross-section of time. Exploratory & descriptive studies are often these. Cause-effect studies. Longitudinal Studies- observations over an extended period Trend studies- study changes within some general population over time Cohort studies- relatively specific subpopulations or cohorts as they change over time. Often an age group. Panel studies- examine the same set of people each time. Ex: voters every month SPACE IN RESEARCH DESIGN Field Dependent Research- reaction in a physical setting. Evaluation research often does this, asses the impact of social interventions such as public health media campaigns. (specific actions is particular contexts) Field Independent Research- setting is either irrelevant or tightly controlled. Cause/Effect research. Ch 3 Paradigm-fundamental points of view characterizing a science (ex Darwin & Newton) Positivist Paradigm (comte) viewing society the same way science views natural phenoms. Features the belief in objective reality, through empirical observation, study of variables, theories for prediction, explanation and control, the search for generalized laws, and quantitative data observations. Focuses on the relationship between variables and their attributes Objective reality – out there independent of the researcher. Cause and effect way to explain phenomena Attributes- characteristics of qualities that describe an object or phenomena (ex loud & chatty) Variables-features or characteristics about individuals or phenomena. Logical groupings of attributes. So someone who is obnoxious (variables) has the attributes of loud and chatty. -independence/exogenous/predictor (IV) or X Attitude similarity -dependent variables/endogenous (DV) or Y Relational satisfaction X -- Y (x is a predictor of Y) -moderator Variables whose value(s) alter the strength and/or direction of the relationship between the IV and DV -mediator Explains the relationship between the IV and DV Positivist Theory – seek to explain the cause-effect relationship between variables. An attribute on one variable should cause an attribute on another variable. Causality- proved through 1) a causal relationship between two veriables in which the cause precedes the effect (IV before DV) 2) the two variables be empirically related to eachother
Anonymity cannot link a given response to a given participant Confidentiality the researcher can link the responses but will not do it publicly Deception- sometimes needed for research. If disguise self muts do it for compelling scientific or administrative concerns. Deal with through debriefing: interviews to discover any problems generated by the research so that these problems can be corrected -try not to describe participants in ways they can identify themselves in research IRB ensures that the risks face by humans is minimal Tearoom study- Humphreys was watch dog for gay hookups. Then tracked down these men by license plate number & went to their homes disguised to ask them more questions. Never been resolved Milgram – had 40 men be the “teachers” and administer shocks to pupils in another room. Many men shocked the pupils into possible death. The study had no real pupils or shocks but still emotionally distressed particpants Ch 6 -has only one value Properties -quantitative/continuous or qualitative/discrete -manipulated (some would argue that all manipulated are discrete) or not manipulated -manipulation is putting it in different groups??? -quantitative do not have active control over our subjects Types of variables -independence/exogenous/predictor (IV) or X Attitude similarity -dependent variables/endogenous (DV) or Y Relational satisfaction X -- Y (x is a predictor of Y) -moderator Variables whose value(s) alter the strength and/or direction of the relationship between the IV and DV -mediator Explains the relationship between the IV and DV Causality -there is a relationship between the variables -the IV must precede the DV (time order) -non spuriousness (something else explains the relationship between the IV and the DV) Research Questions Types of Research and Research Questions -exploratory or descriptive studies --how do people define family?
-explanatory studies --what is the relationship between perceptions of “normal” and ideal family communication? -when do you advance a research question as opposed to a hypo? --when you either don’t have enough info from the literature or from your own conclusions from the literature to make some sort of claim as to what should or could happen between a couple of variables Hypotheses -tentative answers to research questions -importance There are falsifiable and non falsifiable theories MUST PICK FALSIFIABLE THEORY FOR OUR PROJECT MUST INCLUDE A RESEARCH QUESTION IN PROJECT Form of Hypotheses -phrased as an affirmative statement (the null hypothesis is what is not) -assumes a relationship exists -speculated about the nature and form of the relationship Can not decide a relationship just use a research question -states the relationship in associative (not casual) terms (a hypothesis does not establish time order) -is consistent with prior research and theory TYPE ONE of hypotheses: Continuous statements H1: greater feelings of marital satisfaction (continuous) will be associated with lesser feelings of job satisfaction. (Continuous) -is not just one thing from start to finish TYPE 2 Difference statements H1: Married couples will report high levels of relational satisfaction than dating couples. (the dependent variable should not be categorical) Categorical variable = married couples over dating couples TYPE 3 DIRECTIONALITY Directional H1: married couples will report higher levels of relational satisfaction than dating couples. -it is directional because it specifies the direction of the nature of the relationships (also called one tailed) TYPE 4 Non directional H1: married couples will report different levels of relational satisfaction than dating couples. -it is nondirectional that a relationship or a difference will be found. Two tailed. DIFFERENCE STATEMENTS WITH DIRECTIONAL NESS DO NOT HAVE POSITIVE OR NEGATIVE Three issues with research ethics: -how research participants should be treated -how data are collected, analyzed, and reported -how research results are used (unethical use leads to career lose)
The interchangability of indicators: if several indicators all represent the same concept then all of them will behave the same way that the concept would if it were real and could be observed Dimensions a specifiable aspect or facet of a concept ex: feeling dimension of compassion The progression of measurement (vague – specific) Conceptualization, nominal, operational, ratio Conceptualization, nominal definition, operational definition, measurements in the real world Nominal definition (4 requirements) Specify indicators (sign of presence or absence) Specify multiple dimensions Identify related concepts Are consistent with past usage in theory & research Operational choices: what range of variation to consider, what levels of measurement to use, and whether to depend on a single indicator or several. Attribute a characteristic of quality of something ex feminine Variables logical sets of attributes ex gender Every variable should be exhaustive and mutually exclusive (classified in terms of one and only attribute) Levels of measurement Flirting can be defined as messages and behaviors that are received by a recipient as purposely endeavoring to gain the attention of the recipient and intentionally revealing an affililtative desire for the recipient Conceptual (nominal) definitions -specify indicators (sign or presence or absence) -specify multiple dimensions -Identify related concepts -Are consistent with past usage & theory Importance of conceptual definitions -lead to operational definitions (a definition that explains how something is measured) -informs units of analysis Individuals, groups, organizations -inform research design Cross-sectional specific cross section of the population in a specific time and place & history Longitudinal (trend, cohort, panel) over a period of time -nominal measures: variables with attributes that have only the characteristics of exhaustiveness and mutual exclusiveness are being measured at the nominal level ex birthplace, major, hair color. Offer only names or labels. -ordinal measures: attributes can be ranked logically ex social class or class standing. -interval measures: when the actual distance separating attributes matters -ratio measures: the attributes composing it are based on a true zero point. Ex: age, number of times married & number of children. --you can make a ratio an interval but you cannot make an interval a ratio. -some indicators are single (one observation or question) others are multiple that combine several pieces of information to form a composite, several pieces that measure the variable. Ex college performance.
PAGE 119 HOW TO ACTUALLY PERFORM
-reliability does not ensure accuracy -for reliable measures ask only about things the respondent knows and be clear. Test/retest method you should get the same results every time. So retesting helps. Alternate form method researcher develops two different forms or versions of the same measure from the same pool of measurement items. Use with paper and pencil questionnaires. Internal consistency method various items in a given measure should correlate positively with one another, or consistently. Split half reliability two sets of a test that measure the same ex: a quiz on love that has two different sets of qualities that determine love the same Item total reliability compares the consistency of each item in a measure with the total score. Ex: how likely selecting one characteristic means determining love
- use previously established measures to ensure reliability -correlation coefficients: 1 being the most desirable. Acceptable is around. Content validity how well a measure covers the range of meaning or the dimensions included within the concept Face validity particular empirical measures may or may note jibe with our common agreements and our individual mental images concerning a particular concept. Expert panel validity group of experts in the area evaluates a measures adequacy. Ex census Factor analysis analyzing the multidimensional structure of their measures. Criterion validity asses measures validity through its relation to some external criterion Predictive validity measure must predict future behavior Concurrent validity a new measure corresponds to the measurement made by an established measure Construct validity based on the logical relationships among variables Convergent construct validity the measure converges positively with measures of theoretically related variables Discriminant construct validity the measure diverges from or correlates negatively with measures of theoretically different variables SUGGESTED FORM representational validity whether a measures categories are meaningful for the people who are being assessed. Types of Concepts (family of conceptions created by a researcher) based on how hard it is to define Direct observables - directly measurable (hair color etc) Indirect observables – indirectly measurable (can’t see it but can figure it out) Constructs – based on observations but not directly or indirectly measurable (ie IQ) not a “real” characteristic Writing Research Reports -introduction -literature review Justification of the study Discussion of theoretical bases Review of relevant concepts Hypotheses and research questions
N ominal level O rdinal level, rank ordering variables from greatest to least or vice versa (likert scales) I nterval level allow us to categorize things and can also rank order but in addition interval level variables allows me to determine something about that rank ordering. Can infer the distance between things. (likert scales when analyzed) R atio level, identify categories, rank order, say something about the distance between each categories, has an absolute zero point. Nuances of measurement Higher level v lower level Converting measures Need for multiple indicators (sometimes) multiple measures of something & in different ways Determining trustworthiness of measures -validity “refers to the extent to which an empirical measure adequately reflects the real meaning of the concept under consideration. Does it tap the phenomena? -reliability is whether a particular technique, applied repeatedly to the same object yields the same result as last time. -cannot have categorical/categorical. Whatever the fuck that means. Example: 1- not satisfied, 4 satisfied, 5 very satisfied. Problem is 4 & 5 are both satisfied. So it is not utilizing the scale. (you can never know the true value) an instrument can be reliable and valid, just reliable, or just valid. You would want one that is reliable over one that is valid if you had to choose. But if a measure is not valid it is not reliable. Types of Validity Content -face validity, just look at it & if it looks like it measures what its supposed to measure -expert panel validity Criterion -concurrent validity. Does the measure discriminate between groups that are known to differ? -predictive validity, does it produce results Construct -convergent validity, does this measure produce outcomes similar to measures of the same contraint -discriminant validity, does this measure produce different outcomes than measures of related but conceptually distinct constructs Types of Reliability Paper pencil measures Test-retest reliability. Has a priming (pre exposing them) effect Alternate form reliability (asking a slightly different question) Internal consistency
Cronbach’s alpha (α )
Ranges from o -. If it is .85 or greater we can say it is a reliable measure. But this is arbitrary so many argue that above a .7 is reliable too Observed measures
Inter-observer or intercoder agreement. Don’t want coder bias. Formula: observed agreement – expected agreement over expected agreement Basic Research ( trying to find out some unknown thing about a phenomenon) vs Applied Research ( what we want to know because we want to fix something) Population- the aggregation (the group of people for whom our study would apply) of elements from which a researcher selects a sample (babie, 2004) Theoretical population Accessible population Sample A sample is a subset or portion of a population (keyton, 2006) Why does sampling matter? Ann landers asked her readers: if you had to do it again would you have children? 70% of parents say kids not worth it? Sample Size Factors determining sample size Heterogeneity of the population (have to have equal representation) Desire precision (the more precise you need to be the larger the sample size you need) Sampling Error Sampling error is the degree to which a sample differs from population characteristics on some measure Sampling error ALWAYS occurs because researchers do not collect data from the entire population Sampling error, sample size, and population size are all linked, such at that the closer the sample size is to the population size the smaller the sampling error Sample size in relation to sampling error Sampling error depends directly on sample size Acceptable sampling error is 5 Thus the goal is a 95% confidence interval- or a 95%(degree of accuracy) chance of predicting the result for the population based on the result from the sample Generating a sample Determine theoretical & accessible populations Determine the sampling frame List all cases (if possible) (cases are all members of the population from which you could sample) Decide on a rule defining members Select units based on sampling design Probability Nonprobability Probability sampling- “using a statistical basis, and the most rigorous way, for identifying whom to include as part of a sample” (Keyton, 2006, p.122) Often for experimental research this is needed. Convenience sample is often used for surveys. Can be hard to meet. It is the ultimate goal though Nonprobability sampling is sampling that does not rely on any form of random selection (keyton, 2006, p. 125)
Typically self-report method Procedures are systematic Unstructured Structured Research is descriptive or explanatory. Basic or applied Produces associative data Produces primarily quantitative data Allows for both primary and secondary research Questionnaires Delivery: face to face, mail/phone, internet, computer assisted (means that the computer has all the text & you can read the question or based on your responses to one particular question the computer decides which is next) Pros Accessible to general public Ease of sampling Economical Offers anonymity Standardizes responses Cons Impersonal Restricted range of responses Response rates (about 30%) Response bias (because of the little response, only a certain type are willing to fill) Can be labor intensive Quantitative interviewing cant be paper and pencil Delivery Face to face Telephone Computer assisted Pros Holds respondents accountable Interviewer is present Excellent for examining complex, personal or latent phenomena (something that is an underlying idea ) Flexibility Opportunity for structured or semi-structured interviews Cons Expensive (have to compensate, train/.pay interviewers, facilities) Researcher has less control, cant control interviews or respondents Potential researcher (interviewer) bias Data may be difficult to analyze Typically requires a large staff Requires careful monitoring of standardization Survey research designs Cross sectional- looks at a cross section of the population at a particular point in time. One measure at one time point
General cross sectional Contextual cross sectional Longitudinal- could do it over a period of years or months. Means multiple measures at different time points Trend Panel Cohort Developing good survey questions Expert understanding of the phenomenon under study Knowledge about the population of interest Knowledge of the data collection procedures Basic Assumptions of survey research The survey should Not make unreasonable demands on the respondent Not having a “hidden agenda” Be used to collect data that you cannot get from other sources Indicate that it is backed by the authority of some institution or individual Clarity of Instructions The survey should Be prefaced by explicity, directed instructions Use clear and simple language Offer sample responses when necessary Offer clear and concise directions on how to complete and return the survey (if applicable) Questions should be.. Straightforward and non biased Singular (no double or triple barreled questions) Do you agree that the department of comm. studies has great faculty and is understaffed? Might agree with one but not the other Question Construction Questions should be Non-threatening: have you ever cheated on an exam in your major classes ( THIS IS THREATENING) Non-leading: don’t you think the best way to get a job when you graduate is to major in communication studies? (this is leading) Question order Easy to hard Thematically (all the question from one instrument should be grouped together) Add filler questions when necessary (example: to surround a threatening question) Demographic questions should be first or last Last if a lengthy survey (what he recommends) Rules for questionnaires The items in the questionnaire should be presented in an attractive, professional, and easy-to-understand format. All questions and pages should be clearly numbered. Separate into sections restarting at one so it does not seem so daunting
d. what would count as evidence for or against the hypothesis.
