Com 309-Notes 309 - Comstrat - Quantitative Research Methods
Popular in Quantitative Research Methods Notes
309 - Comstrat - Quantitative Research Methods
verified elite notetaker
verified elite notetaker
verified elite notetaker
Popular in Communication Studies
verified elite notetaker
This 55 page Bundle was uploaded by Ashton Zeth on Monday February 23, 2015. The Bundle belongs to 309 - Comstrat - Quantitative Research Methods at Washington State University taught by Graham Dixon in Fall. Since its upload, it has received 363 views. For similar materials see Quantitative Research Methods Notes in Communication Studies at Washington State University.
Reviews for Com 309-Notes
Can you just teach this course please? lol :)
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 02/23/15
Com 309 Notes MWF 210300 PM 09242014 Homo egaster Early ancestor of homo sapien o 16 million years ago 0 Built res and hand spears o Closely living relatives chimpsorangutan What is research 0 Coreunderstand world see associations causes 0 Making sense of causation o Fixing amp creating based on casual interpretations Types of research 0 Academic universities For the sake of understandingnot much implementation Publication in journals 0 Businessmarketing eld Business invest l understand message Communication with customers stakeholders Rely on work by academic o MedicalPublic Health Cures pro tnonpro t Foundations of Research 0 Epistemology How is knowledge acquired O O O O O Method to pursue Use experiment use interviews Ontology nature of being how we exist Nature vs Nurture Violent video games Axiology values in research Motivation for performing research E run study of test aggression after playing violent video games 0 belief that aggression can occur from outside in uences A research done due to concern about affecting adolescent behavior 0 Theory 0 O O 0 Based on collection of studies Scienti c theory and hunch not the same Can provide direction for how things work Will lead guide to making hypothesis Ex Social norms theory l behavior is largely in uences by our perception of what others do Based on SNT advertisements that use narrative messages ie everyone drinks coke will effectively persuade people 2 buy 0 Empirical Research 0 O O Empiricism an epistemic view that knowledge is derived from measurable experiments usually sensory Empirical Research research that involves collection of data based on directindirect observations Descriptive describe Ex 25 of Americans know the earth revolves around the sun 0 Analytic inferring relationships estimating effect of x and y making causalassociative claim experiments surveys correlations Ex quotwatching American Idol increase the likelihood of buying coke due to product placementquot 0 Study Overview 0 First watch Ted Talk Laptop users verbatim note taking 0 Second same as rst with no verbatim note taking Same as previous results 0 Third does different style affect test scores YES hand notes takers had better retention than laptop takers Research Process Part 2 0 Scienti c Journals Primary Sources of research 0 Impact Factor the average of citations per article over a two year period 0 General Journals have larger IF than smaller more speci c journals 0 Secondary Sources 0 Not considered peer reviewed sources 0 Report on peer reviewed sources 0 Often exclude important details from scienti c studies variables methods limitations 0 Peer Review 0 Manuscript is submitted to journal o Is it signi cant 0 Journal EditorAssistant Editor accept for peer review or reject 0 Editor invites experts in eld to review paper 0 Important Correct 0 Reviewer makes recommendations to editor About 90 of papers are rejected 0 Decisions 0 Accept extremely rare 0 Accept w minor revisions very rare 0 Review amp Resubmit most published articles were rst asked to revise signi cantly Can request certain reviewer not to review 0 Reject most common outcome with feedback journalAcceptancetruth 0 Daryl Bem s ESP study l psychic phenomenon in prestigious psychology study Multistudy paper found evidence of precognition and premonition Failed replications 710 repetitions all but 1 fail to replicate Jury is out Not uncommon for article to retracted Ethical violation Article about vaccinations and autism retracted for not following ethical protocol Retractionwatchcom medical psychological Predatory Open Access Journals 0 Charge fee to publish 2000 0 Lax or nonexistent peer review 0 Not indexed in major directories 0 Free access to all journals for everyone lssue desperate professors can pay to publish OOOOOO O O O O Scam Alert check for predatory access Fake study with fata error ie no control group sent to hundreds of open access journals Over 50 accepted paper Shows some quotpeerreviewed journals lack of rigor Bigfoot l DeNovo only article created own journal 0 As StratCom professional researcher 0 Know difference between peer reviewed journal and secondary source 0 Article include original data 0 Be able to discern the goodbad journals ls article open access ls article indexed in reputable directory Strong impact factor ALWAYS examine a peer reviewed with critical eye lf retracted could ruin career criminally convicted Research Process Part 3 0 Types of Scienti c Articles 0 Original Research Articles Presents empirical findings ndings based from directindirect observation Uses theory in order to make compelling rationale for doing study Make amp test hypothesis Offers recommendations for future research Ex Longhanded note taking vs laptop Review Article Provide a substantive review of a particular domain s used for understanding what the current state of evidence is for a particular theory of research area DOES NOT involve gathering data DOES NOT involve data analysis Ex Impact of alcohol advertising and youth use Meta Analysis Collection and analysis of empirical studies Unlike review meta involves statistical analysis Controls for biases such as publication bias the tendency for researchers to publish signi cant ndings rather than non signi cant ndings Determines whether based on collection of research there is evidence of X Ex Link between media use and body fatness Research Briefs Similar to original research article Much shorter 1 page1 paragraph Typically dose not include indepth discussions of theory Represents works in progress Theory Article Presents theoretical propositions No collection of data Suggests a new or alternative theory Abstract summarize entire paper Intro why being conducted Literature review previous research Methods Testing hypothesis Results ndings Discussions Interpretation 0 Conclusion Sum up nding Dissemination of Research Findings 0 Narrative Journals Press release USA Today Journalist might interview authors Journalist might bring in other experts 0 News Media Traditional Media Alternative Media Problems with News Coverage 0 News of scienti c research often acks important context limitations author info journal info methodology 0 Sometimes highlights controversy where none exists eg early reporting of climate change autism amp vaccine contro Why are there issues with science reporting New Org 0 Audience needs 0 Economic needs 0 Controversy sells 0 Risk ses Science journalism down for prints Reduced of skilled science journalists More quotbeat reportersquot tasked with reporting on scienti c issues Back and Forth OOOO 0 Scientists and researchers often bad at communicating their research to lay audiences o Journalists dif culty in presenting nuanced research within nite space with editorial demands Understanding articles formation 0 Ethics 0 Textbook De nition moral principles that govern a person39s or group39s behavior branch of knowledge that deals with moral principles 0 Ethics in Research 0 Participants Safety Is the participant safe from harm Are risks minimized Con dentiality Is the participation kept con dential Informed Consent Is the participant made aware of study and is asked for consent 0 Researchers Con icts of Interest Is the researcher gaining anything from this research Plagiarism Is the research done with proper citation Fabrication Are methods done without manipulation and fabrication 0 Important Moments in the History of Research Ethics 0 19305 1945 Nazi Experiments 0 lnfected children with tuberculosis for research After the research was conducted on the children and they got the information they needed they murdered the children 0 Hypothermia research Wanted to test new devices for pilots that needed to eject from the airplane dunked participants in tubs of cold water o Josef Mengele39s twin research at Auschwitz Many of the twins ended up dead 0 19305 19705 Tuskegee Experiment O 0000 0 Clinical study 19321972 by US Public Health Service Understanding progression of syphilis Never told they had syphilis Never treated for syphilis Researchers recruited mostly African American men in Tuskegee who had Syphilis Told those men that they had quotbad bloodquot Researchers wanted to see how syphilis progressed naturally The African American men didn39t know they had syphilis The researchers never treated the men and let them die naturally Led to an official apology in 1997 by Clinton 0 19405 Guatemala Syphilis Experiment 0 0 Study by US Public Health Service Researchers Infected people with syphilis and other diseases 0 Wanted to see the effectiveness of penicillin on the diseases 0 US formally apologized in 2010 o This study was formally kept quiet until 2009 0 Stanford Prison Experiment 1971 O O 0 Prison abuse experiment taken place in Stanford psychology building Randomly assigned college males to be prisoners or prison guards in mock prison Psychological and physical abuse occurred Prison guards took over their roles very quickly 0 Experiment abruptly ended only after 6 days Prisoners ended up with emotional issuesbreakdowns post experiment Whole purpose of the study prison guards and prisoners were not required to do ANYTHING they fell into their roles and did what they were told This study crossed the line in harming participants Milgram Study 1963 O A quotteacherquot would ask a quotstudentquot questions quotstudentquot was in another room from the teacher If the student got the question wrong they would receive an electrical shock each time they got a question wrong they would have to increase the level of the electrical shock 15 bolts up to 450 bolts The highest bolts given would equal death quotStudentquot wasn39t actually being shocked in the other room the quotteacherquot giving the shocks was listening to a recording of a quotstudentquot yelling 39No Stop I quit39 Researchers wanted to see if the quotteacherquot would continue with the shocks because the researcher told them to continue Study examined how people respond to authority If person tells you to do something immoral would you do it because they have authority People delivered what they thought were real electric shocks to a study partner Researchers wanted to see if the quotteacherquot would stop giving electrical shocks to the quotstudentquot because it was an immoral act to shock them Outcomes of the Studies Stanford Prison Experiment Milgram Study 0 00000 Creation of the Institutional Review Board IRB Reviews all research on human and animal subjects Ensures participants are not harmed psychologically or physically Risk posed by the study are minimal and controllable Ensures participants are aware of risks a study might pose Ensures that participants knowingly participate in the study and give consent Informed Consent Procedures 0 Provides information about study Risks and Bene ts Con dentiality What the participant is asked to do 0 Discusses risks associated with study 0 Ask for participants39 consent Provides documentation that you agreed to do this study IRB says that special human populations require extensive research review they are very critical of your research if it involves 0 Children age 17 or under 0 Prisoners Ethics in nonacademic research 0 Governing boards FCC Federal Communication Commission govern advertisements how ads are portrayed FDA Trade Organizations 0 Example in questionable research ethics OK CUPID 0 Pairs told they had 90 match when their match was actually 30 compatibility Answer 17 increased chance of chatting 0 Pairs told they had a 30 match they their match was actually 90 compatibility Answer likelihood of conversation dropped from 20 to 16 0 Because they were real people they did not know they were part of a study is this an ethical breach 0 Facebook Emotion Study 0 If you manipulate people39s news feeds with more positive posts or more negative posts what type of post would you be more inclined to post in the future 0 Questions you need to ask yourself before conducting research 0 What population do you prefer using Is it a vulnerable population 0 prisoners children under 17 How will you obtain informed consent Will they be aware they are participating in a study How much information will you tell them Potential risks that participants are exposed to in your study How can you minimize these risks How could your ethical practices impact your study39s validity Surveys 0 O O 0 Provide selfreport data Used in both formative and summative research Formative Research measurement of phenomena at one point in time ex Cross sectional surveys Summative Research measurement of an intervention ex Experiments 0 Content Analysis 0 Common method that shows frequencies and relationships between quotcodesquot Codes are variables that help de ne the properties of media Media can be in text or visual forms All very subjective ones opinion of quotviolencequot will be different than another39s opinion of quotviolencequot Experiments 0 All helpful for making causal claims 0 Can use selfreport surveys but also behavior 0 Experiments fall under the summative research category Qualitative Focus Groups 0 Provide insight into people39s decision making 0 Allows for more intimate connection with participants a real chance to talk to people 0 Fairly inexpensive 0 PR and Ad groups use focus groups for insight Theory 0 Based on collection of studies Scienti c theory and a hunch not the same Can provide direction for how things might work Will leadguide one to making a hypothesis 000 Theory in Action Based on cultivation theory we can then make more speci c predictions People who watch crime shows are more likely to believe crime occurs in real life Grounded Codnition Theorv humans represent abstract concepts in terms of concrete experiences in physical world Ex Stop light colors a Red Stop more negative connotations a Yellow Caution D Green Go more positive connotations Based on grounded cognition theory we can then make more speci c predictions n Ex Using green labels will be more effective than yellowred labels due to coor39s associations Hypotheses 0 Based upon prior research and theory 0 Make predictions 0 Direct and Precise 0 Good example quotthere will be a positive relationship between watching Law and Order and perception of real life crimequot 0 Bad example quotthere will be a relationship between watching Law and Order and perceptions of real life crimequot Making prediction after data is collected is called The Texas Sharpshooter Procedure 0 Hypothesis should not predict null ndings Bad hypothesis watching televised sports does not increase your likelihood of playing sports Good hypothesis watching televised sports increases your likelihood of playing sports 0 NULL FINDINGS Dif cult to prove a negative Not interesting 0 Hypothesis testing Now that you39ve made the hypothesis what could be the outcomes Using statistics we are interested in whether there is enough evidence to reject the null hypothesis in favor of the alternative hypothesis If the null hypothesis is true there is NO EFFECT If alternative hypothesis is true there is AN EFFECT o What determines whether we reject the null hypothesis ex Finding support for our alterative hypothesis In quantitative analysis we use SIGNIFICANCE TESTING based on statistics and probability Requirements for quotsigni cancequot are arbitrary 0 Typical requirements for signi cance in social sciences PValue the percent likelihood of rejecting the null hypothesis when in fact the null hypothesis is true Alpha Level pvalue cutoff point for signi cance test Typical Alpha Level 05 meaning if we get a plt05 nding we consider the nding signi cant The smaller the pvalue the more evidence there is that you can reject the null hypothesis Ex P 04 there is a 4 chance the null hypothesis is true 0 Research Basics variable amp measurements 0 Variable Operationalization 0 Start with unanswered questions 0 Make theory guided prediction hypothesis 0 Now time to identify and operationalize our variables 0 From a hypothesis to measureable variables 0 After creating a hypothesis measurable variables must be constructed o A precise and direct hypothesis makes this process easy Ex quotThere will be a positive relationship between watching Law and Order and perception of real life crimequot Variables a characteristic or value that varies or changes 0 Independent Variable the variable that is controlled andor manipulated o Dependent Variable connected to the independent variable measure as an outcome of independent variable The relationship variables the in uence of X on Y where xindependent and ydependent Hypothesis green food labels will have greater positive in uence on nutrition perceptions than other color labels 0 Independent color of label 0 Dependent nutrition perception 0 measuring the effect of color on nutrition perception Hypothesis a soda campaign that eicits disgust will have higher reca than other campaign messages 0 Independent disgusteliciting campaigns 0 Dependent information recall 0 measuring effect of disgusting campaign message on information recall Independent variable should always precede the dependent variable Once you ve given identi ed the variable how do you measure them 0 Operationalization taking a variable and narrowing into a de ned measurable unit Ex Perception of crime estimated likelihood of being crime victim Variable l Operationalization oAggression l aggressive acts punching hitting cursing aggressive thoughts hateful 0 Political Ideology support for a political party republicandemocrat views on issues gun control abortion 0 Religiosity How often do you attend a place of worship Belief in supreme being willing to donatetithe Operationalization l Measurement 0 After we ve operationally de ned the variables we then must consider measurement Measurement refers to the way in which data are collected selfreport behavioral text observations audiovisual Data the representation of measured units 0 Types of Data 0 Nominal categorical data made into categories Gender ethnicity political party 0 Continuous data measured on scale Age emotional arousal attitude 0 Ordinal data is made ordinal scale Socioeconomic status grade level 0 Issues in Research 0 Types of Validity 0 Study Validity lnternal are the variable relationships true External can you results occur in other populations ie occur outside of a lab 0 Measurement Validity Are your measurements valid This is particularly important for social science research 0 Internal Validity o Are the relationships between the independent and dependent variable true o Is the relationship a casual one and without confounding o A challenge for survey research O O O Hypothesis playing violent video games lead to real life aggression Survey collect variable measuring violent video game playing and selfreporting aggression Results the more one plays violent video games are causing the aggression or it could be that aggressive individuals play more video games It could be another factor not measured that plays a greater role Therefore this study has LOW INTERNAL VALIDITY Generalizability can the results from a study be generalized to other persons times places a Difficult for experimental research 0 External Validity O O O O Hypothesis playing violent video games lead to real life aggression Experiment randomly assign people to either play a violent video game or nonviolent game Results violent video games players report greater aggressive tendencies than nonviolent players Problem do these results occur in real world where people self select into whether they play or don t a video game Therefore experiments often have LOW EXTERNAL VALIDITY Some studies allow for greater internal validity but 0 Measurement Validity 0 Construct Validity your measure is accurately labeled Ex Does the frequency of attending a place of worship actually measure religiosity 0 Content Validity extent to which the content of the measure contains the full scope of the phenomenon you intend to study Ex How many measures might you need to capture the full scope of religiosity o Criterion Validitv extent to which your operationalized variable performs the way it should according to an external standard Ex Create political ideology scale You test with actual self identi ed liberalconservative to see how well your operationalized variable performs o ReHabH y o TestRetest Reliability concordance of results when same 0 instrument administered to same individualgroup on separate occasions InterRater Reliability degree to which different ratersobservers give consistent estimate of the same phenomenonsubjectgroup Two people rating same picture l agreement on rating 0 lntraRater Reliability degree to which some raterobservers give consistent estimates of the same phenomenonsubjectgroup at different times One person rating one picture twice l same rating Threats to lnternal amp External Validity Testretest Reliability same results with the same person using the same instrument InterRater Reliability Do different people give consistent estimates of the same phenomenon lntraRater Reliability Does the same person give consistent estimates of the same phenomenon at different times Bias 0 Bias caused by systematic errors that can in uence study validity 0 Different from random errors which occur by chance alone Selection Bias 0 Systematic Error in how people participate in a study Are biases in measurement which lead to the situation where the mean of many separate measurements differs signi cantly from the actual value of the measured attribute 0 Self Selection 0 Selective Attrition uneven dropouts The tendency of some people to be more likely to drop out of a study than others Information Bias 0 Bias resulting from measurement error 0 lnvalid instruments 0 Recall bias 0 Social desirability response bias The tendency of respondents to answer questions in a manner that will be viewed favorably by others Favoring good behavior and underreporting quotbadquot behavior Confounding Bias o Is relationship between X and Y causal 0 Or is the relationship due to chance 0 Or is the effect due to some third unobserved factor spuriousness Ice Cream and Drowning 0 Fact when ice cream sales go up drowning increase 0 Does ice cream cause drowning o No the lurking variable is a hot summer day which boosts ice cream sales AND swimming Criterion for Causation o X and Y must be associated Association 0 X must precede Y temporality o No other factors cause Y speci city Strategies to Minimize Confounding o Attempt to control for all possible third factors 0 Randomize participants eliminates selfselection Sampling Part 1 0 Types of Measurements o Behavioral type of population People 0 SelfReport type of population People 0 TextAudio Type of population Text pictures audio etc 0 Sampling Structure 0 Theoretical Population who do you want to generalize to US Adults Children 1216 etc 0 Study Population What populations can you get access to 0 Sampling Frame How can you get access to them Telephone directory records patient les etc o The Sample Who is in your study 0 Different Ways of Sampling 0 Random Probability Sampling Each member of the theoreticaltarget population has SAME LIKELIHOOD of being picked Random sampling and randomization ARE DIFFERENT About 10000 per question Opinion polls 0 Convenience Sampling Not Random Using a particular population Threats to validity Approachingasking people Cheap George Gallup Sampled only 50000 US voters Correctly predicted Roosevelt landslide victory Catapulted his company into political stardom Featured in Time Magazine El El El What went wrong for literary digest Large sample size didn t mean a representative sample size Response rate of readers was low out of 10 million questionnaires sent only 26 million returned Gallup s sampling method corrected for overrepresentation of Landon s supporters Landon s supporters were more likely to return survey than Roosevelt supporters Nonscienti c means not random 0 Snowball Sampling Not random Population de ned by its participants Threats to external validity Cannot be sure that sample is good representation Researcher contacts three individual participants l Participant 1 l 11 12 13 recruited by participant 1 Participant 2 l 21 22 23 recruited by participant 2 Participant 3 l 31 32 33 recruited by participant 3 Sampling Part 2 Sampling and Validity 0 External Validity How generalizable is your sample to your theoretical population Some sample sizes might exclude either due to chance or by selection people of certain backgrounds Random Sampling Margin of Error Larger the sample the smaller the error Usually error of is acceptable for political polls 0 Internal Validity Are the relationships between variables true Underpowered Sample sizes can lead to type 1 false positive and type 2 false negative errors Why Small sample sizes have greater chance of overrepresentation Overrepresentation of one factor can lead to false positive Increasing sample size can help reduce overrepresentation of a factor 0 Sampling Procedures Convenience snowball random 0 Power The sample size 0 Determining Our Sample Size 0 Signi cance Is the effect statistically signi cant 0 Effect Size While the effect is signi cant what magnitude is its size 0 Larger the effect size the smaller the sample size needed 0 Effect Sizes 0 Small The Facebook emotion study n700000 0 Medium The effect of violent media and aggression n300500 0 Large The difference between average height of US Men 1776 cm and women 1632 cm n4060 Think of what sized effect you expect assuming on exists 0 Where do we come up with these predictions 0 Prior research and theory Calculating Sample Size 0 Make an assumption that an effect exists 0 Power assuming an effect exists the probability that an effect will be detected based on the sample size you choose 8 power 80 likelihood that you will nd signi cant results Never have 100 power because can never be 100 Power corresponds with sample size and effect size Closer power is to 1 the larger the sample size For small effect sizes if you want your study to have 9 power you will need large sample size OOOOO Equidistant same distance between extremely satis ed very satis ed and somewhat satis ed 0 Good example very dissatis ed dissatis ed somewhat dissatis ed RecaH 0 Survey information bias due to invalid instrumentation Semantic equidistance Suggestive questions Doublebarrel questions When you ask two different questions but expect one answer Leading questions Different responses depending on how question is written 0 Information Bias 0 Invalid instruments O 0 Recall bias Social desirability Self Report 0 0 Bene ts Quick responses No need for behavioral measures Can quickly compile data Drawbacks DunningKruger effect Social desirability response Recall bias DunningKruger Effect O 0000 CO Perceived knowledge actual knowledge Overestimate Underestimate Do a certain type of people overestimate or underestimate Prompted by quotAmerica s Dumbest Criminalquot In 90 s man devised quotfoolproof plan rub lemon juice on face he would be invisible David Dunning and Dr Kruger Is there an effect where people with very low knowledge actually over estimate level of knowledge Generally yes people with low knowledge over estimate their knowledge As well as reverse l those with high knowledge underestimate their knowledge Remedy to DunningKruger effect Ask items that tap into knowledge Make sure these items are good operationalization of your target variable Ask questions related to topic Knowledge of Science amp Technology quotSunscreen protects from what radiationquot 0 Social Desirability Bias 0 O O O O Overunder estimating response due to social pressures Threatens measurement validity Example mental health questionnaire Lying in online dating pro les Men overreport height and women underreport weight Overcoming social desirability bias Ensure participant con dentiality Make surveys anonymous 0 Recall bias 0 O O Often bad at recalling events Sometimes false memories occur lmplanting memories Subjects were implanted with a false memory of a hot air balloon ride l subjects brought with them a picture of themselves as a child and the subject s picture was pasted onto photograph of hot air balloon Research called parents and ask if child ever experienced hot air balloon ride those that had were removed form study 50 of subjects recalled that the ctitious event Remedy to recall bias Make sure survey items aren t suggestive or leading Priming can lead to false recollections quotMany families go on vacations to Disneyworld As a child how often did you go on vacation to Disneyworld 0 Survey Item Reliability 0 Operationalization l measurable variable 0 Religiosity l 1 Religious service attendance 2 Belief in supreme being 3 Donation to religious institution 0 Question Does a survey respondent rate these three different items in a similar way 0 Survey Measuring Religiosity 0 Good reliability l same survey respondent rates the following in similar ways 1 Religious service attendance 2 Belief in supreme being 3 Donation to religious institution October 13 2014 0 Survey Item Reliability 0 Operationalization gt Measureable Variable o Religiosity gt 1 religious service attendance 2 belief in supreme being 3 donation to religious institution 0 Question does a survey respondent rate these three different items in a similar way Are these items consistent Ask yourself that to measure reliability 0 Good reliability gt same survey respondent rates the following in similar ways 1 religious service attendance 2 belief in supreme being 3 donation to religious institution 0 Analyzing Reliability 0 Cronbach39s Alpha statistical test that determines internal consistency of items Excellent a gtor equal to 09 Good 07 ltor equal to a lt 09 Acceptable 06 ltor equal to a lt 07 Poor 05 ltor equal to a lt 06 Unacceptable a lt 05 0 Analyzing Survey Results 0 Parametric Statistics Assumptions 0 Types of Statistical Tests ttests Correlation Chisqua re Parametric Statistics Assumptions 0 Probability distribution 0 Middle Mean symmetrical 0 Standard Deviations Middle 341 and 341 Outer Middle 136 and 136 Outer outer middle 21 and 21 Outside 1 and 1 0 Sample mean u 0 As sample size increases the margin of error decreases TTest O O 0 Part of parametric statistics Used to test differences between two sample means Ex Surveyed attitudes about a product Question do men score differently than women 0 Assumptions 1 Data is normally distributed 2 Homogeneity of Variance both samples are of the same in variation If variance scores were different ex P lt 05 then this assumption is violated 3 Independence both sample means independent of one another 0 Example COO 0000 Independent variable type of notetaking Dependent variable interest in comstrat 309 Research question are there differences in comstrat 309 interest scores based on type of notetaking 1 normal distribution of dta 2 homogeneity of variance p 21 3 groups are independents note takers vs laptop note takers t263p01 0 Recap from last class 0 ttest between two independent means 0 Correlation relationship between one independent variable and one dependent variable 0 Multiple Regression relationship between multiple independent variables and one dependent variable 0 Chi Square difference between the expected frequencies and the observed frequencies in one or more categories 0 Correlation O 0000 00 Tests the relationship between two variables Positive correlation as X goes up so does Y on a graph Negative Correlation as X goes up Y goes down on a graph Data must be nominal The strength of a relationship is measured with Pearson39s r coefficient Pvalue determines signi cance Strong correlations has the points on the graph pretty much in a straight line and clustered together Weak correlations have the points scattered all over the graph shows no trend like a straight line Case study Attitudes about diet water affect people39s intentions to purchase diet water 0 Correlation versus Causation O O 0 Fact married men more likely to live longer than unmarried men Question does marriage positively in uence life expectancy Answer no men who get married are generally healthier and of higher SES 0 SURVEY ANALYSIS Confounding Bias 0 O 0 Is a relationship between X and Y causal Or is the effect due to some third unobserved factor spuriousness ln survey research you can control for possible quotthirdquot factors ex Prior health SES 0 Multiple Linear Regression O O O O O 0 Same as correlation analysis Allows for multiple independent variables instead of only one With multiple independent variables you can control for confounding factors Example Correlation being married is positively related to life expectancy But what happens when we include peoples39 1 health prior to marriage 2 socioeconomic status B beta standardized coefficient tells you how strong the relationship is between your X and Y variables RSquared the amount of variance explained by your independent variables in predicting dependent variable 0 Chi Squared Test 0 0 Only have categorical independent and dependent variables Example IV gender malefemale DV purchase energy drinks yesno Question Do men purchase more energy drinks than women 0 Chi Squared tests difference between the expected frequencies and the observed frequencies in one or more categories 0 Do the number of individual or objects that fall in each category differ signi cantly from the number you would expect Paper Critique Assignment Prospectus is due Nov 5th Content Analysis 0 Changes in Media Content 0 Has media become more violent o Concern in 80 s after WWII determine if there is more violent media o Is there greater representation of minority groups in media 0 Content Analys a systematic procedure to examine the content of recorded information data 0 Content Analysis the Foundation 0 Systematic content is analyzed using explicit and consistently applied rules 1 Sampling must follow explicit procedures for generalizability 2 Evaluation all content evaluated in same manner 0 Objective researcher s biases should not in uence ndings Analysis should yield same results for other researchers Ex For content analysis on violent media what you think is violent should also be viewed as violent by others Need high interrater reliability 0 Quantitative 1 provides a way of accurately representing a body of messages 2 allows for results to be summarized and reported succinctly Content Analysis Uses 0 1 Describing communication content Cataloguing characteristics of given body of communication content at one or more points in time Can identify developments of media content 0 2 Testing hypotheses Content analyses can be used for hypothesis testing Can examine relationships between different types of media content Ex Lowry and Naser 2010 examined relationship between winning political candidates and their political ads 0 3 Comparing media content to the quotreal worldquot Media portrayals assessed against quotreal lifequot Ex Does crime depictions on news mirror real life crime 0 4 Assessing the image of particular groups in society Representation of minority groups in popular media Ex 2003 study found that Latinos Asian Americans and Native Americans were rarely seen in local television newscasts o 5 Framing analysis Understanding the frequency of certain news frames A news story can be framed in different ways Obama and Ebola scare analyzed his ineptness on handling the threat Ex Poverty can be framed thematically or episodically Study show stories shown episodically more likely to perceive that poverty is individual issues 0 6 Starting point for media effects study Identi es trends in media content Show current media landscape Ex Can then lead one to do a study on media effects 0 Content analysis process 0 1 research question 0 2 population 0 3 sample 0 4 analyze sample via coding o 5 interpret results 0 Sampling Structure 0 1 Theoretical population who do you want to generalize to o 2 Study population what populations can you get access to o 3 Sampling frame How can you get access to them 0 4 The sample who or what is in your study 0 Random Sampling 0 Allows for greater generalizability 0 Random sampling allows for sample to more accurately represent its population 0 Computer programs can randomly select media from population 0 Determining Margin of Error 0 Margin of error the error in your sample s results o Represented by a range given a con dence interval 0 Example your sample has a margin of error of 3 This means that you are 95 con dent that your error is between 3 0 Larger the margin of error the less con dence that your sample s results re ect the population with which it was drawn 0 Reducing margin of error improves generalizability 0 Increasing sample size reduces the margin of error Coding 0 Now that we have a suitable sample its time to code 0 First we select a unit of analysis 0 Unit of analysis is the thing that you are coding 0 Ex words videos characters etc 0 Unit of Analysis Example A newspaper article A movie scene Newspaper headlines 0 Next we create coding categories 0 Codind Catedories Manifest Content straightforward information that is either objectively present or not present Latent Content subjective theme or information Manifest Code Example 0 O O 0 Content analysis of advertisement of people in ads of times product name appears Color of advertisement Latent Code Examples Content analysis of advertisements Depictions of good health Depictions of friendship Depictions of cultural gures OOOO Coding Recap o Coding Categories 1 Manifest content straightforward information that is either objectively present or not present 2 Latent content subjective themes or information Coding 0 Example Coding categories 1 Framing episodic framing thematic framing other 2 Depiction of violence Present versus not present 0 Review of Reliability 0 Latent content is inherently subjective o How do we ensure coded content is what we think it is 0 Ex Is something you code humorous perceived as humorous to others Answer we must do coder reliability 0 Interrater reliability 0 Agreement between coders o lntrarater reliability o Consistent individual response 0 Reliability Process in content analysis Interrater reliability when two or more independent coders evaluate the same content lntrarater reliability when one coder evaluates the same content at different periods 0 Measuring reliability Is there good agreement between coders Cohen s Kappa Cohen s Kappa is better than simple percent agreement scores It takes into account agreement occurring by chance 0 O O O O O The closer to 1 means its very reliable What s a good Cohen s Kappa Score 0 no agreement 002 slight agreement 02104 fair agreement 04106 moderate agreement O61O8 substantial agreement 0811 almost perfect agreement 0 Experiments continued 0 Experiments l great for internal validity Causality can be demonstrated why Elimination of selection bias Controlling temporal direction of variables Reducing likelihood of confounding variables 0 Method Comparisons 0 Cross sectional surveys Data taken at one point in time Can show correlations between variables Cannot determine causality 0 Experiments Participants exposed to a set of conditions Can show differences between conditions Can determine whether one condition is a causal agent 0 Independent Variables 0 These represent the thing that you are manipulating 0 They represent different conditions in your experimental design Example Do positive or negative images in news articles in uence article recall Dependent Variable 0 These are the things you are measuring following exposure to the independent variable 0 Types of Measurements o Selfreport surveys o Behavioral Virtual cutting down tree and using napkins 0 Biological 0 Types of Experimental Designs 0 One shot case study X gt O Something is made to happen then observe the results Example 0 Expose participants to an ad measure attitudes 0 However design limits causal claims 0 No control group to compare 0 One shot pretestposttest 01 X l 02 Observations made both before and after exposure to our independent variable Example First measure attitudes toward product Then expose participant to an ad Then we measure attitudes toward product again Any changes might be the result of the ad However other reasons could be at play People s attitudes might change without an ad overtime 00000 0 True Experiment X O Control O Or 01 X 02 01 control gt 02 Observations made in comparison with control conditions Example 0 Does watching crime shows in uence reallife perception of crime o If effect occurs we can determine precisely why that effect occurs 0 Comparison of experimental designs 0 One shot case study Low in internal validity o PretestPosttest Better internal validity than quotone shot case studyquot because looking at changes over time 0 True Experiment Best internal validity Factorial Designs 0 lnvolves manipulation of multiple independent variables 0 Examines main effects of each IV on the DV Does humor have a direct effect on an advertisements effectiveness 0 Also examines interactions between the multiple le on the DV Humor Present Not Present Music Present Not Present Possible Outcomes Main effects on humor I Humor affects the DV Main effects on music a Music affects the DV Interaction Effect a Only content with both humor and music affects the DV QuasiExperiment Designs 0 Sometimes key foundations of an experiment cannot be met 0 For example might be difficult to randomize participants Example WSU waste services observes changes in recycling behavior based on different signs One location they install new experimental signs another they leave the same Cannot randomly assign people to one condition or another Must then take into account this limitation 0 Experiments and Validity o Minimizing threats to validity Manipulation check How do we ensure construct validity for our independent variable Manipulation checks provide empirical evidence that our independent variables are label correctly Ex The effect of humorous versus non humorous ads Selective attrition Sometimes one experimental condition leads to participant attrition ie dropouts Experimental conditions that provide an undue burden on participants a How does quotWalking Deadquot in uence TV viewer s advertising recall a Walking Dead too scary participant dropouts n Leftover participants might be those who really like violent shows I Effects might not be due to the show but people who watch the show 0 Comparison of Experimental Designs 0 One Shot Case Study Lowest in internal validity o PreTestPostTest Better internal validity than quotone shot case studyquot 0 True Experiment Best internal validity o Factorial Designs Allows for multiple NS and their combined effects 0 QuasiExperiments Lacks one important feature of experiment typically randomization 0 Experiments and Validity o Minimizing threats to validity Manipulation Check provides evidence of construct validity Selective Attrition shows if one condition leads to greater participant dropouts than others 0 Independent vs Dependent Variables 0 Independent Variable IV Manipulated content You have control over what you showask the participants o Dependent Variable DV measured after exposure to IV 0 Examples of IV and DV 0 Does the type of food advertising affect nutrition quality of food purchases Independent Variables Healthy food ads Antiobesity ads Unhealthy food ads Mixed all 3 No ads Dependent Variables Calories Fat Sodium Carbs quotquotquotOf the items purchased for lunch Findings advertising for healthy foods antiobesity ads and mixed ads increased probability of selecting more healthy items and fewer unhealthy items from a lunch menu 0 Does eating popcorn in a theatre affect advertising effectiveness Independent Variable Eating popcorn Dependent Variable Advertising effectiveness Findings eating popcorn reduces the effectiveness of the advertisements 0 Experiments and Statistical Analysis 0 Statistics help us determine many things 1 Determine Threats to Validity Manipulation checks Selection attrition 2 Determine Causality Basic Statistics for Experiments 0 TTest Independent samples ttest Measures differences between two independent sample means Paired samples ttest Measures differences between two dependent sample means 0 Chi Squared 0 Analysis of Variance ANOVA Basic statistics for experiments 0 TTest Independent samples ttest Measures differences between two independent sample means X O Control O Paired samples ttest Measures differences between two dependent samples means 01 x mm 02 0 Independent Sample Ttest example 0 O O O 0 People randomly assigned to one of two conditions Average ad recall scores signi cantly higher in popcorn condition than control condition Popcorn recall score 23 Control recall score 35 P04 Paired samples Ttest example 0 O O 0 Does average attitudes toward candidate differ following exposure to campaign ad Pre score average 45 Post score average 32 P01 ChiSquared Test 0 O O Tests difference between the expected frequencies and the observed frequencies in one or more categories Do the number of individuals or objects that fall in each category differ signi cantly from the number you would expect Only have categoricalnominal independent and dependent variable Question Do people who watch a healthy food ad purchase a healthy lunch more than those who watch an unhealthy food ad lV healthy food ad versus unhealthy food as DV purchase healthy lunch yesno Can be used to detect selective attrition uneven dropout rates Question do your observed number of dropouts differ the expected number of dropouts 0 Review lecture slides for example 0 ANOVA Analysis of Variance o Ttest measure difference between TWO means 0 ANOVA means difference between 3 or more independent means Example 0 Can be used to test a manipulation check 0 Question Do our manipulations actually capture what we wanted to manipulate o Manipulation check results show evidence of construct validity Example 0 Dependent Variable Advertising effectiveness 0 Question Are there differences 0 Answer ANOVA results suggest Funny ads and Scary ads produce stronger recall scores than the neutral ads 0 Recap o Ttest differences between two means Independent samples differences between Paired sample differences within 0 Chisquared categoricalnominal data only 0 ANOVA differences between three or more independent means 0 Experiment Part 3 0 Comparison of experimental designs 0 One shot case study X O Lowest in internal validity o PretestPosttest 01 X 02 Better internal validity than quotone shot case studyquot 0 True Experiments Best internal validity o Factorial Designs Allows for multiple le and their combined effects 0 Quasiexperiments Lacks one important feature of experiment typically randomization Stats recap o Ttest difference between two means Independent samples differences between Paired samples differences within 0 Chisquared categoricalnominal data only 0 ANOVA differences between three or more independent means Qualitative Research 0 Research Paradigms o Empiricism Researchknowledge is generated through direct or indirect observation 0 Interpretive Understanding how individuals interrupt events and create meaning 0 Critical Examines power structures and their in uence on things Key Differences 0 Role of the researcher 0 Design 0 Setting 0 Measurement 0 Theory Building Purpose of Qualitative Research 0 Used by those who have an interpretive or critical paradigm 0 Goal is to develop rich understanding of peoples subjective expedences Qualitative Assumptions 0 Reality is socially constructed 0 Each person s experience is unique therefore cannot be generalized 0 Value is in understanding a situation deeply rather than being able to apply it to other situations quotTrustworthinessquot of Data 0 Qualitative research is NOT concerned with Reliability and validity of measurement Internal and external validity 0 Instead focus is on the researcher interpretation Credible study rings true quotCon rmablequot wellrounded conclusions Transferable ideas for reader 0 Types of Qualitative Methods 0 Participant observation 0 lndepth interviews 0 Case studies 0 Participant Observation 0 Observer no interaction 0 Observer as participant minimal interaction 0 Participant as observer often are acting as member of group 0 Participant fully acting as member of group 0 Participant Observation Process 0 Observations Setting Participants Events Acts Gestures Dialogue 0 Field Notes Descriptive Portray context Analytic Write during process of observation Expand upon later that day Autobiographical Your own behaviors amp emotions 0 Steps in Participant Observation Choose the research site Gain access Sampling Collecting data Analyzing data Exiting 0 Interviews 0 Interviews are an interaction 0 Types Structured ndepth Open Depth probing Need notes from 121 0 Focus Groups 0 Help examine people s perceptions opinions beliefs and attitudes toward a 0 Focus group structure 0 People involved Moderator ObserverNotetaker Participants 0 Protocol Structured questions Tend to last hour to hour and a half long 0 Focus Group General Flow 0 Welcome participants unstructured Have them eat get comfortable in the space 0 Introductions First names only moderator should note names 0 Begin Protocol Introconsent lntro questions TransitionKey questions Wrap UpSummary Anything else Dismiss 0 Room Set Up Comfort of participants is key Moderator should display leadership Note takers should be present but should be inconspicuous Role of the Moderator o Keeps group on track Encourages all group members to share equally Discourages domination of a single participant 0 Keep group members comfortable Need to be comfortable in front of small groups Dress like your subjects 0 Serves as a leader Types of Questions 0 Main research questions 0 Factual questions 0 Feel questions 0 Silence Special Considerations 0 Steer discussion toward their experiences The moderator should be aware that some individuals may be uncomfortable discussing certain issues 0 Seek out descriptions of why they feel the way they do 0 Have participants open up to one another Common Moderator Problems 0 Personal bias 0 Unconscious need to please the client 0 Need for consistency 0 Focus Groups and Boas 0 Social desirability response bias 0 Group norm bias 0 Focus Groups and Bias 0 Social Desirability Response Bias comes from society expectations 0 Group Norm Bias comes form group expectations 0 Role of Note TakerVerbal o Serves to help with transcription noting speaker changes 0 Note quotes that seem to stand out 0 Note general conclusions or themes 0 Good idea to assign two or more group members as note takers Role of Note Taker Physical 0 Same purpose as verbal note takers but more important if video recordings were not done 0 Note reactions of other participants Other s comments General nature 0 Focus Group Results Now What 0 Goals of Qualitative Research 0 lndepth understanding of a group of people s subjective expedences 0 Theory Development for academic research 0 Productmessage developmentfor industry research Focus Group Data 0 Focus group transcripts o Observational notes Creating a Transcript 0 Complete record for verbal comments 0 Should include ums ahs pauses grammar errors etc 0 Typically will not include body language gestures These need to be included in supplementary documents Identifying Themes 0 Become very familiar with the text you are analyzing 0 Start creating categories based on frequent comments sentiments conclusions 0 This process should continue to be re ned and evolve the more you look at the text Cut and Paste Technique 0 Read the transcript several times 0 Note important or recurring categories 0 Mark text which correspond to different categories Content Analysis o Is it possible to quantify information gained from focus groups Most interpretive scholars are horri ed contrary to purpose of qualitative research However yes Focus group data can be quanti ed and coded similar to content analysis Drawing Conclusions 0 Look at the these you identi ed 0 Look at comments of participants surrounding those themes o Is there a consensus or not 0 Where do people appear to be similardifferent 0 Writing the Report 0 Introduction 0 Method Method used and why Research Setting Participants backgrounds How data was collected 0 Findings lndicate general conclusions Provide detail and examples Make sure to highlight inconsistencies as well as consensus 0 Discussion Strengths Limitations Future Study
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'