RESEARCH DESIGN PSYC 3980
Popular in Course
Popular in Psychlogy
This 88 page Class Notes was uploaded by Spencer Smitham on Saturday September 12, 2015. The Class Notes belongs to PSYC 3980 at University of Georgia taught by Staff in Fall. Since its upload, it has received 10 views. For similar materials see /class/202469/psyc-3980-university-of-georgia in Psychlogy at University of Georgia.
Reviews for RESEARCH DESIGN
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: 09/12/15
Ch5 Collecting Descriptive Data 2122009 110100 AM I Types of descriptive research o Descriptive research 0 Systematically describes the characteristics of a given population o A Surveys 0 Used to gather info about Ps experiences attitudes or knowledge 0 1 Status surveys vs survey research Status surveys describe the characteristics of a population Survey research also tries to discover relationships among variables a Correlational not descriptive o 2 Methods of administration Interviews n Facetoface n Telephone Questionnaires n In person a Via mail a Via the internet o Quick and inexpensive o Little control over selection of sample o Ex surveying adolescents after divorce 0 3 Survey designs for descriptive and correlational research A Crosssectional design a Sample is cross section of the population a Can make comparisons between groups 9 how various groups differ in opinion a Difficult to argue for developmental trends by looking at Ps age B Successive indegendent salees design a Two or more samples of Ps answer the same questions at different points in time n Important to select samples in consistent manner n Ex monitoring quality of public schooling over span of years C Longitudinal design aka panel design a A single sample of Ps is questioned more than once a Can monitor change across time n Dropouts may be a problem B Demographic research 0 Describes patterns of basic life events and experiences 0 Eg birth amp death marriage amp divorce employment amp unemployment migration C Epidemiological research 0 Studies the occurrence of disease in different groups of people 0 Describes the prevalence prop With disorder and incidence rate of new cases of psychological disorders 0 Examines the behaviors and lifestyles associated with illnesses and injuries 0 Often conducted by medical researchers 0 Highlights problems that need attention II Sampling issues A Sampling the process by which a sample of Ps is selected from a population 0 Sample subset of population from which data are collected 0 Population a defined set of people animals or events eg Americans mice traumatic events 0 1 If your sample is representative of a population you can draw accurate unbiased estimates of the population s characteristics 0 2 Big questions To what groups do I want to generalize my findings Is it possible to attain a representative sample Is it worth the time money and effort B Probability sample a sample for which the researcher knows the probability that any individual in the population is included in the sample 0 Characteristics of sample should be the same as the characteristics of the population 0 Eg a sample selected from our 3980 class from UGA from GA residents registered voters o In an egsem design all cases in the population have an equal probability of being chosen for the sample Equal Probability Selection Method 0 Samgling error the extent to which characteristics of individuals in the sample differ from those of the population Results with sample results with whole population 0 When using a probability sample the error of estimation or margin of error can be determined 0 1 Margin of error the degree to which the data obtained from the sample are expected to deviate from the population Margin of error is a function of a Sample size population size variance of the data A sample will be more similar to the population when n The sample is la g o Economic samgle provides reasonably accurate estimate of population at reasonable effort and cost a The population size is smaller n The variance in the data is smaller 0 2 Basic methods of selecting a probability sample A Simgle random samgle every possible sample of the desired size has the same chance of being selected from the population a Requires a samgling frame a list of the population from which the sample is to be drawn a Eg class lists registered voters n More difficult teenagers seniors with arthritis B Stratified random samgling dividing the population into strata and randomly selecting Ps from each stratum n Stratum a subset of the population that shares a particular characteristic O O o Eg gender race location age a Ensures that researchers have an adequate number of Ps from each stratum n Progortionate samgling cases are sampled from each stratum in proportion to their prevalence in the population C Cluster samgling technique in which the population is first broken into groupsclusters and Ps are obtained from those clusters Typically clusters are based on naturally occurring groups Helpful when you don t have a sampling frame of all cases in the population Often involves multistage cluster samgling Divide population into large clusters and randomly sample clusters Then randomly sample smaller clusters within those large clusters Continue to sample from clusters until the appropriate number of Ps is chosen o Eg GA HS student s math performance 0 Counties 9 schools 9 classrooms 9 students 3 Nonresgonse groblem Failure to obtain responses from individuals that researchers select for their sample Respondents and nonrespondents may differ 4 Misgeneralization Occurs when a researcher generalizes the results to a population that differs from the one form which the sample was drawn n n n n o C Nonprobability sampling 0 O 0 Used when the probability that a particular case will be chosen for the sample is unknown Margin of error can t be calculated Most research involves this type of sampling o Is it a valid method Yes when the goal is to test hypotheses and not to describe the characteristics of a particular population 0 If you don t expect significant differences as a function of demographics eg age ethnicity it may be wasteful to seek a probability sample 0 Would a nonprobability sample suffice in testing Does frustration elicit aggression Does the presence of strangers hinder task performance What brain circuitry is associated with fear How do people form impressions o 1 Convenience sampling use whatever Ps are readily available Such samples vary in their diversity more diversity is better a Eg college students passersby friends Results may not generalize to desired population 0 2 Quota sampling try to fill a certain quota using a convenience sample Used to ensure that certain kinds of Ps are obtained in particular proportions n Eg choose equal of men and women young and elderly Ps Christians and nonChristians monolinguals and multilinguals n Eg fill ethnic quotas proportionate to US demographics o 3 Purposive sampling try to choose a sample of Ps that is typical of the population of interest Researchers Rs use their own potentially flawed judgment in deciding which Ps to include in the sample Examples a Rs choose one UGA class to extrapolate to all classes a Rs choose one college to extrapolate to all colleges a Rs look at voting patterns in one state and extrapolate to the whole country Ch5 Collecting Descriptive Data 2122009 110100 AM III Describing and Presenting Data o A Measures of Central Tendency 0 Mean average score 0 Median middle score of the distribution Less affected by extreme scores outliers Mode most frequent score Ex 11111223385 Mean 9 10 Median 9 15 Mode 9 1 o What would be the best measure of central tendency to use in the following scenarios Ex 13333588101215 a Mean no outliers Ex 2222333555666792 O O n Median Ex 11111122222333 n Mode a Political affiliation 1independent 2democratic 3rebupican o B Measures of Variability o Convey information about the speed of data Knowing the variability tells us how typical the mean is of the scores as a set If the variability is small then the mean is representative of the scores 0 Range Difference between the largest and smallest scores in a distribution 0 Variance Takes into account all the scores when calculating the variability Add the squared deviation scores an divide by n1 0 Standard deviation The square root of the variance generally easier to interpret O Ch5 Collecting Descriptive Data amp Ch13 Single Case Research 2122009 110100 AM III Describing and Presenting Data cont C Freguency distribution 0 O O O A table that summarizes raw data Displays the number of scores that fall within each of several categories 1 Simple freguency distribution indicates the number of Ps who obtained each score Scores are arranged lowest to highest Ex Ten people were asked how many close friends they have Friend 1 frequency 3 friend 2 frequency 5 friend 3 frequency 2 2 Grouped freguency distribution Compares samesized class intervals lowest to highest and frequency Used when there are many possible scores Shows frequency of a subset of scores Relative freguency proportion of the total number of scores that falls in each class interval Must have mutually exclusive categories that capture all possible responses Each category should be of equal size 3 Frequency histograms and polygons Responses or response category are on the xaxis and the number of responses is on the yaxis Presented in graphic form When lines are drawn to connect the frequencies you form a freguency polygon D Normal Distribution 0 O O A normal distribution rises to a rounded peak at its center and tapers off at both tails bell curve Most scores will fall toward the middle of the range of scores In a normal distribution about 68 of the scores fall in a range defined by or 1 standard deviation SD from the mean Skewed distributions Positive skew more low scores than high scores a Eg response time on word recognition task Negative skew more high scores than low scores a Eg exam scores o E Zscore O Describes a particular P s score relative to the rest of the data Indicates how far from the mean the P s score falls in terms of standard deviations Zscore score mean standard deviation Eg IQ scores mean100 SD15 n Score of 115 1 n Score of 70 2 Eg Response time mean1000ms SD250 n Score of 750 1 n Score of 1750 3 0 Useful in identifying extreme scores or outliers scores that are 3 SDs from the mean 0 O I Group vs IndividualBased Research o A Different research approaches have different aims o Nomothetic a roach establish general principles with broad generalizability Requires groupbased research 0 Idio ra hic a roach describe analyze and explain the behavior of individuals Focus not on the average person but looks at the behavior of individual people o B Different units of analysis 0 Experimental designs 9 groups 0 Singlecase designs 9 individual P More than one P may be studied but their responses are analyzed individualy 0 Cannot analyze these data with inferential statistics eg t tests and F tests o C Grounds for conducting singlecase research o 1 Error variance Group design argument n Averages provide a more accurate estimate of a variable s general effect a Allows us to estimate the amount of systematic and error variance in our data Singlecase argument n Error variance is created by averaging Ps in a group design interQarticiQant variance variability as a result of individual differences a Rs using groups designs ignore the real error variance within the P intraQarticiQant variance variability in P s behavior in the same situation a No interparticipant variance not an aggregate error variance is only intraparticipant o 2 Generalizability Group design argument n Averaging across Ps reduces the idiosyncratic responses of any one P to show the general effect a Provides an average P Singlecase argument n Averaging responses may not accurately describe any particular P s responses n An average P may not exist in reality o Eg variables with a bimodal distribution o Eg stepwise not gradual change 0 Learning curves 9 Ps are given puzzles that can be solved after figuring out a strategy or trick 0 The average P does not exist 0 3 Reliability Group design argument n Reliability of findings is established by replicating studies Singlecase argument n Reliability of findings should be established via o Intraparticipant replication replication the effects of the IV with a single P o Interparticipant replication seeing whether the effects obtained for one P generalize to other Ps II Singlecase Experimental Designs o A Basic types of singlecase experimental designs 0 1 ABA Design aka reversal designs Behavior is measures Baseline period A IV is introduced B Behavior is measured IV is removed A Behavior is measured Ex effects of teacher reinforcement on disruptive behavior of student in special education class 0 2 MultipleI designs singlecase experimental designs that tests between levels of an IV not just its presence or absence ABCDEFG design a A Baseline n B One level of the IV a C Another level of the IV a Eg does room temperature impact attention o A complete a stroop task at normal room temperature o B complete another stroop at 10 degrees cooler o C complete another stroop at another 10 degrees cooler ABACA design inserts a baseline period between each introduction of a level of the IV 0 3 Multiple baseline designs involve the measurement of two or more behaviors simultaneously Obtain baseline on all behaviors Introduce an IV that is supposed to affect only one behavior n Eg cold temperature should hinder attention but not analytical skills o Measure baseline attention amp analytical skills take P to cold room and do so again Allows the R to show that the IV is causing the target behavior to change and is not affecting the other behaviors o B Examining the Data 0 Results are typically shown in graphs R visually inspects the graph of the P to see if the IV had an effect Criticized for having no explicit criteria for deciding when an effect significant no inferential stats o C Uses of these designs 0 Operant conditioning eg schedules of reinforcement and punishment Psychophysiological processes eg effects of drugs Behavior modification techniques for changing problem behaviors based on operant conditioning o Demonstrational purposes o D Problems in using these designs 0 1 Effects are not necessarily generalizable Individual Ps may not be representative of the population at large 0 2 Difficult to assess interactions among variables Eg temperature x mood 9 attention maybe temperature only impacts attention when in a good mood o 3 Ethical issues Do you withdraw an effective treatment from a particularly troubled client in an ABA design 0 O Ch13 SingleCase Research amp Ch6 Correlational Research 2122009 110100 AM III Case Study Research o A Case study a detailed study of a single individual group or event 0 May use information from numerous sources Observation interviews questionnaires news reports and archival records 0 All information is compiled into a narrative description o B Uses of the Case Study Method 0 Source of insight and ideas Do celebrities more frequently suffer from bipolar disorder eg Vivian Leigh Carrie Fisher Britney Spears 0 Describe rare phenomena Who are serial killers o Psychobiography applying concepts and theories from psychology in an effort to understand famous people Why was Dr King an inspirational leader 0 Illustrative anecdotes o C Limitations of the Case Study Approach 0 Case studies provide little more than anecdotal support for a theory Failure to control extraneous variables a Alternative explanations cannot be ruled out Observer biases a All observations may be conducted by a single researcher a No way of determining reliability and validity of these observations I Correlational research o Describes the relationship between two or more naturally occurring variables 0 No manipulation of an IV 0 Do scores on the 2 variables covary 0 Examples As temperature increases does mood change Is age related to political conservatism II Correlation coefficient o Correlation coefficient a statistic that indicates the degree to which two variables are linearly related 0 A Pearson correlation coefficient r is the most commonly used measure of correlation Ranges from 100 to 100 0 B Sign direction of the relationship Positive correlation a Direct positive relationship between variables a As one variable increases the other increases a Eg SAT scores and GPA Negative correlation n Inverse negative relationship between variables a As one variable increases the other variable decreases n Eg class absences and course grade 0 C Numerical value magnitude or strength of the relationship When r00 the variables are not related rs of 78 gt 30 Magnitude is unrelated to the sign ofr n In terms of magnitude 80 80 80 gt 50 o D Graphic Representation of Correlations Scatter plot a graph of Ps scores on two variables a When there is a perfect correlation r 100 all data fall along a straight line a When r00 there is no M relationship between the two variables c There could be a curvilinear relationship between them 0 E Calculating Pearson s r One variable x values Another variable y values N sample size Z sum III Coefficient of Determination r2 o Difficult to compare correlation coefficients r o So we square rto obtain the coefficient of determination r2 o r2 the proportion of variance in one variable that is accounted for by the other variable 0 Example Correlation between children s and parents neuroticism scores is r 25 r2 25 25 0625 n 625 of the variance in children s neuroticism scores can be accounted for by their parent s scores 00 r200 shared variance0 r 40 r216 shared variance16 r 80 r264 shared variance64 r 100 r2100 shared variance100 IV Statistical Significance of r c When is a correlation coefficient statistically significant 0 When the r calculated on a sample has a very low probability of being zero in the population 0 Usually an r is significant if there is a less than 5 chance that such a correlation could have come from a population with a true r of zero Imagine that Extroversion amp IQ are truly unrelated r00 in the US Researchers want to determine if extroversionIQ relationship exists and for a sample of n150 150 Americans complete measure of extroversion amp IQ With this sample will the r00 Maybe Maybe not sources of error variance may produce a nonzero r Can we trust the r we get from our sample Not necessarily but we can calculate the probability that the r from our sample would be zero if we had tested the whole population I W O O If the probability is lt5 then our r is statistically significant 0 Statistical significance ofr is affected by 1 Magnitude of the correlation n The larger the r the more likely it will be significant 2 Sample size n More confident about rs from larger samples n Eg you find a r50 between age and liberal attitudes with n10 vs n100 n The smaller the sample the larger the r has to be to be statistically significant plt05 3 How careful you want to be not to draw an inaccurate conclusion about whether r00 n plt01 and plt001 are more stringent than plt05 Ch6 Correlational Research 2122009 110100 AM V Factors that Distort Correlation Coefficients rs o 1 Restricted range data in which Ps scores are confined to a narrow range of the possible scores on a measure 0 Artificially lowers rs below what they would be if the full range of scores were present o 2 Outliers scores that obviously deviate from the remainder of the data 0 They don t seem to belong in the data set 0 Scores that are 3 505 away from the mean are considered outliers 0 Online outliers fall in the same pattern as the rest of the data and tend to artificially inflate r o Offline outliers fall outside of the pattern of the rest of the data and tend to artificially deflate r o 3 Reliability of measures 0 The more unreliable a measure is the lower its correlation with other measures will be 0 Imagine that In the population the true correlation between neuroticism in children and in their parent r45 In your study you assess neuroticism with an unreliable scale The obtained rwill not be 45 but rather near 00 VI Correlation and Causality o A correlation between two variables does not imply that one causes the other Even when two variables have a perfect correlation we can t conclude that one variable causes the other variable o Ex Does having many friends make people happy 0 Imagine that happiness and friends r35 o Two alternative explanations in correlational findings 0 1 Thirdvariable problem Perhaps being extroverted causes both happiness and having friends 0 2 Directionality reversecausality problem Perhaps happy people are more likely to have many friends o Criteria for inferring causality o Covariation changes in one variable are associated with changes in other variable same as correlation Directionality the presumed causal variable preceded the presumed effect in time o Extraneous variables all other variables that may affect the relationship between the two target variables are controlled or eliminated o Correlational research satisfies the first and sometimes the second criterion but never the third VII Partial Correlation O o Partial correlation the correlation between two variables with the influence of one or more other variables statistically removed 0 What can you infer if You partial out the influence of a third variable and the correlation between two other variables drops a Correlation between the two variables is at least partly due to the third variable 0 Example If hunger amp math performance r50 Is some of this relationship due to mood n Hunger 9 bad mood 9 poor performance If you partial out the variance accounted for by mood and the partial correlation between hunger amp math is 20 a Yes some of the hungermath relationship can be explained by mood VIII Other Indices of Correlation o SQearman rankorder correlation 0 Correlation used when variables are measured on an ordinal scale Eg Are runner s height associated with the order in which they completed the race o Phi coefficient 0 Correlation used when both variables are dichotomous having two levels Eg Is gender associated with whether someone drops out of school o Pointbiserial correlation 0 Correlation used when only one variable is dichotomous Eg Is gender associated with IQ Ch7 Advanced Correlational Strategiesz122009 110100 AM I Regression Strategies o These techniques allow researchers to understand how and why sets of variables are related o Goal of regression analysis 0 To develop a regression equation that allows us to predict one score on the basis of 1 other scores o Regression provides a mathematical description of how the variables are related o A Linear Regression o What does a correlation indicate A linear relationship between 2 variables Ex Job performance rating vs test scores a This regression line shows the relationship between the 2 variables 0 The linear relationship between variables can be described with the equation for a straight line 39 Y 50 51X y the variable we would like to predict a Called the outcome variable dependent variable or criterion variable x the variable we are using to predict y n Called the predictor variable lo the yintercept of the line that best fits the data in the scatter plot a Also called the regression constant 31 the slope of the line that best represents the relationship between the predictor x and the outcome variable y n Also called the regression coefficient Example Does frustration predict aggression n Ps complete frustrating puzzles on paper and tear each up and throw them in the garbage n x frustration using a selfreport scale a y aggression of times papers are torn n If o is 5 and 31 is 3 n y 5 3X o If P reported frustration level of 2 how much aggression would he likely display 0 Answer 11 o If frustration is a 5 0 Answer 20 o B Multiple Regression Analysis 0 Multiple regression analyses use more than one predictor variable 0 1 Types of multiple regression A Simultaneous or standard multiple regression all of the predictor variables are entered into the regression analysis at the same time n The resulting equation provides a regression constant and separate regression coefficients for each predictor EX o Outcome life satisfaction o Predictors health social support amp income o Satisfaction 30 1health 2support 3income B Stepwise multiple regression builds the regression equation by entering the predictor variables one at a time n n Step 1 The predictor that most strongly predicts the outcome variable is entered into the equation Step 2 The predictor that accounts for the greatest amount of variance in the outcome variable above and beyond the variance accounted for by the predictor in Step 1 is entered o This variable may or may not have the second highest Pearson correlation with the outcome variable Continue stepbystep entering predictor variables in the order of their ability to uniquely predict the outcome variable n The procedure stops when o a either all of the predictors have been entered into the equation c or b none of the remaining predictors uniquely contributes to predicting variance in the outcome variable C Hierarchical multiple regression the predictor variables are entered in a predetermined order a As each new variable is entered the researcher tests whether the new variable uniquey contributes to variance in the outcome variable a Example o Outcome recovery from injury o Predictors general health age time in hospital amp optimism o If you want to know whether optimism predicts recovery above and beyond these other factors 0 Enter health age and time in hospital in one step 0 Enter optimism in a second step If optimism is a significant predictor what does that mean a Two common uses o Eliminating confounding variables o Testing mediational hypotheses An aside The difference between mediation and moderation n Mediation A 9 B 9 C o B is the causal mechanism that explains the relationship between A and C o Time on beach 9 exposure to UV rays 9 sunburn n Moderation D influences the A9 C relationship o Time on beach 9 sunburn o Moderator use of sunscreen n Moderation or Mediation o EX Stress 9 health moderation 0 Moderator social support o EX Stress 9 depression mediation 0 Mediator rumination 0 Stress 9 rumination 9 depression 0 2 Multiple correlation coeffcient R and R2 Multiple correlation coefficient R describes the degree of relationship between the outcome variable y and the set of predictor variables a R can range from 0 to 100 n The larger the R the better job the regression equation does of predicting the outcome variable from the predictors the proportion of variance in the outcome variable that can be accounted for by the set of predictors II CrossLagged Panel Design o In this design the correlation between two variables X and y is calculated at two different points in time o Correlate the scores on X at Time 1 with the scores on y at Time 2 Correlate the scores on y at Time 1 with the scores on X at Time 2 IfX causes y then the correlation between X at Time 1 and y at Time 2 should be larger than the correlation between y at Time 1 and X at Time 2 EX TV violence vs aggressiveness n Correlation between TV violence at Time 1 and aggressiveness at Time 2 31 gt correlation between aggressiveness at Time 1 and TV violence at Time 2 01 B These results support the hypothesis that watching violent TV increases later aggression O O Ch7 Advanced Correlational Strategiesz122009 110100 AM III Structural Equation Modeling SEM Structural eguation modeling or analysis tests whether the relationships observed among variables conform to a hypothetical model 0 When multiple measures of each construct are used 9 latent variable modeling Example type A personality 0 When single measures of constructs are used 9 path analysis Example smoking behavior in teen girls A researcher may predict how a set of variables are causally related 0 EgX9Y92vsY9X9Z This implies a particular patern of correlations o X amp Y would yield a higher r than X amp Z c This predicted pattern is compared to actual pattern of correlations o Fit index indicates how well the hypothesized model fits the observed data 0 If not the hypothesized model is not likely to be correct 0 By comparing fix indices for various models the researcher can determine which model best fits the data IV Multilevel Modeling o Used to analyze data sets with a nested structure 0 Examples Students may be nested within classrooms which are nested within schools Ps in a study may be nested in collaborative work groups 0 Problematic because most statistical techniques require independence of observations o Multilevel modeling separates the various influences that are operating at various levels of the nested data structure 0 For example it would allow us to examine the separate influences of Students personal capabilities features of the classroom and aspects of the school Ps personal attributes and the joint characteristics of the work group Ch8 Basic Issues in Experimental Research3262009 110900 AIV I Overview o What is an experiment 0 They go beyond description and correlation 0 Experiments allow us to examine the causes of behavior Does X cause Y o In a well designed experiment we 0 Manipulate an independent variable Randomly assign Ps to condition Control extraneous variables Of four types of research descriptive correlational experimental and quasiexperimental only experimental research provides conclusive evidence regarding causeand effect relationships II Experimental Manipulation o A IVs must have 2 or more levels 0 IV manipulated by researcher to assess their effects on Ps behavior levels different values or conditions 0 Examples Mood induction happysad qualitative Fear prime subliminal exposure to images of spiders or office supplies qualitative Exposure to noise 20dB40dB60dB quantitative 0 These levels can reflect either quantitative or qualitative difference o B Types of independent variables 0 Environmental manipulations experimental modifications of P s physical or social environment Ex temperature lighting number of people around Confederates accomplices of researcher who pose as bystanders 0 Instructional manipulations through verbal instructions that Ps receive Ex framing a task in different ways 0 Invasive manipulations creating physical changes in P s body through surgery or drugs Changes in person s body or physiological state 0 O O o C Experimental vs control groups 0 Experimental group Ps who receive a nonzero level of the IV 0 Control group Ps who receive a zero level of the IV or the absence of the variable of interest o D Making sure your manipulation is effective 0 1 Use pilot testing test on other people first 0 2 Use manipulation checks rating scale to determine whether IV was manipulated successfully if manipulating mood use mood measure III Other Variables o Subject variables 0 Subject or participant variable a personal characteristic of Ps that cannot be manipulated reflect existing characteristics of P Ex age gender region or origin presence of disorderillness or personality trait Subject variables are not true independent variables a They are a subset of quasiindependent variables o Dependent variables 0 De endent variable the response being measured in a study Typically a measure of Ps thoughts feelings behavior or physiological reactions Most experiments have several dependent variables IV Participant Assignment to Conditions o Goal is to manipulate only your IVs o The only difference between groups should be your IV 0 Betweensubjects design interest in differences in behavior between different groups of Ps o A In a betweensubjects design where each P is only in one condition researchers my use 0 1 Simple random assignment Ps are randomly places in conditions O o B Withinsub39ect or re eated measures desi Every P has an equal probability of being in any condition Experimenters may flip a coin or use a list of random numbers R can be confident that the groups are roughly equivalent at the beginning of the study 2 Matched random assignment Ps are first matched into homogenous blocks and then participants within each block are assigned randomly to conditions Rs obtain Ps scores on a measure known to be relevant to the outcome of the experiment a If you anticipate some subject variable to be important pretest it and take that into account when assigning participants to conditions a Ex pretest for age assign two oldest Ps to condition X and condition Y n This helps to ensure that the groups will be similar along some specific dimension n an experimental design while each P serves in all conditions of the experiment 0 O O O O 0 Interest in differences in behavior across conditions within a single group of Ps Each P is measured more than once Every P is tested at every level of the IV Ex does loud noise impact test performance Advantages Are more powerful than betweensubjects designs because every P is in every condition a Power the ability of a research design to detect the effect of the IV Require fewer Ps Disadvantages Order effects n n n The sequence of experimental conditions may impact your DV Types o Practice effects P s performance improves over time only because DV is completed several times o Fatigue Ps become tired or less enthused as experiment progresses o Sensitization suspicion Ps begin to realize what hypothesis is o Carryover effects when effect of particular condition persists after condition ends 0 Ex making sure first caffeine dosage wears off before giving second dosage Counterbalancing c To protect against order effects Rs use counterbalancing o Involves presenting the levels of the IVs in different orders to different Ps o Ex administering caffeine 0 Participant 1 0 mg 100 mg 300 mg 0 Participant 2 0 mg 100 mg 600 mg 0 Participant 3 0 mg 300 mg 100 mg Ch8 Basic Issues in Experimental Research3262009 110900 AIV Introduction o Describe the problem under investigation o Discuss relevant aspects of the existing research literature o Discuss the purpose and rationale of your research o Explicitly state you research questionhypothesis USE A CLEAR STRUCTURE 0 Outline your intro before you begin 0 Structure the intro like a funnel start broad and narrow until you get to your precise question 0 Use headings o Cite you sources in the intro and provide a list of references at the end 0 Update your references when you add your methods sections later No block quotations No reason to use quotations paraphrase O O V Experimental Control o Experimental control holding constant extraneous factors that might affect the outcome of the study o Main question in an experiment 0 Of the total variance in behavior X is there systematic variance due to your IV 0 Example People display varied amounts of aggression Can some of that variance be explained by the amount of frustration they are experiencing If frustration elicits aggression we should see systematic differences between the scores in the different experimental conditions in which frustration is manipulated o A Systematic variance portion of the total variance in Ps responses that reflects differences among the experimental groups 0 1 Treatment variance potion of the systematic variance that is due to the IV o 2 Confound variance portion of the systematic variance that is due to extraneous variables that differ systematically between the experimental groups aka confounds Confound variance must be eliminated by controlling for extraneous variables Ex if height pics of tall men vs short men 9 perceptions of attractiveness amp tall men were older than short men then age was confounded with height o B Eliminate confounds to improve the study s internal validity 0 Internal validit the degree to which a researcher draws accurate conclusions about the effects of the IV on the DV VI Threats to Validity o A Major confounds o 1 Biased assignment of Ps to conditions Nonequivalent groups not IV could impact DVs Can occur when random assignments fails Ex conscientious Ps end up in condition A but not B o 2 Differential attrition Attrition loss of Ps during a study Ps drop out of experimental conditions at different rates Thus the experimental groups are no longer equivalent Ex Ps in the mind numbingly dull condition drop out of study and Ps those in the engaging condition remain o 3 Pretest sensitization Completing a pretest leads Ps to react differently to the IV than they would have reacted had they not been pretested Ex pretest on racial prejudice before reading about Black or White job applications 0 4 History effects Extraneous events occurring outside of the research setting have an effect on Ps responses Ex before or after 911 an assault receipt of an award or admission into grad school 0 5 Instrumentation Any change in the calibration of the measurement instrument during the study may alter scores on the DV Ex using different observers to rate the behavior of Ps in different experimental groups 6 Regression to the mean Tendency for Ps who are selected because they have extreme scores to be less extreme in followup Ex select highlow socially anxious Ps in pretest in later session scores will be more moderate 7 Diffusion of treatment Change in the response of Ps in one condition because of their knowledge of the other conditions Especially problematic when Ps are in close proximity Ex Ps asked to make phonetic word judgments may hear other Ps instructions make semantic judgments 8 Sequence effects or carryover in withinsubjects designs Effects on a P s performance in later conditions due to the P s experience in previous conditions Ex writing about a sad experience may impact what P s later writing about a happy experience 9 Maturation in withinsubjects designs Ps go through agerelated changes Changes in Ps responses between Time 1 and Time 2 posttest are due to the passage of time rather than to the IV Ex aging fatigue hunger 10 Other design confounds Something other than the IV differs systematically between the experimental conditions Ex placement of an attitude object symmetry of an attitude object Ch8 Basic Issues in Experimental Research amp Ch9 Experimental Design 3262009 110900 AM VI Threats to Validity o A Major confounding variables o B Subject effects any changes in the behavior of Ps that are attributable to being in the study 0 1 Demand characteristics cues given to P5 on how they are expected to behave o 2 Placebo effect a physiological or psychological change that occurs as a result of the belief that an effect will occur Placebo control grouQ may be used in which some P5 are administered an ineffective treatment a placebo Ex degree of pain reduction real drug95 sugar pi50 none05 o C Experimenter effects any biasing effects due to the actions of the experimenter o 1 The experimenter s expectancies about the outcome of the study may bias the results expectancy effects Eg influencing P s behavior selecting data using statistical techniques and interpreting results in biased manner to support one s hypotheses o 2 Minimize the impact of subject and experimenter expectancies by using doubleblind procedures neither R nor P know which condition the P is in at the time of the study VII Sources of Error Variance o Error variance the portion of the total variance in Ps responses that remains unaccounted for after systematic variance is removed o Sources of error variance 0 A Individual differences preexisting differences between people 0 B Transient states at the time of the experiment Ps differ in how they feel 0 C Environmental factors differences in the environment in which the study is conducted D Differential treatment experimenters may not treat all of the Ps exactly the same 0 o E Measurement error unreliable measures increase error variance c Total variance treatment variance confound variance error variance 0 Systematic variance treatment variance confound variance 0 Unsystematic variance error variance VIII External Validity o External validity the degree to which the results of one study can be generalized to other samples settings and procedures o Tradeoff between internal amp external validity o A tightly controlled experiment has strong internal validity 0 But a high degree of experimental control lowers external validity I Oneway Designs o A Onewax experimental designs include only 1 IV that is manipulated o The simplest one includes two experimental groups ie 2 levels of your IV Eg if hypothesis is noise 9 anxiety manipulate noise quiet vs loud and measure anxiety 0 At least 2 levels of the IV are required but more are okay 2 noise quiet vs loud NOT 1 noise loud Also okay 3 noise none vs moderate vs loud o B Betweensubjects oneway designs 0 1 Using randomized group assignment Ps are assigned randomly to 1 of 2 conditions 0 2 Using matched subject assignment Ps are matched into on the basis of a variable then randomly assigned to conditions o C Withinsubjects oneway designs repeated measures 0 Each P serves in all experimental conditions make sure to counterbalance o D Pretestposttest designs 0 Previous examples are of Qosttest only designs DV is only measured after the experimental manipulation of the IV Eg does receiving a hug impact blood pressure 1 manipulate hugno hug 2 measure BP 0 Pretestlgosttest design DV is measured before and after the experimental manipulation Eg 1 measure BP 2 manipulate hugno hug 3 measure BP 0 1 Advantages of pretestposttest designs Can show that Ps in the various conditions did not differ on the DV at the beginning of the experiment a Eg compare pretest assessments of blood pressure Can show how much the IV changed Ps behavior from pretest to posttest n Eg measure change from pretest to posttest n Posttest only interpretation o Being hugged produced higher BP than not being hugged n Pretestposttest design allows this conclusion o Being hugged decreases BP whereas not receiving a hug does not impact BP More powerful than postonly designs in detecting the effects of the IV 0 2 Major disadvantages of pretestposttest designs Pretest sensitization administering the pretest may lead Ps to respond differently to the IV than they would have otherwise a Via suspicion demand characteristics practice effects Pretestposttest designs are not necessary a Posttest only designs provide enough information to determine whether the IV has an effect on the DV II Factorial Designs o Factorial design an experimental design in which 2 or more IVs are manipulated o A Independent variables IVs are referred to as factors 0 Eg in a study assessing the impact of noise and lighting on test performance Two factors IVs noise amp lighting o B Factorial nomenclature 0 Special terms are used to describe the size and structure of factorial designs 0 A 2x2 factorial read 2by2 is a design with two IVs each with two levels Eg does having another person present impact performance on simple or complex tasks a 2 an audience present vs absent x 2 task type simple vs complex betweensubjects factorial design 0 A 3x3 factorial has two IVs each with three levels Eg does having a friendstrangerno one present impact performance on math verbal or analytical tasks a 3 an audience friends vs strangers vs no one x 3 task type math vs verbal vs analytical betweensubjects factorial design Ch9 Experimental Design 3262009 110900 AM II Factorial Design o A Independent variables factors in a design o B Factorial nomenclature cont 0 A 2x2x4 factorial has three IVs two with two levels and one with four levels Eg does exposure to different musical genres at different volumes impact performance on simple or complex tasks a 2task type simple vs complex x 2volume soft vs loud x genre rock vs hip hop vs country vs opera betweensubjects factorial design 0 Tells you about The IVs manipulated factors The number of levels in your IVs The nature of the P assignment between or within subjects 0 Allows you to calculate the number of experimental groups needed How many cells are in the design In a 2x2 4 In a 2x2x4 16 0 You can then estimate the number of Ps needed for your study Using 15 Psgroup a ruleofthumb a 2x2 between subjects design 60 Ps o C Betweensubjects factorial designs 0 1 Randomized grougs factorial design Ps are assigned randomly to one combination of IVs o 2 Matched grougs factorial design Ps are matched on a relevant variable and randomly assigned to one combination of IVs o D Withinsubjects factorial designs repeated measures each P participates in every experimental condition 0 Eg do people recognize biologically prepared objects that are positive or negative more than those that are learned 2nature of object biologically prepared vs learned x 2valence positive or negative withinsubjects factorial designs o E Mixed factorial design Ps are randomly assigned to only one level of some IVs but receive every level of other IVs 0 Combination of betweensubjects and withinsubjects factors 0 Eg are individuals slower in categorizing positive or negative words when they are cognitively busy 2word valence positive vs negative x 2processing capacity no cognitive load vs cognitive load mixed factorial design with valence as a withinsubjects factor and processing capacity as a betweensubjects factor III Interpreting your Data o A Main effects 0 A main effect is the effect of one IV on a DV while ignoring the effects of other IVs A factorial design will have as many main effects as there are IVs Eg 2sleep deprivation 4 hrs vs 8 hrs x 2caffeine intake 0 mg vs 100 mg betweensubjects factorial design in which motor coordination is measured a How many main effects may result 2 o B Interactions 0 An interaction occurs when the effect of one IV differs across the levels of another IV If the A98 relationship is different under one level of variable C than another level of variable C an interaction is present Eg 2time studying 3 hrs vs 6 hrs x 2context alone vs with friends with test score as DV in betweensubjects design a Does studying for a short or long time differentially impact test performance as a function of the study context alone or with friends n Perhaps 6 hrs of studying produces better scores than 3 hrs only if the studying was done alone Perhaps time studied was irrelevant among those who did so with friends o C Using tables and figures 0 Data can be presenting in tables 0 It s easier to interpret data in a figure eg a line or bar graph An interaction is depicted with nonparae lines Ch9 Experimental Design amp Ch10 Analyzing Experimental Data 3262009 110900 AM III Interpreting your Data o A Main effects o B Interactions o C Using tables and figures o D Higherorder interactions interactions with more than 2 levels 0 3way designs a highorder design examine 3 main effects of IVs A B amp C 3 2way interactions n AxB interaction ignoring C n AxC interaction ignoring B n BxC interaction ignoring A 1 3way interaction of AxBxC n 2caffeine 0mg vs 100 mg x 2nightly sleep 4 hrs vs 8 hrs x 2nights deprived 1 vs 4 IV Combining Independent and Subject Variables o A Expericorr expcu39 39 orrelatinnal factorial desi ns experimental designs involving both manipulated IVs and measure subject variables ex sex age intelligence ability etc 0 NOT quasiexperimental designs 0 Ex what are the effects of a manipulated IV on the reactions of men and women a subject variable Allows researchers to Determine whether certain effects may generalize only P5 with particular characteristics Examine how personal characteristics relate to behavior under varying experimental conditions Reduce error variance by making the groups more homogeneous Subject variables may be discrete or continuous Discrete gender college graduate region of origin Continuous age income years of higher education personality traits o B Historically continuous subject variables have been split into discrete groups for analysis 0 Ways of splitting Ps into groups 0 O 1 Make into meaningful groups n 1 Ex income 9 above or below poverty line Ex years of higher education 9 college graduates 2 Mediansplit procedure R identifies the median of Ps scores on variable of interest a n 1 Ps scoring below the median are classified as LOW and all Ps scoring about the median are classified as HIGH Ex income 9 split into highs and lows Ex selfesteem scores 9 split into highs and lows 3 Extreme groups procedure a Preselect Ps who score extremely high or low on a particular subject variable Ex pretest for income 9 very poor or very affluent Ex pretest for neuroticism 9 extremely neurotic and nonneurotic o C Interpreting findings from expiricorr designs 0 Splitting continuous subject variables is problematic It may bias the results by missing effects that are actually present or obtaining effects that are not real Researchers lose important information by using this strategy a n Reduces power to find significant relationship between the subject variable and DV Ex Ps who are 18 23 28 and 35 may be young but they are not really truly equivalent in age 0 Rather than splitting Ps into groups based on a continuous variable Use multiple regression analyses that allow researchers to keep the subject variable continuous Ex 2mortality salience salient vs not salient x age betweensubjects design with patriotism as DV 1 Enter continuous subject variable age and the dummycoded IV 0mortality not salient group O 1mortaity salient group as predictors in a multiple regression analysis with patriotism as the outcome Be cautious in interpreting results from expiricorr designs If the IV affects the DV we can conclude that the IV caused this effect But we cannot infer causation with subject variables If a subject variable is involved in an interaction we say that it moderates Ps reactions rather than causing them a Moderator variable subject variable that moderates Ps reactions to IV a Ex selfesteem moderates individuals response to positive and negative feedback n Ex gender moderates the impact of alcohol on perceptions of dates I Rationale for using Inferential Statistics Do the responses of 2 or more experimental groups differ O O 0 If the IV has an effect on the DV the means for the experimental conditions should differ But error variance can also cause the means to differ So the condition means could be different even if the IV had no effect Then how do we know whether the difference in means is caused by the IV or by error variance 0 O Inferential statistics determine whether observed differences between the means of the conditions are greater than expected based on error variance alone Inferential statistics allow researchers to estimate how much the means should differ if the IV has no effect If the observed difference exceeds this amount then the IV may be having an effect We cannot be certain that the difference was caused by the IV but we can calculate the probability that the IV caused the means to differ II Hypothesis Testing o A Null and experimental hypotheses o Null hypothesis the IV did NOT have an impacteffect on the DV 0 Experimental hypothesis the IV DID have an effect on the DV 0 Although we are really interested in the experimental hypothesis inferential statistics test the null hypothesis 0 Researchers do NOT accept or reject the experimental hypothesis They Reject the null hypothesis H A researcher concludes that the null hypothesis is wrong and the IV did have an effect Fail to reject the null hypothesis H A researcher concludes that the null hypothesis is correct and that the IV did not have an effect We cannot accept the null because the null can never be proven There are four potential outcomes of a study a If null is actually false o Reject the null correct decision o Fail to reject the null Type II error a If null is actually true o Reject the null Type I error o Fail to reject the null correct decision o B Type I error a researcher rejects the null when it is actually true 0 You think there is an IV 9 DV effect but you are wrong 0 The difference in means is actually due to error variance 0 1 Alpha level the probability of making a Type I error Most commonly used at is 05 plt05 When 0c05 a researcher may reject the null when there is less than 5 chance that the difference in means is due to error variance If there is very little chance lt5 the difference is due to error variance we can be confident that our IV had an effect a We can sat the difference in means is statistically significant n The lower the alpha level the more confident we can be C Type 11 error a researcher fails to reject the null hypothesis when it is false 0 You conclude that your IV does not influence you DV but it really does 0 1 Beta level the probability of making a Type II error Error variance within groups contributes to Type II errors a Ex unreliable measures data entry errors poor control Might have too few Ps in each cell to have enough betweengroups variance To avoid Type II errors researchers try to improve the study s power 0 2 Power the probability that a study will reject the null when it is false and thus detect effects that actually occur Power 1 3 Power increases with the number of Ps in the study Power analysis is used to decide how many Ps are needed to detect a significant effect a Researchers should conduct studies with sufficient power so that they are able to reject the null if in fact there IV has an effect in their DV If you are looking for a small effect sadnessetest performance you ll need more Ps to have sufficient power than if you are looking for a large effect gendereheight What defines power n at of Ps and effect size o D Effect size the proportion of variability in the DV that is due to the IV 0 0 Range 00 to 100 Ex if the effect size 32 then 32 of the observed variability in the DV is due to the IV Common measures Cohen s d etasquared omegasquared Ch10 Analyzing Experimental Data 3262009 110900 AM II Hypothesis Testing o E Direction of hypotheses o Directional h otheses state which of the two condition means is expected to be larger Use a onetailed test more liberal Ex rates of recovery will be higher among Ps exposed to a pet dog than those not exposed to a pet dog 0 Nondirectional h otheses state that the two means are expected to differ but does not specify which will be larger Use a twotailed test less liberal Ex rates of recovery will be influenced by exposure or lack of exposure to a pet dog III The t test o ttests are used to test the difference between means derived from two independent groups o A At a conceptual level 0 Estimate how much the means of the conditions would differ due to error variance alone Calculate the observed difference between then means Compare observed difference with the estimated difference that could be due to error variance Observed difference gt estimated difference due to error variance 9 reject the null hypothesis o B 5 steps in conducting a t test an overview 0 Step 1 calculate the means of the two experimental conditions 0 Step 2 calculate the standard error of the difference between the two means Step 3 find the calculated value oft Step 4 find the critical value of 139 from a table of t values Step 5 comparing the calculated value of t to the critical value oft If the calculated value oft gt critical value oft 9 reject the null IV The Paired t test for Matchedsubjects and Withinsubjects Designs o Used when the experiment involves O O O O O 0 Between subjects with matchedsubjects assignment 0 Withinsubjects design Takes advantage of reduced error variance and provides a more powerful test Ch11 Analyzing Complex Designs 3262009 110900 AM Research Proposals o Method sections due Tuesday April 21st 0 Turn in Method and References including sources from the Introduction o Anticipated Results sections due Tuesday April 28th 0 Turn in everything 0 Title page Introduction Method Results References I Rationale for the Use of More Complex Tests of Experimental Data o A When the IV has more than two levels a t test is inappropriate 0 Why 9 inflation of Type I error When one ttest is conducted what is the probability of making a Type I error 5 Two t tests 25 10 After 20 tests you d expect one to yield a significant effect plt05 that is due to chance o B Minimize Type I error by using more complex tests or adjustments 0 Bonferroni ad39ustment divide the desired alphalevel eg 0c05 by the number of statistical tests that will be conducted The overall likelihood of making a Type I error across all tests is 05 Ex if you conduct 5 t tests use 0c01 Drawback for using this adjustment a By decreasing the alpha level for each test the probability of making a Type II error increases o C Analysis of variance ANOVA a statistical procedure used to analyze data from designs that involve more than two conditions 0 Analyzes the differences between all condition means simultaneously 0 Holds the alpha level at 05 regardless of the number of means being tested Different types of ANOVAs for different study designs 0 Betweensubjects oneway ANOVAs ANOVAs for factorials Withinsubjects repeated measures ANOVAs Mixedfactorial mixedfactorial ANOVAs Multiple DVs MANOVAs multivariance ANOVA Designs including covariate ANCOVA II Using ANOVA with Data from Oneway Designs betweensubjects o A An overview 0 ANOVA uses a statistic called the F test o Ftest the ratio of the variance betweengroups eg levels of an IV to the variance withingroups Variance between groups is systematic variance attributed to the IV and confounds n Ex difference between mean anxiety of Ps in the cat condition dog condition and no pet condition Variance within groups is error variance n Ex difference in anxiety level within conditions eg among those exposed to a pet o The larger this ratio the larger the calculated value of F and the less likely that the differences among the means are due to error variance o B Components 0 1 Computing total variance What is the total variance in Ps responses a Total variance systematic variance error variance Calculate the total sum of sguares n 1 subtract mean from each score 2 square these differences 3 sum them ANOVA partitions the total sum of squares into n Sums of sguares betweengroups is systematic variance that reflects differences between the experimental group due to IV a Sums of sguares withingroups reflects error variance n Conceptually 39 SStotal SSbetweengroups SSwithingroups o 2 Computing variance withingroups Sums of sguares withingrougs SSWg the sum of the sums of squares for each condition a 1 calculate sum of squares for each group 2 sum them a Because the variance within each group is error variance the sum of the variances within each group is the total error variance in the data Withingrougs degrees of freedom dfwg N k total Ps of groups Mean sguare withingrougs MSWg an estimate of the average withingroups variance n MSWg SSWg dfWg o 3 Computing variance betweengroups Sums of sguares betweengrougs SSbg the variance due to the IV or confounds n Think of it as SS at the group level o 1 Calculate the grand mean the mean of group means 0 Ex no pet M5 cat M4 dog M3 grand M4 o 2 Subtract the grand mean from each of the group means 0 Ex 541 440 341 o 3 Square the differences 0 Ex 1 0 1 o 4 Multiply each squared difference by the size of the group 0 Ex If n10 in each condition 10 0 10 o 5 Sum across groups 0 ex 10O1020 SS20 Betweengrougs degrees of freedom dfbg dfbg k 1 or of groups 1 n Ex 3 groups cat dog no pet 312 Mean sguare betweengroups MSbg reflects the systematic variance between the groups 393 MSbg SSbg dfbg n Ex 202 10 MSbg 10 o 4 Conducting the Ftest F MSbg MSWg Compare calculated Fto critical F n Must know the alpha level and df If calculated F gt critical F then we conclude that at least one of the means differs significantly from the others III Using ANOVA with Data o A ANOVAs can be used when you have multiple IVs o B Instead of having one SSbg you include SS for each individual group 0 Ex in a twoway factorial AxB the total variance is composed of four parts 1 Main effect ofA 2 Main effect of B 3 AxB interaction 4 Error variance 0 Conceptually I SStot SSA SSB SSAXB SSwithingroups IV Followup Tests o If an Ftest is significant we don t know which of the means differ 0 Ex no pet M5 cat M4 dog M3 o Followup tests are needed to determine which means differ significantly o A Further examination of main effects 0 Post hoc tests aka multiple comparisons are used to determine which means differ significantly Common tests Tukey s test Scheffe s test and NewmanKeuls test If the Ftest is not significant followup tests are not conducted o B Further examination of an interaction o If the Ftest shows that an interaction is significant tests of simple main effects are conducted 0 A simple main effect is an effect of one IV at a particular level of another IV 0 For a twoway interaction of AxB 1 Simple main effect ofA at Bl noise when dog present 2 Simple main effect ofA at B2 noise when dog is not present 3 Simple main effect of B at A1 dog presence when quiet 4 Simple main effect of B at A2 dog presence when loud Ch11 Analyzing Complex Designs amp Ch12 QuasiExperimental Research 3262009 110900 AM V Using ANOVA with Data from Withinsubjects or Mixed Factorial Designs o Using withinsubjects ANOVA aka repeated measures ANOVA when IV is manipulated withinsubects o Eg attention to images 9 3neutral vs disgusting vs pleasant o Eg perceptions of people on the basis of their handwriting 9 2legibility legible vs illegible x 2style print vs cursive fonts 0 Eg recognition of voices in various contexts 9 2context noisy office vs noisy kitchen x 2voice stranger vs family member o Withinsubjects ANOVA o Reduces the estimate of the error variance increases power and makes it easier to detect significant effects 0 Can also be run with a mixed factorial VI Using MANOVA with Multi Related DVs o MANOVA multivariate analysis of variance tests the differences between the means of 2 conditions on 2 DVs o Eg does sun exposure to a pet cat dog none reduce anxiety self report blood pressure 0 Eg does music rap country none impact cognitive performance math verbal spatial o MANOVA is used when the DVs are conceptually related to one another 0 The researcher expects the IV to impact DVs similarly o If not the researcher would not run a MANOVA o Conducting multiple ANOVAs will increase the risk of a Type I error VII Using ANCOVA with a Covariate o ANCOVA analysis of covariance allows researchers to statistically control for extraneous variables 0 Covariate variable that is correlated with the DV 0 This lets you account for variance not associated with IVs Thereby reducing error variance 0 Ex 3treatment placebo vs drug 1 vs drug 2 on recovery from injury Control for the severity of the injury by measuring severity and entering it in an ANCOVA as a covariate 0 Used when researchers are not interested in the impact of the covariate itself but simply wants to control for its impact VIII Using ANOVAs with Nonexperimental Data o Nonexperimental data derived from a quasiIV can be entered into an ANOVA o Eg gender male vs female region of origin GA vs out ofstate 0 Main effects and interactions involving quasiIV can be assessed Eg gender manipulated IV and gender x IV 0 Typically only discrete variables with few levels are entered into an ANOVA Continuous variables eg age can be split and entered into an ANOVA bad idea Better idea 9 continuous variables can be entered in a multiple regression as a predictor along with other IVs o Data analysis options 0 Ex mortality saliencewalking by a funeral home or not x agecontinuous 9 patriotism 1 2mortality salient vs not x 2age lt35 vs gt35 betweensubjects ANOVA 2 Multiple regressions with mortality salience dummy coded 0not salient 1salient and age as predictors with patriotism as an outcome I Overview o A Quasiexperimental designs involve 0 Comparisons between group means 0 Group assignment is not manipulated Often used when variable of interest cannot be ethically manipulated Lack of control over extraneous variables o B Quasiindependent variables Not a true IV May be subject variables eg gender having a disorder May be an event of condition that naturally occurred eg the introduction of a new law o C Issues of internal validity 0 Without control over the IV and random assignment the internal validity is questionable o Quasiexperimental designs are more internally valid if researchers Assess and statistically control relevant variables Match treatment and control group as much as possible a Eg looking at new seatbelt law in Oregon Compare with similar state II Pretest Posttest Designs o A Onegroup Qretest Qosttest design 0 1 Measure behavior 0 2 Enact treatment or change 0 3 Measure behavior again Eg measure mood turn on music measure mood again a 01 X 02 Changing in behavior are attributed to n Maturation normal changes over time n Regression to the mean extreme scores become moderate over time a History effects an even cooccurs with the quasiIV Because these designs lack sufficient control and fail to eliminate most threats to internal validity you should never use them 0 Common threats to internal validity How could maturation explain the impact of an SAT prep course on SAT performance a Take SAT take prep course take SAT again o Could be affected by other classes 0 O O o Add control group Regression n Tired red bull tired n Go back to original level of tired History effects a Weight shortest day of the year weight n Shortest day is near holiday Christmas o B Nonequivalent control group design 0 Researcher obtains one or more groups of Ps who are similar one group receives the quasiIV o 1 Noneguivalent groups posttest only design measure both groups after one receives the quasiIV Ex select two states 1 state with new speeding law measure speeding in both states a QE group X 0 NEC group 0 Cannot be sure the groups were the same before treatment 0 2 Noneguivalent groups pretest posttest design both groups are measured before and after the quasiIV Ex assess speeding in 2 states 1 state institutes new speeding law assess speeding in both states a QE group 01 X 02 NEC group 01 02 0 Potential threat to internal validity Local history effect something else may happen to one group that does not happen to the other a Ex large deadly crash makes headlines in one state III Time Series Designs measure the DV of several occasions before and after the quasiIV occurs o A Single interrupted time series design 0 01 02 03 04X05 06 07 08 0 Taking several pretest measures before introducing IV and taking several posttest measures 0 Should be able to tell whether or not an effect is due to the quasiIV as opposed to aging or maturation Ex introducing drug education program 0 Possible threat to internal validity Contemporary historl possibility that the observed effects are due to some other event that occurred at the same time as the quasiIV o B Interrupted time series with a reversal o 01 02 03 04 X 05 06 07 08 i 09 010 011 012 0 Shows the effect of the quasiIV on the target behavior and what happens when the quasiIV is removed 0 Helps rule out historical influences o C Interrupted time series design with multiple replication 0 01 02 03 X 04 05 06 i 07 08 09 x 010 011 012 i 013 014 015 0 Further allows researchers to rule out history effects o D Control group interrupted time series design Ch12 QuasiExperimental Research 3262009 110900 AM Research Proposal o Anticipated results due Tuesday April 28th 0 Turn in everything 0 Title page introduction methods and results III Time Series Designs o A Simple interrupted time series design o B Interrupted time series with a reversal o C Interrupted time series design with multiple replications 0 Limitations of timeseries designs that include reversals Le a return to baseline Ability to remove quasiIV n Eg repeal a new law Effects of the quasiIV may carry over after reversal n Eg realizing how effective a new management style is managers may continue to use it despite instructions to revert back to their old style Removal of the quasiIV can produce a change a Eg if a candyreinforcement program was stopped students might become angry that they no longer receive a reward for participation o D Control group interrupted time series design 0 QE group 01 02 03 04 X 05 06 07 08 0 NEC group 01 02 03 04 05 06 07 08 0 Includes a nonequivalent control group that does not receive the quasiIV o Helps rule out some history effects If both groups are exposed to the same events and a change is seen in only the treatment group the quasi IV likely had an effect Cannot rule out local history effects events specific to one group IV Comparative Time Series Design o Comparative time series design 0 Follow 2 variables over time in order to examine how changes in one are related to changes in the other Eg economic uncertainty and attitudes about immigration Eg TV watching and aggression Eg length of days depression Provides indirect evidence that the change in one variable may be causing the change in the other Cannot make causal conclusions V Longitudinal Designs o A In comparison to crosssectional designs 0 O In longitudinal designs the quasiIV is time Nothing has occurred from one observation to the next except for the passage of time 01 02 03 04 05 Mostly used by developmental psychologists to study agerelated changes in cognition affect and behavior Ex how strategies children use to remember things change as they get older Crosssectional designs compare groups of different ages at a single point in time Drawbacks of longitudinal designs Difficult to obtain Ps who agree to be in a study over a long period of time Difficult to keep track of Ps Repeatedly testing a sample requires time effort and money Advantages of longitudinal designs Allow us to distinguish agerelated effects from generational effects a Ex measure age and risktaking in a cross sectional study a If the variables are negatively related there could be an agerelated effect or a generational effect a Better measure risktaking over time longitudinal study Allow us to observe how individuals change with age VI Program Evaluation o Program evaluation using behavioral research methods to assess the effects of behavioral interventions or programs 0 Used to inform decision makers eg government administrators legislators school boards and company executives VII Evaluating QuasiExperimental Designs o Can we infer causality based on a quasiexperimental design 0 These designs allow researchers to demonstrate that The presumed causal variable precedes the effect in time The cause and effect covary 0 These designs do not Eliminate all other alternative explanation of the results through randomization and experimental control o Researchers can be more confident of their results if they 0 Measure both the effects of the quasiIV and the processes that are assumed to mediate the relationship Eg gender 9 help in a crisis potential mediator perceived ability to enact help in the situation Eg child of divorce 9 selfesteem potential mediator time spent with supportive parents 0 Employ multiple approaches that may yield convergent evidence critical multinism 0 Consider the potential threats to internal validity History effects impact all Ps Maturation Regression to the mean Pretest sensitization Contamination Selection bias normally ruled out via random assignment Local history effects impact Ps in one group Ch1 Research Design Overview 1132009 105900 AM I Drawing Conclusions about Behavior o How do we come to understand the world 0 Intuition 0 Personal philosophy or religion 0 Logic Basic research 9 increasing knowledge Applied research 9 finding solutions 0 Scientific approach II The Scientific Approach o Three criteria 0 A Systematic empiricism 9 systematic instead of casual observations practice of relying on observation to draw conclusions about the world 0 B Public verification 9 retesting of hypothesis findings of one researcher can be observed replicated and verified by others 0 C Solvable problems 9 answerable given current knowledge and research techniques Pseudoscience 9 violates basic criteria of science III The Science of Psychology o Although biology chemistry physics and psychology differ in content they all use a scientific approach o Psychology is the scientific study of behavior and mental processes 0 Wilhelm Wundt was the founder of a scientific psychology 1870s 9 first science laboratory 0 Isn t it just common sense Surprising findings conflicting axioms hindsight bias IV Goals of Psychological Research o A Description 0 Common methods Surveys observational studies 0 Examples What type of cereal certain people buy Voting preferences Changes in behavior across the lifespan Patterns of aggression among chimpanzees o B Prediction 0 If X condition is present Y will happen 0 Examples Who will do well in a particularjob What personalities put people at risk for criminal behavior Factors that predict happiness o C Explanation 0 Looking beyond what happened to see why it happened causation 0 Examples Why do depressed individuals respond more to one treatment than another Why are group with charismatic leaders especially susceptible to groupthink Why are Southerners more likely to lash out after a transgression than Northerners V Theories and Models o Theory 0 A set of propositions that attempts to specify the interrelationships among a set of concepts May be on a grand or small scale M0de o Attempts to describe how concepts are related but not why VI Formulating Hypotheses o A Hypothesis 0 A proposition that follows logically from a theory a prediction All hypotheses should be made a priori made before testing Beware of post hoc explanations conclusions made after testing o B Deduction 0 Process of reasoning from a general proposition to specific implications of that proposition general to specific 0 Theory 9 hypothesis o C Induction O O O O O Abstracting a hypothesis form a collection of facts specific to general Data 9 hypothesis Empirical generalization 9 previously observed patterns of results VII Testing Hypothesis Powerfully o A Methodological pluralism 0 Using a variety of different methods and designs to test a theory Gives us greater confidence than findings that are based on a single method o B Strategy of strong inference 0 Designing studies to pit two or more opposing predictions against one another 0 Data will confirm one theory while disconfirming the other VIII Definitions o A Conceptual definition 0 Much like a dictionary definition Sometimes too fuzzy for precise scientific communication o B Operational definition 0 Specifies precisely how a concept is measured or manipulated in a particular study o Ex Is balding related to selfesteem 0 Conceptual 9 balding loss of hair selfesteem view of oneself 0 Operational 9 balding scale of hair selfesteem questionnaire o Provide operational definitions for O O O O O O H 9159 9 Happiness surveys measure of smiling Aggression measure heart rate in stress Attention eye contact Obesity measuring calorie intake Patience situation to strain patience Anxiety observation heart rate IX Proof and Disproof in Science o A Logical impossibility of proof 0 Confirming a hypothesis with empirical findings does not logically indicate that the theory from which the hypothesis is derived is correct Ex Tiny green men living under the skin 0 You don t prove a theory you provide support for it o B Practical impossibility of disproof o Failing to find empirical support for a hypothesis does not necessarily imply that the theory is incorrect 0 Many things can lead to a failure to obtain supportive data 0 Ex Aspirin for headaches o C How does science progress 0 Theories gain merit with accumulated supporting evidence from several studies 0 All ideas pass through a scientific filter if they are to become widely accepted by the field All ideas 9 initial research projects 9 research programs 9 published research 9 secondary research literature established knowledge 0 Null findings 9 results showing certain variables not related to behavior X Strategies of Behavioral Research o A Descriptive research describes the behaviors thoughts or feelings of a particular group of individuals 0 Foundation for other research 0 Examples Public opinion polls 9 describe attitudes of particular group of people Developmental changes in behavior over age Marketing studies of consumer preferences Incidence of particular mental disorder o B Correlational research investigates the relationship between two or more variables 0 Correlation does not im l causation How not why are concepts related 0 Examples 0 Is selfesteem related to how shy people are Is the ease with which people can be hypnotized related to their conformity in social situations Is an individual s mood related to their task performance o C Experimental research can determine whether certain variables cause changes in behavior thought or emotion 0 Experiments involve Manipulation of at 1 independent variable IV Measurement of causal effects on a dependent variable DV Control of extraneous influences 0 Example Manipulate mood measure task performance control for extraneous factors 0 Term experiment 9 researchers control independent variable to assess its effects on behavior cannot be used as synonym for research or study o D Quasiexperimental research examines the effects of naturally occurring events 0 Used when unable to manipulate the IV or control extraneous factors 0 Examples Do women prefer sugary foods more than men n Men amp women are given the opportunity to choose a sweet or salty snack Do teenagers have better memories than the elderly a Teens and senior citizens complete a recall task Ch2 Behavioral Variability 1132009 105900 AM I Variability and Research o A Behavior variability the degree to which scores in a dataset differ or vary from one another o B Five proposition that involve the relationship between variability and research 0 1 Psychology involves the study of behavioral variability Behavior varies across situations among individuals and over timedevelopment Psychologists try to understand how and way behavior varies o 2 Psychological research questions are about behavioral variability Examples a To some extent dodoes SES explain variability in voting behavior Situational factors explain variability in memory recall Gender explain variability mate preferences 0 3 Research should be designed to best test questions about behavioral variability We want to unambiguously identify constructs that explain behavioral variability a Measure constructs validly and reliably n Cleanly manipulate a construct a Take other factors into account 0 4 The measurement of behavior involves the assessment of behavioral variability We quantify behavior measure behavior by assigning numbers to behaviors n Intelligence IQ test a Persistence seconds spent holding a handgrip n Embarrassment degree of blushing So the variability in the numbers should reflect the variability in the behavior a Great variation in intelligence wide range of IQ scores 0 5 Statistical analyses are used to describe and account for the observed variability in the data How much variability is there in our data What is the variability related to What cased the variability Two types of statistics n A Descriptive Statistics used to summarize and describe the behavior of participants in a study o Reducing large number of scores or observations to interpretable numbers such as averages and percentages o Eg What is the average score What is the variance n B Inferential Statistics used to draw conclusions about the reliability and generalizability of one s findings o Used to help answer questions o Eg How likely are my findings due to random extraneous factors rather than my manipulation II Assessing Variance o A Variance statistic used to indicate the amount of variability in participant s P responses 0 Q To what extent do you favor gun control laws 1 not at all 5 very much Participant Score a Lauren 4 n Sophia 1 n Dorothy 2 a Rose 2 n Heidi 4 n Blanche 3 o B Range the difference between the highest and the lowest score 0 The range for these scores is 41 3 0 Two sets of data could have the same range but vary significantly 1333333335 vs 1122334455 Not much variability vs a lot of variability 0 Range is insufficient because it only takes the highest amp lowest scores into account 0 Variance s2 is a statistic that uses all the scores in a dataset Variance is assessed by seeing how much the scores vary around the mean u If the scores are tightly clustered around the mean then the variance of the data will be small If the scores are more spread out from the mean then the variance will be larger C Statistical Explanation of Variance 0 Five steps for calculating the variance 1 Calculate the mean the average a 4122436267 2 Calculate a deviation score how much each score differs from the mean n n Positive deviation scores 9 Lauren 4267 133 Negative deviation scores 9 Dorothy 2267 067 3 Square each deviation score a n n n n n n This eliminates negative values Lauren 177 Sophia 279 Dorothy 045 Rose 045 Heidi 177 Blanche 011 4 Calculate the total sums of squares the sum of the squared deviations of the scores of the mean u 177 279 045 045 177 011 734 5 Divide the Total Sums of Squares by n1 n n the number of participants in your sample n 734 61 147 n The variance for this data set is 147 Ch2 Behavioral Variability cont 1132009 105900 AM III Total Variance Systematic Variance Error Variance o A Systematic variance the portion of the total variability in P s behavior that is related in an orderly predictable fashion to variables under investigation 0 The search for systematic variance Is variability in one variable systematically related to the variability in another variable Researchers try to account for or explain the variability they observe In a correlational study predictor error outcome In an experiment IV error DV Examples bus schedule responses of romantic partner variation in voting behavior o B Error variance the portion of the total variance in P s behavior that is unrelated to the variables under investigation in the study 0 The remaining variance that is not accounted for 0 Due to Poor measurement P inattentiveness Other factors 0 Figure amount of social support is positively related to physical recovery o Distinguishing Systematic from Error Variance 0 Statistical analyses are used to partition the total variance into systematic and error components 0 The more error variance or noise the more difficult it is to determine whether the variables of interest are related to variability in behavior Minimize error variance as much as possible in order to detect the systematic variance in the data a Minimize noise by controlling or holding constant other factors operationalizing variables well etc a If you control for P gender type of injury optimism overall health etc you ve reduced the noise in your study and can better reveal an association between social support and recovery IV Effect Size o Effect size is a measure of strength of association the strength of the relationship between two variables 0 Ex Pearson s r Cohen s d Hedges g 0 Effect size indicates the proportion of the total variance that is systematic o Cohen s criteria for comparing effect sizes Small 15 of variance is systematic Medium 615 Large gt15 V MetaAnalysis o MetaAnalysis a procedure to quantitatively assess a relationship across studies 0 Examine every study that has looked at a particular topic or relationship 0 Compare effect sizes across many studies to generate a general estimate to reflect the strength of the relationship 0 Note potential moderators difference in procedure measures sample one of most important steps 0 What is the purpose of metaanalyses in terms of variance To explain variability in effect sizes and findings Ch3 Basic Issues in Measurement 1132009 105900 AM I Types of Measures o A Selfreport measures 0 People s replies to questionnaires and interviews 0 Cognitive what people think and affective how people feel 0 Can measure Thoughts attitudes Feelings Actions o B Physiological and neurological measures 0 Used to study the relationship between bodily processes and behavior 0 Involves the use of specialized equipment to measure outcomes like Heart rate blood pressure body temperature brain activity hormonal changes o C Observational aka behavioral measures 0 Involve the direct observation of behavior 0 Observation in person or via AV recordings or the computer 0 Used to measure anything that can be observed Ex Via video movement action spatial location Ex Via audio sound language tone of voice Ex Via computer speed of responding o Converging operations 9 different kinds of measures provide same results yields higher confidence in their validity II Scales of Measurement o Four scales of measurement 0 1 Nominal scale objects or individual are assigned labels or categories Any numbers assigned are meaningless No mathematical operations possible For example a Gender male1 female2 a Political party Dem1 Repub2 Ind3 n Mental illness none1 depression2 o 2 Ordinal scale involves the rank ordering of categories behaviors or characteristics Examples rank order n In which runners complete a race a Of students by class performance a Of preferred foods Describes order of items but not the distance between ranked items units are not equal a Ex 1blueberries 2kiwi 3radishes o 3 Interval scale intervals between numbers are equal in size but there is no true zero point Can perform some mathematical operations but not ratios ex 100 F is not twice as hot as 50 F Examples n Scores on an IQ test a Ratings on a 5point agreedisagree scale 0 4 Ratio scale contains a true zero point that indicates the absence of the variable being measured Math operations including ratios can be performed Examples a Weight a Number of questions answered correctly a Time it takes to complete a task Ratio scales provide the greatest amount of information 0 Identify the scale of measurement for 1 Zip code nominal scale Grade of egg large medium small ordinal scale Reaction time ratio scale Score on the SAT interval scale Class rank ordinal scale Price of an insurance policy ratio scale Number of DVDs purchased last month ratio scale 8 Tshirt size ordinal scale III Discrete and Continuous Variables o Discrete variables consist of whole number units or categories made up of distinct units 0 Change in value occurs a whole unit at a time I IOXU39lPWN 0 Ex gender ethnicity siblings classes variables of nominal scale o Continuous variables usually fall along a continuum and allow for fractional amounts 0 Ex age height weight speed distance o Are the following discrete or continuous 1 Number of academic degrees discrete variable 2 Age continuous variable 3 Number of close friends discrete variable 4 Temperature discrete variable 0 5 Marital status discrete variable IV Components of Observes Scores o Observed Score True Score Measurement Error 0 True score the score obtained if the measure were without error Actual level on a measure P s true IQ score not necessarily valid or intrinsic to the individual P s true intelligence 0 Measurement error inaccuracy found in the measurement of a variable Distorts observed score so it is a less reliable indicator of the P s true score o Five sources of measurement error 0 1 Transient states a temporary unstable state of the participant Ex mood health level of fatigue anxiety 0 2 Stable attributes enduring traits of the participant Ex illiteracy paranoia hostility conscientiousness o 3 Situational factors Characteristics of the researcher eg rude friendly intimidating or the lab eg cold dark crowded o 4 Characteristics of the measure Long difficult ambiguous misleading or painful o 5 Mistakes in recording a P s score Typos failure to note behavior mechanical problems 0 O O O V Reliability o A Reliability the consistency or dependability of a measuring technique 0 Measurement error ME undermines reliability of a measure A measure s reliability is an inverse function of its ME If a measure is highly reliable near 100 Ps observed scores will be very close to their true scores 0 Testing reliability requires analyzing the variability in a set of scores Total Variance in a set of scores Variance due to true scores Variance due to measurement error 0 Reliability is the proportion of total variance that s systematically associated with Ps true scores Reliability Truescore variance Total variance 0no reliability 1perfect reliability o B Assessing reliability 0 To what extent do two or more measurements of the same behavior object or event yield similar scores 0 Researchers usually use a correlation coefficient to make those estimates 0 Correlation coefficient expresses the strength of the relationship between two measures Can range from 100 to 100 n 00 no relationship between the variables a 100 or 100 perfect relationship The sign indicates whether the relationship between the variables is positive or negative o C Types of reliability 0 1 Testretest reliability consistency of Ps responses on a measure over time Administer measure on two separate occasions Examine the correlation between the scores obtained on the two occasions Correlation gt 70 indicates acceptable reliability Useful only if the attribute being measured should not change over time o 2 Interitem reliability consistency among the items on a scale Are all of the items on a scale measuring the same thing If not averaging scores across the items creates ME and lowers reliability Indices of interitem reliability n A Itemtotal correlation the correlation between one item and the sum of all the other scale items a B Splithalf reliability divide the scale items into two sets and examine the correlation between sets o Correlation depends on how the scale was halved n C Cronbach s alpha coefficient at equivalent to the average of all possible splithalf reliabilities o Most frequently used c Adequate interitem reliability if X exceeds 70 o 3 Interrater reliability consistency among two or more researchers observations Examine the degree of agreement among 2 people who observe and record Ps behavior 2 raters must be independent and blind to the other s score o D Increasing measure reliability 0 Eliminate sources of measurement error Train observers Minimize errors in recording data Make instructions and questions clear Standardize the administration of a measure VI Validity o A Validity the degree to which a measurement instrument measures what it is intended to measure 0 Alternatively it could measure something else or nothing at all 0 Does the variability in test scores reflect variability in the characteristic or behavior we are trying to assess Are we measuring what we think we re measuring B Types of Validity o 1 Face validity the extent to which a measure appears to measure what it s supposed to measure Just because something has face validity doesn t mean that it is valid Many measures without face validity are valid Some are designed to lack face validity so as to disguise the purpose of the test 0 2 Construct validity the extent to which a measure of a hypothetical construct relates as it should to other measures Hypothetical constructs entities that cannot be directly observed but are inferred on the basis of empirical evidence a Ex intelligence status motivation love self esteem attachment style To have construct validity a measure should display a A Convergent validity a measure correlates with other measures that it should correlate with o Ex Embarrassability should be correlated with shyness but correlated with self confidence n B Discriminant validity a measure does not correlate with other measures that is should not correlate with o Ex Embarrassability should not correlate with IQ o 3 Criterionrelated validity A Concurrent validity B Predictive validity Ch3 Basic Issues in Measurement cont1132009 105900 AM VI Validity o A Validitv o B Types of Validity o 1 Face validitv o 2 Construct validitv A Convergent validitv B Discriminant validitv o 3 Criterionrelated validitv the extent to which a measure accurately predicts behavior or ability in a given area Are behavioral outcomes related to scores on the measure as expected Kinds of criterionrelated validity n A Concurrent validitv scores are related as expects to a criterion that is assessed at the time the measure is administered Ex An embarrassability scale administered today predicts stage fright in the current situation I B Predictive validitv scores are related as expected to a criterion that is assessed in the future o Ex An embarrassability scale administered today predicts whether students signup for public speaking classes next semester C Reliability vs Validity 0 Can a measure be Neither reliable or valid Yes Both reliable and valid Yes Reliable but not valid Yes a A measure may be consistent but not assess the intended construct 1 Ex Head circumference as a measure of intelligence Valid but not reliable No a If a test truly measures a construct it will also be reliable n Ex IQ test VII Fairness and Bias Test bias occurs when a particular measure is not equally valid for everyone 0 The question is not whether various groups score differently on the test Rather test bias is present when the validity of a measure is lower for some groups than for others Ch4 Approaches to Psychological Measurement1132009 1059 I Observational Methods o Observational methods making observations of human and animal behavior o A Natural vs Contrived Setting 0 Naturalistic observation observing human and animal behavior in their natural habitats No intrusion or intervention by the researcher 1 Yields good ecological validity the extent to which research can be generalized to reallife situations 2 Participant observation vs nonparticipant observation 9 researcher engages in same activities as people heshe is observing Examples a Risk taking choosing specific hiking path Advantages and disadvantages n Advantages ecological validity n Disadvantages only so many things that can be observed 0 Contrived observation observing behavior in settings that are arranged specifically for observing behavior Most take place in the lab but may occur outside the lab Examples n Aggressiveness kids attachment Advantages and disadvantages n Advantages more control specific situation a Disadvantages participants respond differently as they know they are in a study o B Undisguised vs Disguised Measures 0 Undis uised observation Present researcher or obvious AV recording Problems n Reactivity Ps act differently because they know they are being watched a Role of demand characteristics aspects of a study the clue Ps in to the researcher s expectations n Presence of observers impacts behavior via distraction anxiety evaluation apprehension social desirability concerns Advantages ethical easier 0 Disguised observation aka nonreactive observation Use of hidden AV recorders oneway mirrors a Ex Garbage graffiti obituaries content of parked cars Problems a May violate right of informed consent a Potential violation of privacy 0 Minimize reactivity by Partial concealment n Allow Ps to be aware of being observed but not the specific behaviors of interest Knowledgeable informants U Get people who know the Ps well to rate their behavior Unobtrusive measures a Use indirect measures that can be taken without Ps awareness o C Behavioral Recording 0 1 Narrative records full description of a P s behavior Used by Piaget in studying children s behavior 0 2 Checklists marks whether particular behaviors or attributes were observed Must formulate clear operational definitions a What are displays of affection aggressive responses and signs of fatigue 0 3 Observational rating scales Researcher rates the quality or intensity of a certain behavior Ex Rating a child s crying as a slight b moderate or c extreme Important to clarify operational definitions Must assess interrater reliability o 4 Temporal measures Reaction time the time that elapses between the stimulus presentation and response Task completion time the length of time it takes Ps to solve a problem or complete a task Interbehavior latency the time that elapses between the performance of two behaviors Duration how long a particular behavior lasts II Physiological Measures and Neuroimaging o Neuroscience 9 studies biochemical anatomical physiological genetic and developemental processes involving nervous system c Cognitive affective and behavioral neuroscientists focus both on physiological processes and psychological phenomena o Five types of psychophysiological and neuroscientific measures 0 1 Measures of neural electrical activity Eg EEG measures brainwaves o 2 Neuroimaging 9 allows researcher to see activity occurring within brain Eg fMRI provides an image of brain activity oxygenated blood flow 0 3 Measures of autonomic nervous system activity Eg heart rate blood pressure skin temperature 0 4 Blood and saliva assays Eg cortisol adrenaline testosterone o 5 Precise measurement of overt reactions Eg EMG to measure contraction of muscles related to eyeblink or smiling III SelfReport Methods o Selfreport measures ask Ps to directly report their thoughts feelings amp behavioral tendencies o A Questionnaires a written series of questions 0 Closeended vs openended questions 0 Ratin scale Different scales can influence responding n Ex 3pt 7pt 100pt scales end points labels o B Issues in making questionnaires O O 0 Specificity and precision in phrasing Use of difficult words unnecessary jargon and cumbersome phrases Unwarranted assumptions about the respondents eg literacy family structure Hypothetical information should come at the beginning of the Q Avoid doublebarreled questions Questions that ask more than one thing Avoid loaded questions Questions that includes nonneutral or emotionally laden terms Avoid leading questions Questions that encourage a particular answer Pretest the questions o C Experience sampling methods ESM Ps report what they are thinking and feeling right now 0 0 Diary method Ps keep a daily record of certain information Computerized experience samplinq Ps carry PDAs that prompt them to answer Qs throughout the day o D Interview schedule the series of questions that is used in an interview 0 How to make interviews more effective Create a friendly nonjudgmental atmosphere Maintain an attitude of friendly interest Conceal personal reactions to the respondent s answers Order the sections of the interview to facilitate building rapport and to create a logical sequence Ask questions exactly as they are worded Don t lead the respondent o E Problems with SelfReports O 1 Response biases Acguiescence bias the tendency to agree with statements regardless of content Naysaying bias the tendency to disagree with statements regardless of content o 2 Social desirability concerns the tendency to answer Qs in a socially acceptable way 0 3 Transparency and demand characteristics that clue Ps in to the researcher s expectations IV Archival Data o The archival method involves describing data that existed before the time of the study o Sources of existing data 0 Eg Census data court records personal letters old newspapers magazine articles government documents o Useful for studying o Phenomena of the past 0 Changes over time 0 Topics that involve articles ads or speeches V Content Analysis o Content analysis a set of procedures designed to convert text to more manageable data 0 Goal is to classify words phrases or other units of text into meaningful categories o Steps to Content Analysis 0 Decide what units of text will be analyzed words phrases sentences Define how the units of text will be coded Must classify or rate the text Raters code the textual material for all participants 0 O O Ch14amp15 Research Ethics 1132009 105900 AM I Researchers Two Obligations o 1 To enhance our understanding of behavioral processes that improves human or animal welfare o 2 To protect the rights and welfare of the human and nonhuman Ps studied 0 What do we do when these obligations collide II Approaches to Ethical Decisions o A Deontology ethics must be judged in light of a universal moral code 0 Eg lying is immoral so deception is unethical o B Ethical skepticism concrete and inviolate moral codes cannot be formed 0 Ethical rules as arbitrary and relative o C Utilitarian judgments regarding the ethics of an action depend on its consequences 0 APA and federal government guidelines 0 Weighing costs and benefits costbenefit analysis o CostBenefit Analysis Potential benefits Basic knowledge Improvement of research or assessment techniques Practical outcomes Benefits for researchers Benefits for research participants Potential costs Time and effort of participants Participants mental and physical welfare Money Deception and the creation of a climate of distrust III Institutional Review Board IRB o Members 0 From scientific and nonscientific disciplines 0 1 member from the community o Submission of a written proposal 0 Purpose 0 Procedures H O Iquot O 0 Potential risks and benefits o IRB approval required before study initiation IV Informed Consent o Ps must be informed of the study procedure risks confidentiality alternatives o Ps must explicitly agree to participate o Problems with obtaining informed consent 0 Compromising the validity of the study 0 Ps who are unable to give informed consent 0 Impractical circumstances V Other Issues o A Coercion to participate 0 When Ps agree to participate because of real or implied pressure from an authority figure 0 Excessive payment can be coercive o B Right to privacy 0 Right to decide when where to whom and to what extent behavior is observed 0 When privacy is not expected in public and some lab settings observation is not an invasion of privacy o C Confidentiality a P s data may be used only for purposes of the research and may not be divulged to others 0 The easiest way to maintain confidentiality is to ensure P anonymity o D Mental and physical risk 0 A study may produce pain stress failure anxiety or other negative emotions 0 Minimal risk is no greater in probability and severity than that ordinarily encountered in daily life or routine examinations o If more than minimal risk strong justification is necessary o E Deception 0 Use of deception Presenting Ps with a false purpose of the study Using an experimental confederate Providing false feedback Presenting two related studies as unrelated Debriefing stimulus material incorrectly Ethical objections eg lying is immoral Pragmatic objections eg fostering distrust Importance of debriefing 9 clarifying nature of study removing stress study may have induced obtaining participants reactions to study itself leaving participants feeling good about participation VI Ethical Research with Animals o Required monitoring by a person with animal care experience o Available veterinarian for consultation o Healthy and humane housing o Minimized discomfort o Strict APA guidelines VII Scientific Misconduct o Categories of scientific misconduct o 1 Fabrication or falsification of data plagiarism o 2 Questionable research practices Contribution does not justify authorship Failing to report inconsistent findings 0 3 Other unethical behavior Sexual harassment Abuse of power Discrimination O O O Failure to follow government regulations VIII Suppression of Findings o Research on controversial or unimportant topics may be halted by legislators 0 Despite the approval of IRBs and funding agencies o Suppression of controversial findings Eg metaanalysis of sexual abuse findings in Psychological Bulletin Ch14amp15 Research Ethics 1132009 105900 AM Wrestling with Ethical Issues o For each study answer 0 1 Are there any ethical issues that should be raised concerning the research described 0 2 How could the ethical problems be remedied o Examples 0 1 Students told to sing song in private though singing was recorded Played back to cause embarrassment Ethical problems Loss of trust in researchers and scientists Remedies Consent after participant has sung o 2 Participants coaxed to cheat on test Some did but all were accused of cheating by experimenter and told of possibility of being taken to review board Ethical problems Lying and threatening Remedies Conduct questionnaire that isn t so accusational o 3 Shocking dogs while suspended to study conditioning Ethical problems No consent animal cruelty Remedies Minimize amount of suffering o 4 Milgram study every incorrect answer of the participant triggers another participant a confederate to be shocked with increasing voltage Scientific Writing o Introduction 0 Describe the problem under investigation 0 Discuss relevant aspects of the existing research literature 0 Discuss the purpose and rationale of your research 0 State explicit hypotheses a priori but avoid HARKing hypothesizing after the results are known making an ad hoc explanation o Method 0 Participants not subjects Who they are how they were selected 0 Apparatus or Materials
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'