ge PHINMA EDUCATION a psy 084: Field Methods In Module #8 Student Activ Name: \ Class number; Ce . 2 oo ita Date: Lesson title: DESIGN PARE PERIMENTS: AND SINGLE FACTOR Learning Targets: : a = Of the module, students will be able to ‘xplain true experiment to isolate cause and effect; ‘Materials: Textbook, pen and notebook, index card/class list 2. oe independent variables being manipulated and References: 'et Salient variables being controlled; and s Cen : - Coolican, H. (2014). Research methods iple experimental designs and simple two- cool atiatics in psychology (6" ed.) condition designs. hology Press. Psycholo, 3 —_—_—_— A. LESSON PREVIEW/REVIEW As a review of the previous discussion, the instructor will ask the students the following: 1. Whatis a partial and semi-partial correlation? 2. Describe a multiple correlation. 3. When will you use a spearman’s rho? B.MAIN LESSON ‘The experiment is considered one of the most powerful ways of doing this mainly because it has the power i eliminate a lot of alternative explanations which can occur with other kinds of evidence. It is more likely thar ‘other methods to demonstrate a relatively unambiguous connection between cause and effect and thesé connections are what science tries to establish,BB inn coucarion PSY 084: Field Methods in Psychology Module #8 Student Activity Sheet ‘ Class number: Name: Section Schedule: Date: Alternative explanations There are always _alternative causes ‘ngoression interpretations of findings. The trouble here is that as heat goes up, so do a“ certain other factors. For instance, the But heat is related to: warmer it is, the more people there are out in public places, the longer are the | queues for ice cream, the longer are , the trafficjams and soon. Itmay not be Explanatlon 2: Numberofpsopleousite —/Amjokhese NS heat at all that causes the aggression Explanation 3: Lengih of ques cause of ition 4: Amount of road traffic ...etc. but one of these other factors (see F*plana Right Figure). The trouble is that with this kind of non-experimental design we can't exclude these other factors as they operate in everyday life. The point being made here is that we can gather data that supports a hypothesis but very often, in studies like these, we cannot rule out competing explanations; that is we cannot confidently point to a clear cause and effect Explanation 1: ‘Alternative explanations of the heat-aggressin lnk. relationship between events and human behavior — we can't be at all sure that it is © Experiment ‘Study in which an independent variable is manipulated under strictly controlled conditions. ¢ Randomization Putting stimulus items or trial types into random order for the purpose of elimination of order effects. Dependent variable 7 The dependent variable is the variable in the experiment whose changes depend on the manipulation of the independent variable. Itis the variable we measure to see if it has changed as a result of the diferent conditions in the experiment. We do not know the values of the DV until after we have manipulated the IV. cow the values of the dent varia Independent variable We kn imenter. ‘depends on The independ. i i i peri pendent variable isthe variable which is manipulated by the experimerte® Csr open IN before we start the experime See periment Here, then, the numberof aggressive esPereSs responses dens I aim of an experiment is 10.4 and the temperature is the independent variable. The change in the change in temperature — see Figure Below. The essential unambiguous ‘causal link’ between these two variables.BB 3.214 0u00n psy 084: Field Methods in Psychology Module #8 Student Activity Sheet Name: eS Class number: __ Section: Schedule: Date:_— Manipulation of the independent variable |__+f procs | ‘Temperature change Change in the dependent variable Number of aggressive story endings «Independent variable (IV) Vartable which experimenter manipulates in an experiment and which is assumed to have a direct effect on the dependent variable. . ble (DV) | Parte ieh ea tly affected by changes in the independent variable in an experiment. Variable that is assumed to be direc! . sf the 1V) Levels (of the 1) by the independent variable; often, the conditions of an experiment, e.g. levels The different values Teste at 90mg, 100mg and 200mg inthe investigation of memory recall © Baseline measure ppaseite of what would occur if no experimental level of the independent variable were applied: how ‘untreated’ participants perform. «© Presest Moosure of participants before an experiment in order to balance or compare groups, or to assess change by comparison with scores after the experiment. Features of the true experiment; avoiding confounds Ina true experiment the researcher: © Manipulates an independent variable «Holds all other variables constant (including random allocation to conditions/levels of the IV) Measures any change in the dependent variable is the heart of scientific experimentation. We saw that in the Rule et al. experiment participants were randomly allocated to conditions. This was to eliminate any variation between the groups that could be caused by participant variables; there might by accident be more aggressive people in the hot temperature condition. Such a pre-existing difference would confound our interpretation of the findings. The middle point above is crucial and i ‘A confounding variable has the effect of hiding the real effect of our IV. We might conclude that temperature increases aggression when in fact we started off with more aggressive people in the higher temperature condition, The participant variable confounds any actual effect of heat; we now don't know if heat had an effect ‘ornot. Other variables we might hold constant would be the humidity in each room, the amount of noise, and the exact instructions given to each group and, of course, the content of the stories they have to respond to. In doing this we try to leave only one variable (the IV) that is changed by the researcher and one other variable (the DV) This document is the property of PHINMA EDUCATIONPHINMA EDUCATION PSY 084: Field Methods in Psychology Module #8 Student Activity Sheet Name: y Section: Schedule: : Class number: ate: which changes as a result. Other variables which could affect the results are known generically as extraneous variables. * Confounding variable Variable that is uncontrolled and obscures any effect sought, varying with the independent variable in a systematic manner. « Extraneous variable ‘Anything other than the independent variable that could affect the dependent variable; it may or have been allowed for and/or controlled. ds Non-equivalent groups two or more groups in an independent samples design experiment ‘A possible confounding variable where differ on a skill or characteristic relevant to the dependent variable. behavior then we need to compare this behavior Control groups its include a control group to act as a baseline If we want to see wh with what would occ measure of behavior without tre: .d participants’ hether a ‘treatment’ has altere perimen ‘ur without the treatment. Many ext atment. periment we just considered. in an experimen! wie He avr audience reduces performance on 2 word recognition task we obviously need ithout an audience in order to compare. A control group is not abways / behavior alters when they talk to a male baby, 48 with 2 female baby, 1 group. However, it might also be esting to know how people tak when they This would be the group not given caffeine in the ext want to know if the presence ‘a control group who perform If we are noting how people's then we do not need a control don't know the baby's sex at al ‘Control group Group used as baseline measure against which the performance of the experimental group is assessed i as the imilar to a control group in that they experience exactly the same ayers 1 of the independent variable that thought to affect denne caives nothing, the placebo arouP receives someting Placebo group ‘A placebo group is si experimental group except for the level variable. However, whereas the control Group re lke the active level of the IV but is ineffective tal grou? Placebo group Group of participants who don't receive the atitic eretes; used in order to eliminate placebo effects al ‘treatment’ but ~ participants may think they have received an effective treatment.PHINMA EDUCATION. ’ Name: Section: Brief critique of the experiment The strengths of experiments > Can isolate cause and effect because independent variable and extraneous variables (p. 67) are completely controlled. «Alternative explanations of effects can be Aertigated/eliminated in extensions of the original experiment because experiments are easier 10 feplicate accurately than many other studies ‘extraneous influences 50 id alternative explanations weakened. + can control many that validity is high ani of events are eliminated or Strengths and weaknesses of experimental designs- ‘The independent samples design The students here are conducting what is known just what it means. One group of participants is audience) while an entirely different (independent) there is no audience. These two conditions are the level ‘audience’, the levels being present and absent. ‘¢ Independent sample (between group: Each condition of the independent varia oe Schade + Limits the kind of as an independent samp! subject to the experi group of people participates in Js of the independent vari s; groups/subjects independent/unrelated) bie is experienced by only one group of participants. psy 084: Field Metho Module #8 student class 1! Date: critique of experiments + Participants’ no unique vie' unless intervie «reactivity effects (P- know they are in an Ethical problems if they contribu! ww from pal wed after © : 446) occur i experiment: do not. that can be investigated because Variables must Be fight in operational definition, Sand conditions Ierefore be quite artificial: : Can invest results with false ‘scientific credibility’. Jes design experiment. The title ‘ental condition (the presence of the control condition able which we could ‘The generalized experiment depicted in Fig The din Figure below is also an independ (2002) and Can et al 1980) studies described earer are examples otindepandent scmple designs Ol “ tere fe independent samples design re INDEPENDENT GROUI eee seoiat oes tea ae IPS, UNRELATED GROUPSIDESIGN, Experimental indomly for a admini allocated to .. Control = participants — nistered —~ Measured on group exaclly the dependent variable same The classic two Group experimental design with j No treatment Repeated reatmel measures design int and control (no treatment) groups. TWis document isthe property o PHINMA EDUCaTina, lent Is the property of PHINMA Ep) UCATIAN ReneePSY 084: Field Methods in Psychology Module #8 Student Activity Sheet ene : Class number: Date: Section: Rickert conducting ihe audience effect experiment ran into the problem that their two groups of participants folate iy be equivalent. There might have been more people with relevant skills in one group than in the Mig there a way to eliminate these differences altogether? One of ihe ‘students comes up with the idea of having the same people do both conditions. This way. all diference® between participants become irrelevant dad we are really testing whether an audience has an effect on ‘each individual's performance, compared with when they were alone. ‘out to investigate, is sometimes called a within panicipant (or ‘subject. The more usual tem, Cat fe repeated on each participant but under the sunts are the same for both conditions, and all ofhes ct of the manipulated independent ,asures design, which we are abi This is why the repeated me: ooking at differences within eac subjects design, since we are ‘repeated measures’, refers to the fact that the same meas Various conditions of the independent variable. If the particip: variables are controlled, any differences within participants should be the effe« Variable, barring the odd accident affecting maybe one or two participants. sures (within subjects/groups) ols of the independent variable «Repeated mea Each participant experiences all lev Check for Understanding Multiple Choice 4. Variable which experimenter manipulates in an experiment and which is assumed to have & direct effect on the dependent variable. 3. Independent variable b. Randomization cc. Dependent variable d. Extraneous variable f order effects. order for the purpos 2. Putting stimulus items or trial types into random a. Independent variable b. Randomization c. Dependent variable d. Extraneous variable dent variable that 3, Anything other than the indepe! rolled. have been allowed for and/or cont a. Independent variable b. Randomization c. Dependent variable d. Extraneous variable ed by ¢ ‘4, Variable that is assumed t0 De directly affect a. Independent variableLN PHINMA EDUCATION Section: Schedule: MIMENTS AND SINGLE FACTOR Lesson title: EXPE! DESIGN PART 2 Learning Targets: Atthe end of the module, 4, Explain true experiment 2. Identify independent variables Del ‘all other salient variables being CO" 43, Describe simple experimental designs 2 condition designs. A. LESSON PREVIEWIREVIEW students will De able to: in to isolate cause and effec ing manipul ntrolied; and nd simple WO" ———— ‘Materials: rextbook, pen and oe card/class list « oO es References: oe oe Coolican, H. (2014). Res and statistics in ps) Psycholos jents the following: ‘as a review of the previous discussion, the instructor will ask the stud 4. What are the features of a true experiment? 2. Differentiate extraneous variable from confounding variable. 3, What is a placebo group in a study? B.MAIN LESSON Order effects You probably realized that people might improve on the second condition simply because they've had practice (and they may be less anxious about learnit participate in conditions are known as order effects. we ale ee ee het iu ln ed tt is of course possible that people would perform worse in the second condition, perhaps because they were . cause they were disheartened by failure, thro lure, through boredom or through fatigue. Ei * it sane le otec som afte mej dsodenioyes al atepesiod measuesteaioe ures design © Order effect A confoundini 1g effect caused by experiencing one condition, then another, such a: ; 8 practice or fatigue. . Asymmetrical order effect der effect that has greater strength Cees igth in one particular order and where, therefore 2, counterbalancing Dealing with order effects This document is the property of PHINMA EDUCATIONate % Setctremenaane iad PSY 084; Field Methods in Psychology Module #9 Student Activity Sheet Class number: Date: Schedule: lancins : a * Tatas ee inthe lone condition might have been {improved because ot tei pron 7 in the audience condition, it makes sense to have hal 4 the other haf to perform in front ofan audience fst-This i known 2S COUNTERBALANCING the conditions and is very often part of the procedure in @ repeated measures design. Calling the conditions A and & one group does the AB order while the other group does the BA order. 2. Complex counterbalancing ABBA , ] To,balance asymmetrical order effects (to some extent at least), all participants take conditions in the order ABBA. Their score in condition Ais taken as the mean of the two A trials and likewise for By Multi-condition designs If an experiment has three conditions (i., three levels of the independent variabl Participants randomly into six groups and ha le) we might divide ive them take part in the following order: 8 of condition: © Counterbalancing Half participants do conditions in a particular order and the other half take the conditions in the opposite order — this is done to balance Possible order effects. We can present the six conditions to each Participant in a different randomly arranged order. This could be arranged by having a computer arrange the Rumbers 1 to 6 in a random order for each participant and giving the six conditi givin ‘ons according to this set of numbers. This should dissipate any order effects sll further, Randomization of stimulus items Acifferent solution in the last example would be to give participants just one list of anagrams contai all the anagrams from the Six conditions in just one list with the anagrams randomly mixed together. This next 088 eliminate order effects since here, patcipants Jo nat compete one conlion then the Standardized (procedure) esting or measuring behavior with exactly the same formalized routine for all participants. 5. Elapsed time © can leave enough time between conditions for any learning or fatigue effects to dissipate. 6. Using another design This document is the property of PHINMA EDUCATIONPHINMA EDUCATION PSY 084: Field Methods Module #9 Student Name: Section: Schedule: We may have to give up the idea of using the same group for each condition. We could m ‘independent samples design’, but, since this design has important disadvantages, we might try tg this more drastic solution to the problem of order effect Example of a vignette study - Baltazara, Shuttsb & Kinzlera (2012) gave ¢-year-old children vignettes which included a picture of al child, a statement that he/she was ‘nice’ or ‘mean’ and then a short description of what that child had! done (e.g. brought in cookies for everyone or stole everyone's cookies). On later testing the Tesearchers showed that children this age could more easily recall who was mean than who was nice, could recall more specific actions of the ‘mean’ characters than of the ‘nice’ characters and maintained this superior memory for ‘mean’ actions even when the person in the vignette was presented as the victim rather than the perpetrator of the mean or nice act. Note that this was a repeated measures design with each child seeing four ‘nice’ and four ‘mean’ characters, balanced for sex and also counterbalanced in a complex manner that meant order of presentation could not be responsible for any of the differences found. * Vignette A story, scenario or other description given to all participants but with certain details altered and this difference constitutes the independent variable Comparison of repeated measures and independent samples designs Repeated measures and independent samples designs are very much alter egos. The advantages of one are the disadvantages of the other. They are by far the most commonly used of experimental designs and so it i worth comparing one with the other. . Disadvantages of repeated measures — a strength of independent samples * The repeated measures design has the problem of possible order effects, described abo independent samples design avoids. ve, which the * __'ve lose a participant between conditions ofa repeated measures design, we have lst a score fr both conditions since we can't use just one score ftom any participant ~we must have aero independent samples design partcipants are tested jst once soloing @ parispant means me : a score from one condition. 2 * The aim of the experimental research may become obvious tothe participant who takes bot Creating demand characteristics and this makes ‘pleasing the experimenter’ or ‘screwing more possible. th conditions, up the results * In a repeated measures design we may have to walt until practice effects have worn off and the Participant is ready for another test, whereas with independent samples we can run the two conten ® simultaneously with two different groups of participants. * Ifeach participant experiences both conditions of. say, a memory experiment using word lists, we have to use a different list in each condition This docurnent is the property of PHINMA EDUCATION eee!0% PHINMA EDUCATION 4, eae PSY 084: Field Methods in Psychology % Module #9 Student Activity Sheet my foXy} Name: Class number: "YW section: Schedule: Dat Disadvantage of independent samples — a strength of repeated measures : : ‘* The major problem is that of non-equivalence of samples, described above. Differences in performance '; between conditions can be mainly the results of participant differences. * With repeated measures we obtain ten scores in each condition using just ten participants, whereas with independent samples we'd need double that number of participants to get ten scores in each condition Participants can be hard to find and individual testing takes time. 7 * {there is too great a difference between the statistical variances of independent groups we have to ater Storstastical procedure in carrying out a parametric test. Parametric tests are preferred to war aaerratves in torms of their power to provide evidence for significant differences. f the numcee oe Participants in the various conditions are very different we may not be able to Significant difference when, in reality, such a difference exists eds * Matched pairs Each participant in on groupicondition, °UP/Condition is paired on specific variable(s) with a participant in another Single par ‘A design using ju A person being 2 os yale Participant is known, unsurprisingly, as a single participant design. To hear of just one Produce quite unreprese omental research can make the scientifically minded reco in horror. Surely this must resentative results, impossible to generalize from? ant and small n design Quite ‘rightly, they assume, one the generalizations la order to avoid the many turns to objective psychological research 'Y People make from personal anecdotes about single individuals. Single participant “S'0n In which only one participant is tested in several rials at allindependent variable eves. Small N design z r work but acean'n which there is only a small number of participants, typically in clinical or counselling wi ere Participants need substantial training for a highly skilled task This ocument isthe property of PHINMA EDUCATIONBD rinse coucmon Name: So a Section: Schedule: igns imental des! ‘Advantages and disadvantages of the various exper! Advantages No order effects. Design lables not equivalent var can use exact the same Participant ar us lists, etc, In each controlled condition. Participants can’t guess alm of experiment. No need to walt for participants to ‘forget’ first een Counterbalance or a Repeated Pact partpant conditions/stimulus measures control so participant cameos variables have litle effect. ve economn Need different but equivalent conditions. port valor stimulus ists Do independent samples artery of vaience Loss of participants between instead not a problem conditions. (see Chapter 17). May not be able to conduct_—_Ensure testing conditions second condition immediately. are identical for second condition. Deceive participants as to. aim (or leave long time gap). Participants not naive for second condition and may ‘try to quess aim. May be hard to find perfect Matched —_No walt for participants to ‘matches and therefore pairs forget frst condition, time consuming. Can use same stimulus ists, Loss of one member of pair atc entails loss of whole pair. Partcpant variables partly Lond participant variables Randomly allocate pairs to No order effects possible amen Homogeneity of variance 3 ‘not a problem, ingle Useful where f Pantcipant participants avai Retraining required if any cain aerate el arcetaes TM Parra vr i ject required for taining Participant tae OPANt In specialised Relates fated and unrelated designsHIINMA EDUCATION PSY 084: Field Methods in Psychology Module #9 Student Activity Sheet Name: Class number: Section: Schedule! Date: j / ‘ s' design is an unrelated design. We say this simply because each Stove n one group oF ccilon cannot Iv any way be relaed to any specie score in the oer groupe conditions. veral trials in each condition of an experiment ticipant design, very often the participant performs se\ n ' © Unrelated Design in which individual scores in one condition cannot be paired (or linked) in any way with individual scores in any other condition. + Related - : Design in which individual scores in one condition can be paired with individual scores in other conditions, Factors in an experiment Independent variables are also known as factors in the experiment and this is the terminology to use when discussing more complex (multi-factorial) designs. Thus, the audience and task design just introduced would ‘contain two factors each with two levels. Or ne independent variable would be the task that participants are asked wre (oem = simple or complex. This factor would be called task and would have levels of simple and complex while the audience factor has the levels of present and absent, * Factorial design Experiment in which more than one independent variable is manipulated. ‘Check for Understanding Multiple Choice 1. A confounding effect 8. Single participant b. Small N design ©. Order effect 4. Asymmetrical order effect caused by experiencing one condition, then another, such as practice or fatigue. fore, counterbalancing Oeeet effect that has greater strength in one particular order and where, therefore, Would be ineffective. 3. Single participant Small N design ©. Order effect - Asymmetrical or der effect indapandentvoble ves tall 3. Design in wi only one participant is tested in several tials at @._ Single participant THs documents the property of PHINMA EDUCATION