Study Guides (248,605)
Canada (121,634)
Psychology (952)
PSYC 2360 (37)
Final

final review-notesolution.docx

43 Pages
126 Views

Department
Psychology
Course Code
PSYC 2360
Professor
Naseem Al- Aidroos

This preview shows pages 1,2,3,4. Sign up to view the full 43 pages of the document.
Description
final review 1/21/2013 4:08:00 PM CHAPTER 11: Experimental Research: Factorial Designs I  Outline o Factorial experimental designs o 2 x 2 Between-subject factorial designs o Understanding 2 x 2 interactions o Interpreting 2 x 2 interactions o Factorial designs with more than two factors Example: Causal attributions  Finding: o When interpreting another persons behaviour, we take social context into account. o But only when we have the resources (i.e., time, mental energy, etc.).  Gilbert et al. (1988) o Person perception is a sequential process: characterization->correction Factorial Experimental Designs  Example: o Observe person having conversation and behaving anxiously  Factor 1: Social context  Discussing anxious vs. relaxing topics  Factor 2: Resource availability  Low vs. high cognitive load o  designs o  manipulate more than one independent variable  AND  levels of all variables are completely crossed o types:  2 x 2 – two two-level factors (most commonly used)  2 x 3 – a two-level factor and a three-level factor  3 x 3 x 3 x 3 x 3 x 3 x 3 x 3 x 3 – nine three-level factors o Provides:  Number of factors  (given by the # of #’s)  Number of levels if each factor  (given by the value of each factor)  Total number of unique conditions  (2x2=4 unique conditions, etc.) 2 x 2 Between-subject factorial designs  can see: o Main effect of factor 1 o Main effect of factor 2 o Interaction between factor 1 and factor 2  Example: Subliminal messages (Greenwald) o Listen to self-help tape. Asked, do you think it helped your memory?  Factor 1: Subliminal content on tape  Memory help vs. self-esteem help  Factor 2: Label on tape  Memory vs self-esteem  Results o Main effects = the impact of one factor on the DV collapsing across all other factors in the experiment o Determining main effects…  look at your marginal means  means combined across the levels of another factor (collapse across the effects of the other factors).  Main effects o Main effect of Tape Content on perceived memory benefit?  F(1, 86) = 0.79, p = .377 o Main effect of Label on perceived memory benefit?  F(1, 86) = 21.72, p < .001  Interaction o Does the effect of Label on perceived memory benefit vary depending on the Content of the tape?  Yes  Interaction  No  No Interaction o Do we see a significant interaction for type of tape label x tape content?  F(1, 86) = 0.26, p = .610  Bar vs. line graphs Understanding 2 x 2 interactions  Interactions o A significant interaction indicates:  The effect of one IV on the DV differs across the levels of the other IV  How to recognize interactions o Look at a line graph  If the lines are not parallel, then there is an interaction.  How to recognize main effects and interactions o Main effects  look at the table of means  looking at the graph o Interactions  look at the graph  Example: o Effect of height and gender on yearly income  IV1: Height  Tall vs. Short  IV2: Gender  Male vs. Female  Fully crossed design  Every level of IV occurs with every level of the other IV  Tall Males, Tall Females,  Short Males, Short Females  DV: Yearly Income  How to recognize main effects and interactions  ―x‖ if points are different, main effect  ―•‖ if points are different, main effect  If lines are parallel  same ―x‖ no main effect  same ―•‖ no main effect  lines not parallel Interpreting 2 x 2 interactions  interactions: o multiple patterns that led to a significant interaction.  ..how to make specific conclusions..?  Comparing condition means o Easy for a 2 x 2   Use T-test to compare RELEVANT pair of conditions. o Which pairs should you compare?   guided by the experimental question (hypothesis)  examples: o Hypothesis: Goal management training benefits healthy people more than TBI patients.  Factor 1: Patients vs. healthy  Factor 2: Training vs. no training  Results:  Significant main effects of training and population  Significant training x population interaction  o Hypothesis: Practicing a sport improves performance when coached, and worsens performance when not coached (bad habits).  Factor 1: Coached vs. not coached  Factor 2: Before vs. after practice  Results:  Significant main effect of coaching, but not practice.  Significant coaching x practice interaction  (compare 2 points on coached line, and 2 points on non-coached line) o Hypothesis: Attending to categories (e.g., faces vs. houses) activates the brain region that is specialised for processing that category  Factor 1: Attention to faces vs. houses  Factor 2: FFA brain region vs. PPA brain  Results:  No significant main effects.  Significant brain region x type of attention interaction  (compare PPA means & FFA means) o Hypothesis: Taking fish oil pills reduces cholesterol, but it takes time  Factor 1: Fish oil vs. no fish oil  Factor2: 1 vs. 3 vs. 5 weeks  Main effect of fish oil, but not time  Significant fish oil x time interaction o Hypothesis: Taking fish oil pills reduces cholesterol, but it takes time  Factor 1: Fish oil vs. no fish oil  Factor 2: 1 vs. 2 vs. 3 vs. 4 vs. 5 weeks  Main effect of fish oil, but not time  Significant fish oil x time interaction  Comparing condition means o Each individual comparison has an alpha likelihood of resulting in a Type I error.  performing multiple comparisons results in GREATER than alpha chance of making a type 1 error. o Familywise error rate  the probability that one or more of a series of significant tests results in a Type I error.  Simple effects o Hypothesis: Taking fish oil pills reduces cholesterol, but it takes time  Main effect of fish oil  Significant fish oil x time interaction  Results:  Simple effect of time in fish-oil group  (but not no-fish-oil group)  Summary o if only main effect(s) are significant  .. interpret them! o if only the interaction is significant,  … interpret it (i.e through mean comparisons) o if the main effect(s) are significant and so is the interaction  … interpret the main effect as qualififed by the interaction Factorial designs with more than two factors  Effect of gaze and expression on face and object ratings o IV1: Gaze  Looking vs. Averted o IV2: Expression  Happy vs. Disgust o IV3: Rated item  Object vs. Face o DV: Liking Ratings  How much do you like this face/object?  Factorial ANOVA with three independent variables o Seven F-values in total  One for each of the main effects  of which there are THREE; one for each IV  One for each possible two-way interaction  of which there are THREE; each pair of IVs  ONE for the three-way interaction o Significant 3-way interaction, so..  .. split the experiment in half, and analyze each 2x2 interactions separately.  Three advantages of factorial designs o 1) More efficient:  test for more than one ME in the same experiment o 2) More comprehensive:  tells us more of the whole story- how do variable interact. o 3) More valid:  increased external validity  (can generalize conclusions to more situations) CHAPTER 12: Experimental Control and Internal Validity Threats to the validity of research  Threats to statistical conclusion validity  Threats to construct validity  Threats to internal validity  Threats to external validity  1) Statistical conclusion validity o Conclusions regarding research may be incorrect because of a Type 1 or Type 2  Are the statistical inferences legitimate?  -when researchers make insertions, have they made an error? (type 1 or type 2 errors)  2) Construct validity o Claim: measured variables/manipulations relate to conceptual variables of interest  -whether measures/manipulation map on to intended construct? o Reality: they may not  Is the chosen operational definition legitimate?  -does your measure match onto your variable?  ..not just whether your measure matches onto conceptual variables, but also if the manipulations match on.  -correctly changes the variable you were targeting?  3) Internal validity o Claim: changes in IV caused changes in DV  Are there confounds? o Reality: changes in DV may have been caused by a confounding variable.  Have alternate accounts of an effect been eliminated?  -are there confounds?  -are changes in IV being caused by changes un DV or something else?  4) External validity o Claim: results are generalizable  How generalizable are the findings? o Reality: observed effects may only be found under limited conditions or specific groups of people  Do the results generalize to the situation or population of interest?  -verify that the generalization are correct?  -overgeneralizing causes problems with external validity.  (do results really generalize that broadly?)  -trade of between external & internal validity. Experimental Control  Occurs to the extent that…  Experimenter can eliminate all effects on DV o .. other than those caused by the IV!  -we want to limit as much variance in the DV as possible, so that the only thing effecting it is the IV. Extraneous vs. confounding variables  Extraneous variables (additional factors causing changes that have no relationship with the variable at all) o -washes out the statistical test of IV, and decreases the statistical power by masking it by noise.  confound variables (cause changes in DV, but are confounded with IV-mingled together causing changes) o -manipulating IV effects Confounding varible too, and both cause changes in DV o -is an alternate explination, have no way of proving with one is true.  WORSE than extraneous variables! Improving statistical conclusion validity   by controlling extraneous variables ! o -get rid of other things that influence the DV. o -reducing the likelihood of making type 1 & 2 error.  4 methods of Controlling extraneous variables o Limited population designs o Before-after designs o Matched-group designs o Standardization of conditions o 1) Limited population designs  Select from a limited, and therefore relatively homogeneous, population   reduce random error!  -substantial reduction in likelihood of getting difference between groups due to chance.  -limiting your population will improve statistical conclusion validity.  -limiting individual differences (& differences between groups) creates internal validity, at the cost of external validity. o 2) Before-after designs  Dependent variable is assessed before and after the experimental manipulation  -before you manipulate, get a base line measure from all participants. Then manipulate, and get new measurement.  -using a baseline helps us because your no longer comparing peoples performance to others performance. your comparing there performnce to there own (within- subject)  Advantages  Any differences among the participants will influence both the baseline measure and the measure serving as the DV.  the power is increased by controlling for variability among the research participants.  Disadvantages  Although completion of the dependent measure more than once helps reduce random error, it also creates the possibility of retesting effects.   the likelihood participants will be able to guess the research hypothesis is increased o 3) Matched-group designs  Participants measured on the variable of interest before experiment begins. Then assigned to conditions on the basis of their scores on that variable  Ex. get measure of IQ before the experiment, and match people in groups compared to there IQ.  2 smartest people, separate, next 2 smartest, separate, etc.  controls for individual differences.  Advantages  Reduction in random error  Increase in statistical power  Note  -pretty good choice!! not many disadvantages, just that it can be very hard.  Random assignment is typically sufficient to prevent between-group differences  But can improve experiment power by matching known extraneous variables  -can get to equivalence between groups more quickly by doing this. o 4) Standardization of conditions  All participants in all levels of the independent variable are treated in exactly the same way, with the single exception of the manipulation itself   experimental script/protocol  (that is strictly followed)   automatization  after your done with consent form, put in room with computer which takes you through experiment, always same computer o (but sometimes people don’t read instructions) Improving construct validity through effective experiments/manipulations  Construct validity in an experiment o IV and DV are each associated with a conceptual variable o Was the manipulation related to the conceptual IV? o Was the measure related to the conceptual DV?  Good manipulations will create: o 1) Experimental realism  Extent to which the experiment engages the participants, and gets them to take the instructions seriously.  -did the manipulation engage the participants, did they become involve, really take part in it.  ex. Milgram’s study o 2) Impact   have impact when the manipulation creates the hoped-for changes in the conceptual DV.  if it effected the DV it worked.  (when no differences in DV, maybe your hypothesis was wrong, maybe it didn't work, etc)  Manipulation checks o  Determines if the experimental manipulation had the intended impact on the conceptual IV  -an additional way to verify that our manipulation had an impact on something. o Not to be confused with:  suspicion check   if subject guessed manipulation.  confound check  verifying the manipulation worked the way you wanted to. o Example: prediction that arousal will effect memory  -shown negative/happy pictures then give memory test.  manipulation check   include an additional measure, maybe a physiological response to ensure that the manipulation had an impact on some measure.  suspicion check   did you think they had to do with each other, were you suspicious.?  cofound check   did it produce additional differences betwen groups (maybe people in the happy face group was easier because they could create mental stories- did you use any strategies?)  Pilot testing: o Conducting the manipulation on a few participants before beginning the experiment itself  trying to asses qulity of manipulation & measures, you want to do this before yoru experiment. o  Helps to ensure that the manipulation checks and confound checks will show the expected patterns Improving internal validity by eliminating confounds  Confound o A variable other than the IV that differs systematically across the experimental conditions (intertwind). o  Impossible to determine which of the variables has produced changes in the dependent variable  Alternative explanations o Confounding variables always produce alternative explanations for the results o Cannot know which explanation(s) are correct!  Threats to internal validity o Participant expectations  A) Placebo effects  B) Demand characteristics o Experimenter expectations  C) Experimenter bias o A) Placebo effects  participants can have expectations about the likely effect of an experiment manipulation  -this itself can have an effect on the DV  -indepdent of any actual effect of the IV  -when aware of manipulation, they adapt expectation of what they think the manipulation will cause, & effects behaviour. o B) Demand characteristics  Something about a study can prompt participants to guess the hypothesis.  This can impact their responses  …they might test my memory or somehow try to trick me.  Socially desirable responding  Reducing demand characteristics  1) Cover stories  2) Unrelated-experiments technique  3) Use of nonreactive measures  1) Cover stories  A false or misleading statement about what is being studied  Used to prevent participants from guessing the research hypothesis  2) The unrelated-experiments technique   story & measure happen in 2 different tasks.  Story:  Participants told they will be participating in two separate experiments  Truth:  experimental manipulation is presented in the first experiment  dependent measure is collected in the second  3) Nonreactive measures  DV:  participants do not realize what is being measured  or cannot cont
More Less
Unlock Document

Only pages 1,2,3,4 are available for preview. Some parts have been intentionally blurred.

Unlock Document
You're Reading a Preview

Unlock to view full version

Unlock Document

Log In


OR

Join OneClass

Access over 10 million pages of study
documents for 1.3 million courses.

Sign up

Join to view


OR

By registering, I agree to the Terms and Privacy Policies
Already have an account?
Just a few more details

So we can recommend you notes for your school.

Reset Password

Please enter below the email address you registered with and we will send you a link to reset your password.

Add your courses

Get notes from the top students in your class.


Submit