PSY 355 1st EditionExam # 3 Study Guide Lecture: 3/16 One Independent Variable Between-Groups Design -Instead of just one treatment versus a control group or one treatment versus another, we often use designs that compare several levels of the same factor. -T-test cannot be used to analyze the data; instead, ANOVA (Analysis of Variance) is used. -Sometimes we use ANCOVA or MANOVA. All discussed later. One Independent Variable Between-Groups Design Examples: -From Book--No Arousal, Low Arousal, Moderate Arousal, High Arousal -What is IV?-How many levels of IV? -Practice times: 15 minutes, 10 minutes, 5 minutes -Treatment study: Treatment alone; Treatment with spouse or partner; Treatment with whole family involved -Standing 6 inches away from speaker; standing one foot away from speaker, standing two feet away from speaker; standing four feet away from speaker -Hypotheses with each of these studies? (Hint, need to have dependent variables) -What does the “Between Groups” part of the design mean? -How do you decide which participant is given which condition? -How many participants needed for each study? One Independent Variable Between-Groups Design Why NOT do t-tests to compare the means of each group versus the other? p < .05 p < .10 p < .15 One Independent Variable Between-Groups Design Group A versus Control Group Group B versus Control Group Group C versus Control Group Group A versus Group BGroup A versus Group CGroup B versus Group CProbability of Type 1 Error Greatly Increased p < .05 p < .10 p < .15 p < .20 p < .25 p < .30 One Independent Variable Between-Groups Design -Instead Analysis of Variance is used AKA ANOVA -Instead of doing many comparisons with the same data over and over, you look at how the IV created variance instead of the normal distribution. -Variance review: -Remember Normal (Bell shaped) curve? -If you did Nothing to a population, you just assessed them, then (theoretically), you would see a Normal Distribution.-This is equivalent to your Control group. One Independent Variable Between-Groups Design -How much of the variance is due to ordinary, everyday variance among the people who are yoursubjects/participants? Versus -How much is due to the fact that you applied the IV to one or more groups of subjects? -Mathematically, this calls for a RATIO (Hint: that’s your F-ratio) One Independent Variable Between-Groups Design -Partitioning the variance: -How much of the variance here is due to just regular old normal individual differences among people -How much is due to MY application of a new condition to them? (Just how much power and control do I have on other people? BWAHAHAHA!) One Independent Variable Between-Groups Design Within group variance: Individual differences (like what) Sometimes thought of as “error” Between group variance: Application of the IV One Independent Variable- Between-Groups DesignF Ratio F=Between Group Variance Within Group Variance -ANOVA produces an F ratio -The higher the F-ratio, the higher the probability that there is significant variation due to the IV. One Independent Variable Between-Groups Design -How do you know if your F-ratio is significantly different from chance? -Use your degrees of freedom and your pre- set alpha level -Alpha conventionally .05 (could be higher)-df between groups = k – 1 (k=number of gps) -df within groups = k (n -1) (n = # sub per gp) -df total for experiment = N – 1 (N=total S’s) One Independent Variable Between-Groups Design -F ratio reported in a Results section of a paper is written so that any researcher can see the F, degrees of freedom, and the probability level. -Example: F(3, 278) =100.32, p < .01 Review -Instead of just one treatment versus a control group or one treatment versus another, we often use designs that compare several levels of the same factor -One Independent Variable Between-Groups Design -Control group assumed to be what “normal” population would look like -Other groups: application of IV-How to analyze? Partition the variance -Calculate an F-ratio (between group variance versus within group variance) -Using an ANOVA -Report your F ratio with your df and alpha level Lecture: 3/18/15One Independent Variable AKA Repeated Measures Design 1) Post test only Between Groups Design (design that you give between groups- 2 groups- one control/one experiment)-Test the effects of Anxiety versus No Anxiety on Musical Performance -30 participants, 15 in each group (random) -Group 1: Induce Anxiety-Group 2: Do not Induce Anxiety-Test: ability to play music (count mistakes) (everyone plays music and you count the mistakes they make)-IV? DV? (mistakes) Stat test? (t-test) because 2 groups and you are comparing the means) Advantages- you get how a person changesDisadvantages- people might have different levels of anxiety2) One Independent Variable- Between Groups Design -Test the effects of Anxiety versus No Anxiety on Musical Performance -What’s different from Post Test Only Between Groups Design?- you’d have more than one levelof anxiety (3 or 4 groups)-Stat test?- analysis of variance/one way ANOVA-Advantages? Better variability is accounted for-Disadvantages?- requires many subjects to complete 3) Pretest-Posttest only Design -Test the effects of Anxiety versus No Anxiety on Musical Performance -What would it look like? (No anxiety and count mistakes, then induce anxiety and count mistakes)- History, test effects, and maturation are threats to internal validity-Advantage: each participant acts as own control (Why is this good?) -Disadvantage: three threats to Internal Validity New: One Independent Variable- Repeated Measures Design What would it look like? -Have person play piano under different circumstances Even with just two levels, how does it improve on Pretest-Posttest only design? -Takes out order effect (which is basis of maturation problems)What would it look like with three levels?-Use one way repeated measures analysis of variance (ANOVA)Advantages of using? -Each person acts as their own controlDisadvantages: 1. Can’t use if one application of IV forever changes participant -Might have to use Between Subjects design 2. Order effects could be confounding -Counterbalance orders (determine all) -Randomly present orders Lecture: 3/20/15Example 1: -One Independent Variable- Repeated Measures Design The study reported here compares the usability of three types of message input format: Abbreviations, Numbers and Free-Form as alternatives for a
View Full Document