Title | Chapter 11 Textbook Notes |
---|---|
Author | Samantha Elia |
Course | Research Design & Analysis II |
Institution | Sacred Heart University |
Pages | 4 |
File Size | 121.5 KB |
File Type | |
Total Downloads | 109 |
Total Views | 189 |
Chapter 11 Textbook Notes...
Chapter 11: Factorial Designs: 11.1: Introduction to Factorial Designs: In real-life situations, variables rarely exist in isolation. There is often a variety of variables acting and interacting simultaneously. - Factorial designs: research studies that include more than one independent variable (quasi-independent variable); used to examine more complex, real-life situations. - 2 factors can act together, creating unique conditions, different from either factor alone Experimental studies: involve manipulation of two or more independent variables - It is possible for factorial designs to involve variables such as gender & age, that are not manipulated and are therefore quasi-independent variables Factorial designs have: 2 independent variables; 2 x 2 matrix shows 4 different combinations of the variables, producing 4 treatment conditions to be examined. Factors: term for the independent variables when 2 or more independent variables are combined in a single study; denoted by letters (A,B,C…) Factorial Design: study including two or more factors Two-Factor Design: design described the number of its factors Single-Factor Design: design with only one independent variable Levels: number of values EX: 2 levels for one factor, 2 levels for the other factor = 2 x 2 (two by two) EX: 2 x 3 = 2 factorial design with 2 levels of first factor & 3 levels of second factor = 6 treatment conditions EX: 2 x 3 x 2 = three-factor design with 2, 3, & 2 levels of each of the factors for a total of 12 treatment conditions Advantage of factorial designs: creates a more realistic situation which can be obtained by examining a single factor in isolation 11.2: Main Effects & Interactions: Structure of a two-factor design can be represented by a matrix in which levels of one factor determine columns and levels of the second factor determine the rows; then the mean differences among the rows define the main effects for 1 factor, and the mean differences among the columns define the main effect for the 2nd factor. Each cell in matrix = specific combination of the factors (separate treatment condition) Main Effects: Main effect: difference between the 2 column means; mean differences among the columns determine the main effect for one factor. Statistical test determines whether mean difference is significant. The Interaction Between Factors: Some situations where effects of one factor are completely independent of the levels of the second factor. - Neither factor has a direct influence on the other Some situations where one factor does have a direct influence on the effect of a second factor = Interaction between factors (interaction): 2 factors, acting together, produce mean differences that are not explained by the main effects of the 2 factors. If the main effect for either factor applies equally across all levels of the 2nd factor, then the 2 factors are independent & there is no interaction.
Alternative Views of the Interactions between Factors: Different perspective on the concept of an interaction focuses on the notion of independence, as opposed to dependence, between the factors. Interaction: when the effects of one factor depend on the different levels of a second factor When two factor study results are graphed, lines that cross or converge (nonparallel) are an interaction. Constant distance between lines = no interaction Identifying Interactions: Do this by comparing the mean differences in any individual row (column) with the mean differences in other rows or columns. If size & direction of differences in 1 row/column are the same as corresponding differences in other rows/columns, then there is no interaction. Differences = interaction Interpreting Main Effects & Interactions: Mean differences between columns & rows = main effects in 2 factor study Extra mean differences between cells = interaction- must be evaluated by statistical hypothesis before they can be considered significant If analysis results in significant interaction, then main effects, whether significant or not, may present distorted view of the actual outcome Because each main effect is an average, it may not accurately represent any of indiv. effects used to compute average. Independence of Main Effects & Interactions: Two factor study allows researchers to evaluate 3 separate sets of mean differences; mean differences from main effect of factor A, then from factor B, then from the interaction between factors; all three are separate & completely independent. 11.3: Types of Factorial Designs & Analysis Between-Subjects and Within-Subjects Designs: Between-subjects: separate group of participants for each of the treatment conditions] Disadvantages of Between-subjects: can require a large number of participants; individual differences (characteristics that differ from one participant to another) can become confounding variables and can increase variance of the scores Advantages of Between-Subjects: completely avoids any problem from order effects because each score is completely independent of every other score Between-subjects is best suited to situations with a lot of participants, small individual differences, order effects are likely. Within-subjects: a single group of indiv. participates in all of the separate treatment conditions Disadvantages of Within-subjects: many treatment conditions mean participants might get tired and walk away before finishing (participant attrition); many treatment effects mean including potential for testing effects (fatigue, practice effects) which make it more difficult to counterbalance the design to control for order effects. Advantages of Within-Subjects: require only 1 group of participants; eliminate or greatly reduce problem w/individual differences Within-subjects is best suited for situation where individual difference is relatively large & little reason to expect order effects to be large & disruptive Mixed Designs: Within Subjects & Between Subjects:
Mixed study: factorial study that combines two different research designs. Common example is factorial study with one between-subjects factor and one within-subjects factor. Experimental & Nonexperimental or Quasi-Experimental Research Strategies: It is possible to construct factorial study that is purely experimental research design- both factors are true independent variables that are manipulated by the researcher It is also possible to construct a factorial study in which all factors are nonmanipulated, quasi-independent variables Combined Strategies: Experimental & Quasi-Experimental or Nonexperimental: Combined strategy: uses two diff research strategies in the same factorial design- one factor is true independent variable (experimental strategy) and one factor is quasiindependent variable (nonexperimental or quasi-experimental strategy) The second factor may be preexisting participant characteristic (age/gender) or factor may be time and how the diff treatment effects persist over time Pretest-Posttest Control Group Designs: Pretest-posttest design: involves 2 separate groups of participants with one group measured before & after receiving treatment & second group is also measured twice before/after treatment but does not receive any treatment between the 2 measurements Higher-Order Factorial Designs: More complex designs involving three or more factors; researcher evaluates main effects for each of the three factors as well as a set of two-way interactions ( AxB, BxC, AxC & then AxBxC as potential for three way interaction) Statistical Analysis of Factorial Designs: Depends whether the factors are between-subjects, within-subjects, or a mixture. Compute the mean for each treatment condition (cell) & use an analysis of variance (ANOVA) to evaluate the statistical significance of the mean diff Two-factor ANOVA: conducts three separate hypothesis tests: one each to evaluate the two main effects & one to evaluate the interaction. ANOVA conducted using a statistical computer program such as SPSS - Two between subjects factors = independent-measures two-factor ANOVA - Only one of factors is between-subjects = specify which it is & use mixed-design two-factor ANOVA - Two within-subjects factors = use repeated-measures two-factor ANOVA
11.4: Applications of Factorial Designs: Expanding & Replicating a Previous Study: Often, factorial designs are developed when researchers plan studies that are intended to build on previous research results. In a single study, researcher can replicate/expand on previous research Reducing Variance in BS Designs: Individual differences can result in large variance for the score within a treatment condition Solution is to use the specific variable as a second factor, creating a two-factor study. Reduce variance without sacrificing external validity Evaluating Order Effects in WS Designs:
Order effects can be serious problem in within-subjects; confounding variable that should be eliminated It is possible that treatments occur early in the order may influence a participants scores for treatments that occur later in the order. Using Order of Treatments as a Second Factor: Measure & evaluate order effects using counterbalancing which requires separate groups of participants with each group going through a set of treatments in a different order Counterbalance = half participants beginning with treatment I & then move to II; other half of participants begin with treatment II & then move to I. 3 possible outcomes that can occur: 1. No order effects: does not matter if a treatment is presented first or second; data shows a pattern w/no interaction- mean is the same in either case 2. Symmetrical order effects: in this situation, size of the treatment effect (I vs. II) depends on the order of treatments – the effect of 1 factor depends on the other factor; symmetrical = symmetrical interaction in data 3. Nonsymmetrical order effects: for example, diff treatment conditions may produce diff levels of fatigue or practice; in data, lines do not intersect at their midpoints and difference between groups in treatment I is much smaller than difference in treatment II. Nonsymmetrical produce lopsided (nonsymmetrical) interaction between treatments & orders. ...