Download presentation
Presentation is loading. Please wait.
Published byClifton Mosley Modified over 9 years ago
1
Control in Experimentation & Achieving Constancy Chapters 7 & 8
2
Internal Validity This is the extent to which one can accurately state that the independent variable produced the desired effect. Could the observed effect be due to some factor other than the independent variable?
3
Confounding Variable What is a confounding variable?
4
Confounding Variable This is a variable that systematically varies with the independent variable.
5
Internal Validity Because confounding variables can obscure the effects of the independent variable, you need to control for confounds in order to achieve a high level of internal validity.
6
Control of Extraneous Variables Only need to control for the extraneous variables that are confounds. Ideal is to eliminate confounds but this is not always possible. If you can’t eliminate the confound, then ensure that it is constant across all levels of the independent variable.
7
Challenges to Achieving Constancy in the Confounding Variable Not all confounding variables can be accurately and precisely measured. Some confounds change throughout the experiment. You need to identify the confounds in order to know what needs to be held constant.
8
How do we achieve constancy? By controlling the extraneous variables that confound the assessment of the influence of the independent variable on the dependent variable.
9
Threats to Internal Validity Natural history Maturation Instrumentation Statistical regression
10
Origins of the Correlation Coefficient 64”65”66”67”68”69” 70”2455 69”235899 68”361012 2 67”71113141310 66”6811 86 65”346432 Children’s height Parent’s height Correlation between parent’s height and children’s height
11
Threats to Internal Validity - continued Selection Selection bias can interact with other threats to internal validity (e.g., maturation, history) thereby magnifying the problem of confounds.
12
Example of a Selection Bias Brady et al. (1958) tested the hypothesis that monkeys given control over pressing a lever to avoid a shock would be more likely to develop stomach ulcers (‘executive monkeys’) than monkeys yoked to them and who received same number of shocks (‘yoked monkeys).
13
Example of a Selection Bias: Brady et al. 1958 Results – ‘executive monkeys’ were significantly more like to die during the 23- day experiment than the ‘yoked monkeys’ and they were more likely to have stomach ulcers. Selection bias – monkeys were not randomized. Rather those who pressed the lever the quickest on a baseline trial were put in the ‘executive monkey’ condition.
14
Threats to Internal Validity - continued Attrition (mortality or drop-outs) Attrition bias can interact with some of the other threats to internal validity (e.g., selection bias) further confounding the effects of the other confounding variable.
15
Threats to Internal Validity Participant effects that need to be controlled Demand characteristics Positive self-presentation bias (most likely to immerge when participant believes that his/her true intensions, beliefs, or feelings are being assessed). Need to ensure constant participant perceptions through all phases of the study.
16
Threats to Internal Validity Two types of interaction effects between the motive for positive self-presentation and experimental condition: Intertreatment interaction Intratreatment interaction
17
Threats to Internal Validity Experimenter effects that need to be controlled: Experimenter’s desire for certain experimental effects can influence his/her behaviour to the participant, thereby biasing the participant’s response (expectancy effect). Experimenter may unintentionally record data in a way to support the hypothesis. Experimenter may unintentionally misinterpret the results in a way that supports the hypothesis.
18
Threats to Internal Validity Experimenter variables that need to be controlled: Biosocial attributes Psychosocial attributes Situational factors
19
Threats to Internal Validity Magnitude of the expectancy effects: About 1/3 of studies demonstrate some degree of an experimenter/participant expectancy effect. The expectancy effect is often greater than the effect of the experimental condition (e.g., treatment effect).
20
Threats to Internal Validity Sequencing effects (e.g., carry-over effects). Participant sophistication
21
How to Achieve Constancy Across the Independent Variable
22
Control of the Effects of Extraneous Variables Appropriate experimental design Statistical methods (e.g., analysis of covariance) Incorporate control techniques into the experimental design.
23
Control Techniques Randomization Helps ensure that extraneous variables that could affect outcome are evenly distributed amongst the experimental conditions. Random selection Random allocation
24
Random Selection Each person in the population of interest has an equal chance of being selected and selecting one person does not affect the selection of another. What are the challenges to random selection?
25
Random Allocation Most effective way of ensuring that influential extraneous variables are balanced amongst experimental conditions. What are challenges to randomization?
26
Block Randomization Each block contains all conditions of the experiment in a randomized order. E, C, C, E C, E, C, E E, E, C, C Experimental Group N = 6 Control Group N = 6
27
Matching Increases sensitivity in the study by ensuring that experimental conditions are homogeneous with respect to important confounds. Holding extraneous variables constant Building the extraneous variable into the design Yoked control Equating participants
28
Building Extraneous Variable into the Design
29
Matching Increases sensitivity in the study by ensuring that experimental conditions are homogeneous with respect to important confounds. Holding extraneous variables constant Building the extraneous variable into the design Yoked control Equating participants
30
Matching Increases sensitivity in the study by ensuring that experimental conditions are homogeneous with respect to important confounds. Holding extraneous variables constant Building the extraneous variable into the design Yoked control Equating participants
31
Matching by Equating Subjects Precision control Matching precisely on a number of confounding variables Frequency distribution control Determine the frequency distribution of the extraneous variable in one sample (e.g., mean and standard deviation of age) and select subjects for the other group who establishes a similar frequency distribution in the second group.
32
Matching is Often Used When: 1. Small N and so randomization is risky and might yield unequal groups on influential extraneous variables. 2. Matching variable is expected to be correlated with the dependent variable and so exert an effect on it (confound). 3. There is a way to measure participants on the matching variable.
33
Control Techniques: Counterbalancing Counterbalance on the sequence of exposure to the experimental condition/task. Control over order effect Control of carry-over effects
34
Control Techniques: Intrasubject Counterbalancing The ABBA technique to control for sequence effects Each subject are exposed to the experimental condition first in one order (AB) and then in the other order (BA). This is sometimes referred to as a within subject design or repeated measures design.
35
Control Techniques: Intragroup Counterbalancing Groups of participants rather than individuals are counterbalanced. Complete counterbalancing is where all sequences and orders are represented. Incomplete counterbalancing technique is the most common intragroup counterbalancing technique.
36
Example of a Partial Counterbalancing Technique – Latin Square Design ABFCED BCADFE CDBEAF DECFBA EFDACB FAEBDC
37
Control of Participant Effects Need to ensure that participants in each experimental condition have identical perceptions about the experiment, except for the one variable being manipulated. Double-blind placebo model Deception Independent measurement of the dependent variable
38
Control of Participant Interpretation Experimental manipulation check by asking the participant after the experiment is completed what he/she thought the experiment was about. Concurrent verbal inquiry is similar except participant is asked his/her opinion at the end of each experimental trial. Think-aloud technique is also similar except participant is simply directed to speak all of his/her thoughts aloud.
39
Control of Experimenter Effects Control over recording errors Use multiple recorders. Ensure the recorders are ‘blind’ to the experimental hypothesis. Use electronic recording device.
40
Control of Experimenter Effects Control of experimenter attribute errors Hold the experimenter attributes across treatment conditions (e.g., same therapist, same gender). Use multiple experimenters. Control of experimenter expectancy error Blind technique Partial blind technique
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.