SOC200H1 Study Guide - Final Guide: Hawthorne Effect, Internal Validity, Dependent And Independent Variables

83 views22 pages
Published on 25 Dec 2015
Lecture 6
"The ideal"
What are experiments?
Research methods that carefully control variables of interest to study their
Each experiment has at least one treatment variable that the researcher
Types of groups in Experiments
Experimental group
Group of subjects which is administered the stimulus
Control Group
Group of subjects which is administered no stimulus
Should resemble the experimental group
Stimulus or Intervention
Independent variable
Takes dichotomous form
◦ Eect
dependent variable which the stimulus has some impact
◦ Placebo
Inactive treatment (substance) designed to resemble the stimulus being
Assumptions built into paradigms are things that we cannot test
Method wouldn't make sense if this assumption wasn't true
What assumptions do experimental designs make?
Causation is the important thing to study
Concepts can be meaningfully quanti/ed and measures
The causal eects of individual variables can be isolated
What is the eect of this variable subtracting out the eects of every
other variable
Causation as a counterfactual model
If in a particular case, X caused Y, then if X had not happened or not been true
for that person, Y would have been dierent for that person
BUT the counterfactual is imaginary and therefore unobserved
To say that X causes Y we need to observe
People who are the same in all the things we measures did Y under
circumstance X and didn't do Y when the only thing dierent thing we
measures was that X was missing
That cannot happen!
When spuriousness attacks!
The thing that makes people who do X dierent from people who don't
Experiments /ght spuriousness
• Randomization
Our best weapon against spuriousness
Having some both happen and not happen to the same person is
Randomly assigning subjects to the experimental and control groups
• Matching
Unlock document

This preview shows pages 1-3 of the document.
Unlock all 22 pages and 3 million more documents.

Already have an account? Log in
Purposely assigning equal numbers of individuals with similar
characteristics )race, gender, level of education) to both groups
Problem: di;cult to know in advance which individual attributes are relevant in
terms of their possible in<uence on the dependent variable
Pre-experimental designs
Like experiments since researchers administer the intervene
Not like experiments because it cannot show causation
One-shot case rudy
One group of subjects gets the stimulus and post-test
◦ Problems
Without a pretest we can't tell whether the D.V. was the same before the
stimulus was introduces
No variation in independent variable - can't even get correlation
One group pre-& post-test design
Pre test added before the stimulus is introduced
Compare pre- and post-test to see if levels of DV have changed
◦ Problem
Maybe something else happened to aect the DV
Static Group Comparison
Experimental group gets stimulus and control group doesn't
Administer post-tests and compare DV levels for experimental group and
control group
◦ Problem
Maybe they were dierent to begin with
The classical Experiment
Experimental group and control group
A pretest is administered to both
The stimulus is administered to the experimental group
A post-test is given to both
The Post-test is pampered to the pre-test in the experimental group
Post-tests are compared between groups
What kinds of /ndings can be made with experiments?
Can come as close as possible to showing causation, at least in the experiment
Internal Validity
"Internal" means within the experiment
Is what seems to have happened (X caused Y) really what happened?
If anything other than the stimulus aected the DV, that's internal validity
◦ Sources
▪ History
▪ Maturation
▪ Testing
Hawthorne eect
▪ Instrumentation
Statistical Regression
Selection Biases
▪ Mortality
You drop out of the study
Causal time-order
Diusion or imitation treatment
Unlock document

This preview shows pages 1-3 of the document.
Unlock all 22 pages and 3 million more documents.

Already have an account? Log in
▪ Compensation
Compensatory rivalry
What Can't we /nd with experiments?
◦ Process
Meaning, perception or interpretation as experiences by participants in the
social setting or process
Complex causation - the kind that doesn't try to isolate the eects of individual
What happens outside the lab? - External Validity
External Validity
Internal validity shows that what we think happened in the experiment really
Possible threat to external validity is the interaction between the testing
situation and the experimental stimulus
Maybe the stimulus would have had a dierent eect with no pretest
◦ Solution
No pretest
Post test only control group design
Solomon Four-Group Project
Explanations on Blackboard
Four groups
2 get pre tests (classic experiment) and 2 do not, all four get
the post-test
Arti/cial settings
The lab itself is a octal setting
What if the phenomenon in question operates dierently in one social
setting (the lane) than it does in the real world
◦ Solution
▪ Quasi-Experiments
Researchers take advantage of some event to test its eect on
some particular variables
E.g. natural disaster, war, disease outbreak
Some naturally occurring circumstances or institutions divide
people into groups that duplicate treatment and control groups
Field Experiments
Experiments with true randomization carried out in the real world
instead of the lab
Maintaining true randomization can be di;cult
"Mortality" becomes a threat again once the experiment moves
into the real world
Strengths and Weaknesses of Experimental Design
◦ Strengths
Isolation of the experimental variable over time
Experimental can be replicated several times using dierent groups of
◦ Weaknesses
Arti/ciality of laboratory setting
Unlock document

This preview shows pages 1-3 of the document.
Unlock all 22 pages and 3 million more documents.

Already have an account? Log in

Get OneClass Grade+

Unlimited access to all notes and study guides.

YearlyMost Popular
75% OFF
Single doc


You will be charged $119.76 upfront and auto renewed at the end of each cycle. You may cancel anytime under Payment Settings. For more information, see our Terms and Privacy.
Payments are encrypted using 256-bit SSL. Powered by Stripe.