Lecture Four

NONEXPERIMENTAL RESEARCH DESIGNS

CASE STUDIES:
analyses of the experiences of a particular person, group, or phenomenon.
Often focus on people whose extraordinary experiences would be difficult or impossible to recreate in the laboratory. Example: Beethoven
They are usually designed to delineate, corroborate, or invalidate a particular theory or theories.
Example: Phineas Gage
This led to scientists understanding that damage to specific brain areas is in fact associated with specific kinds of psychological deficits.

WHAT MAKES A CASE STUDY SCIENTIFIC?

They are almost always a rich source of ideas for theory building and can be used for theory testing as well
Case studies do not usually lend themselves well to the use of operational definitions or statistical analyses.
Example; Sigmund Freud
case studies can fill in some important pieces of the complex puzzle of human experience.

SINGLE-VARIABLE RESEARCH:

designed to describe some specific property of people or of the social world.
These studies focus on one variable only!
May pave the way for more theoretically oriented multiple-variable studies that require a descriptive foundation.

Example: studies that provide info. about the prevalence of various mental illnesses.

form of single-variable research

refers to the scientific study of the causes of disease

RESEARCH ON PUBLIC OPINION:

important form of single-variable research
If you really want to know what people think about something, you should ask a random sample of most people
research designed to determine the attitudes and preferences of large groups of people such as voters or consumers
False consensus effect — shown that most people overestimate the proportion of other people whose attitudes and behaviors are similar to their own. Example: UCLA study — showed false consensus effects on 25 of a total of 28 questions.

MULTIPLE-VARIABLE RESEARCH:

often done when:
a topic is ethically sensitive,
when a variable is difficult or impossible to manipulate,
or when a researcher is especially motivated to conduct a study that is high in external validity.
Refers to research in which investigators examine naturally existing public records to test a theory or hypothesis.
Advantage: potential for high levels of external validity
can be at least as high in internal validity as many lab experiments —need to look out for potential confounds.
Disadvantage - can’t control the data.

OBSERVATIONAL RESEARCH:

the investigators record the real behavior of people in their natural environments
difference between observational and archival research is that in observational research the researchers are doing the observations.
Gives researchers some control over what they observe and also how they observe it.
Best kind of observational research is usually unobtrusive (researchers do not interfere in any way with people’s natural behavior, people do not realize they are being studied).
Observational studies are about the same as archival studies as far as external and internal validity goes.
Great potential for external validity — taking this right from the environment!

Survey and Interview Research:

Asking people about their thoughts, feelings, and behaviors.
Basic idea is to generate a fixed set of questions and pose them to a large group of people
problem with this type of research is that of reverse causality.
Inferences you draw really depend on how good the survey is.
To minimize threats to internal validity — include measures of potential confounds, write unambiguous questions, and take other reasonable precautions).

THE QUESTIONNAIRE:

use existing questionnaires - can compare results
Determine the purpose of the questionnaire - what do you want to accomplish?
Open-ended versus close-ended Questions

Open-ended questions

make it more likely that the questionnaire will discover something not anticipated by its designers,
harder to code, because the answers are in narrative form.
Need to categorize responses in some way in order to summarize the data.
May be more useful for smaller and preliminary investigations

CLOSE-ENDED QUESTIONS

easier to analyze and there are fewer off the wall answers
Respondents don't have to think as hard.
Tend to put words in the mouth of respondents and get responses they might not have otherwise gotten.
A respondent may misinterpret a question or a clerical error may be made.
Standardization of close-ended questions make them more suitable for large studies.
Sometimes the two types of questions are mixed. Permitting the data to be coded and analyzed easily, but giving some insights into the reasons why the respondents chose the alternative that they did.

WRITING SURVEYS

Address a single issue per item
Make items unambiguous - should be clear and precise
Avoid bias - i.e., "do you believe in killing unborn babies" vs. "should women be forced to bear unwanted children?"
Make alternatives clear - make sure the options are distinctly different from one another and that they cover all of the possibilities.
Tendency for social desirability: when respondents perceive one alternative as more socially acceptable than the other - word questions so that each alternative appears equally socially desirable.
Likert scale - allows you to measure magnitude of opinion, not just direction.

ADMINISTERING THE QUESTIONNAIRE

ADMINISTERING THE QUESTIONNAIRE

QUESTIONS TO ASK BEFORE YOU BELIEVE A POLL:

Many surveys —especially those designed to influence public opinion rather than just record it, do not produce accurate or useful info.
who carried out the survey? Do they have some agenda? Did they hire a professional sample survey firm?
What was the population? Whose opinions were being sought?
How was the sample selected? Random sampling?
How large was the sample? Sample size
What was the response rate? What percent of the original subjects actually provided information?
How were the subjects contacted? By telephone? Mail? Face to face interview?
When was the survey conducted? Was it just after some event that might have influenced opinion?
What were the exact questions asked?

Hosted by www.Geocities.ws