Date post: | 20-Dec-2015 |
Category: |
Documents |
View: | 221 times |
Download: | 1 times |
Chapter TenChapter TenResearch BiasResearch Bias
Winston Jackson and Norine Verberg
Methods: Doing Social Research, 4eMethods: Doing Social Research, 4e
2 © 2007 Pearson Education Canada
The Nature of BiasThe Nature of Bias
Research bias: a preference or predisposition to favour a particular outcome thus indicating a systematic distortion of research conclusions
may be inadvertent or intentional occurs in all disciplines and can influence all
phases of a projects
3 © 2007 Pearson Education Canada
Bias in Research...Bias in Research...
In everyday life, we tend to seek corroboration of our preconceptions Helps us make sense of a complicated world Affirms our pet theories
Science potentially blinds researchers because of the expectations of certain findings, or the belief in certain theories Like culture, science produces blinders e.g., view of women in development of
psychology
4 © 2007 Pearson Education Canada
View of Women in PsychologyView of Women in Psychology
Early research about women began with an assumption of women’s intellectual inferiority Instead of challenging that assumption,
researchers sought ways to understand it Early research focused on:
Gender differences in brain structure Role maternal instinct plays in maintaining
women in passive, subservient role Evidence that culture influences science
5 © 2007 Pearson Education Canada
Sources of BiasSources of Bias
Having ‘valid’ research results requires systematically testing alternative explanations Done to avoid researcher affect Researcher affect refers to the bias that results
from a researcher having fallen in love with some particular explanation or view of the world
May inadvertently use procedures that lead to conclusions supporting one’s preferred explanation or world view
6 © 2007 Pearson Education Canada
Sources of bias in the research process
Bias can enter the research process at any stage of the research cycle
Selection of the problem Sampling design Funding Data collection Data analysis Reporting the findings Use of the findings
7 © 2007 Pearson Education Canada
1. Selection of Problem1. Selection of Problem
Some topics are judged as more important by funding agencies, one’s peers, community Results in selection of variables conventionally
considered important There has been a bias toward quantitative
approaches
8 © 2007 Pearson Education Canada
2. Sampling
Some samples will have a known slant Study of attitudes toward capital punishment
done in a rural communities with a prison
Problem of sexism Androcentricity
presenting the world from an exclusively male perspective as if this perspective were universal
9 © 2007 Pearson Education Canada
2. Sampling (cont’d)
Overgeneralization When a study claims to study all people but
samples only or largely males Study of social class – exclusively male, but
generalized to the pop’n Related to overspecificity: when single-sex
terms are used to describe situations applicable to both sexes
“the nurse … she” ; “the doctor … he”
10 © 2007 Pearson Education Canada
2. Sampling (cont’d)
Gender insensitivity Ignoring gender as an important variable Familism – a form of gender insensitivity
treating the family unit as the unit of analysis rather than the individuals in the family or assuming something affects all family members in a similar way
11 © 2007 Pearson Education Canada
2. Sampling (cont’d)
Double standard Using different measures for women and
men Sexual dichotomism - treating the sexes as
discrete social and biological cohorts rather than as two cohorts with shared characteristics
12 © 2007 Pearson Education Canada
3. Funding3. Funding
Granting agencies claim to fund projects on the basis of merit, but are subject to bias Traditional areas have better funding Special funding available in “hot” areas Science and health research better funded
than social science and humanities research Research done in a social context
Peers and funding agencies value some topics and research methods more than others
13 © 2007 Pearson Education Canada
4. Data Collection4. Data Collection
Experimenter effect: a tendency to produce findings that are consistent with the experimenter’s expectations Rosenthal’s “smart rats” study showed that
research assistants’ knowledge of the hypothesis biased the results
Clever Hans – horse who could count horse took its cues from the audience
Called expectancy – results will be biased in the direction of the expected results
14 © 2007 Pearson Education Canada
4. Data Collection (cont’d)4. Data Collection (cont’d)
What does this mean? The social researcher should avoid specifying
the hypothesis of the study to either the respondent or the research assistant
Research participants who know or guess the expectations of the researcher are likely to respond in the way that conforms to those perceived expectations – helpful subjects These distortions are called demand
characteristics
15 © 2007 Pearson Education Canada
5. Data Analysis
Coding research results is subject to two types of error: random and systematic error
Random error refers to inconsistencies that enter into the coding process but which display no systematic pattern Not threatening to the results of the study
Systematic errors refer to errors that distort the data in one particular direction e.g, how to deal with missing data Influences (biases) the results in one direction
16 © 2007 Pearson Education Canada
5. Data Analysis (cont’d)
Data massaging – researchers may manipulate the coding of the variables to influence results
Hunting for results – new computer technology makes it easy to test various hypotheses with different operational definitions for the variables
Nothing inherently wrong with “exploring” the data however, social scientists need to assess whether coding influenced results and report data massaging
17 © 2007 Pearson Education Canada
6. Reporting of findings
Journals prefer to publish positive results (results that support the hypothesis)
Some non-scientific considerations influence whether results are published Are findings culturally acceptable?
Insensitivity to minorities, sexism, etc. Are findings acceptable to one’s peers?
Corporate interests? E.g., medical researchers must ensure they are
not affiliated with a pharmaceutical company
18 © 2007 Pearson Education Canada
7. The use of findings
The findings of science are powerful tools Scientific evidence taken seriously by
courts, politicians, the media, the public Unfortunately, the literature is easily
misrepresented One may present results that support their
position, ignore contrary results “Scientist-for-hire” – may present results
desired by the person paying for their services
19 © 2007 Pearson Education Canada
A Perspective: Understanding Bias
Research is social behaviour Expectations of others to be met, norms of
behaviour to be followed, and findings to be anticipated
The social component of science is frequently at odds with the fundamental canons of science Science as practice is neither value-free nor
wholly objective Impressive achievements of science offset by
gaps between the ideal and practice
20 © 2007 Pearson Education Canada
Myths and Realities
FPO Table 10.1 Myths and realities of the social science, page 296
21 © 2007 Pearson Education Canada
Rules for Minimizing Bias
1. Educate key players about problems of bias
2. Avoid sexism
3. Identify roles played by the research process
4. Eliminate bias in description phase of project
5. In exploratory research, let disconfirmation be your guide (test alternative hypotheses)
6. Policy recommendations should be identified as value-based
7. Be skeptical of research findings
22 © 2007 Pearson Education Canada
Rules (cont’d)
8. Read literature cautiously
9. Distinguish advocacy from pure research
10. Orient research to disconfirmation
11. Use theory to generate testable hypotheses
12. Be sensitive to your own outcome preferences
13. Do not disclose hypotheses to subjects or assistants
14. Cover the attitudinal continuum
23 © 2007 Pearson Education Canada
Rules (cont’d)
15. Be accepting of all responses
16. Specify data analysis procedures in advance
17. Check for random and systematic errors
18. Report extent of data massaging
24 © 2007 Pearson Education Canada
Social science afflictions: humous look at the problem of bias in social research Theoretical rigor mortis Methodological paralysis True believer fever Good cause syndrome Guerilla raider’s syndrome Scientism ailment Anti-science fever Replicationitis No affliction affliction
25 © 2007 Pearson Education Canada
Afflictions: Theoretical Rigor Mortis
Characterized by a quest to do research which will support a favoured theory, those suffering from this affliction, when asked about it, protest their health, and continue to do research which is rigidly interpreted as supportive of a particular theory. The interest appears to be not in testing the theory's limits but to finding additional support for it.
Cure: Do a project which puts the favoured theory at risk. Publish the results. Use your real name.
26 © 2007 Pearson Education Canada
Afflictions: Methodological Paralysis
An inability to move to different research designs, measurement procedures, or analytical techniques. People who suffer from this affliction spend their time in locating problems that can be done using familiar design and measurement procedures.
Cure: Conduct a study using a design, measurements, and statistical procedures never used before. Enjoy it. Understand it.
.
27 © 2007 Pearson Education Canada
Afflictions: True Believer Fever
Whether an hypothesis or pet finding, these are the people always in search of evidence to support their particular view of the world; strongly challenge competing evidence, ruling it as irrelevant.
Cure: Write a paper enumerating the flaws in your perspective. Mean it.
28 © 2007 Pearson Education Canada
Afflictions: Good Cause Syndrome
These are the researchers with a cause: social science is used where its findings can be made to support the preferred view. Evidence to the contrary is dismissed as methodologically or theoretically flawed. These are the researchers who inevitably show up to help the underdog – or the overdog. They may also suffer from terminal liberalism, terminal political correctness, or terminal conservatism – all for a good cause, all in the name of virtue, all in the name of social science.
Cure: Write an essay defending cannibalism. Try to believe it.
29 © 2007 Pearson Education Canada
Afflictions: Guerrilla Raiders Syndrome
These patients typically do not do original research. Instead they fashion a career out of attacking a particular theoretical or methodological approach. This syndrome is often jointly found in those who suffer from True Believer Syndrome.
Cure: Do a primary research project. High cure rate.
30 © 2007 Pearson Education Canada
Afflictions: Scientism Ailment
The belief that if we follow the models of the physical sciences, that we will eventually understand social behaviour and it will become as predictable as are relations in engineered physical systems.
Cure: For each day during the next month, predict the weather 10 days in advance. Check performance.
31 © 2007 Pearson Education Canada
Afflictions: Anti-Science Fever
Characterized by the belief that social behaviour cannot be understood using orientations borrowed from the physical sciences. Anger is directed at the science practitioners who are favoured in our culture.
Cure: Marriage to someone with the Scientism Ailment is recommended. Check frequently: there is a danger of a double murder.
32 © 2007 Pearson Education Canada
Afflictions: Replicationitis
This terminal boredom condition is characterized by the researcher who keeps replicating the same study—with minor variation in samples, or in the variables involved. This is the person who makes a career out of studying the same set of variables. Incessantly.
Cure: Cut off funds. Get a hobby.
33 © 2007 Pearson Education Canada
Afflictions: No Affliction Affliction
These are the social scientists who think that their research is unbiased—untainted by brushes with their socio-cultural milieu.
Cure: not treatable since the patient is in strong denial. Pray.