Research bias and ethics

Nature of bias

• preference or predisposition to favour a particular outcome

• bias is the systematic distortion of research conclusions

  • inadvertant or intentional

Bias in research

• we tend to seek corraboration of out preconceptions

  • science blinds research because of the expectations of findings and belief in theories

Researcher bias

• researchers have results or theories that they like and want to find

• uses procedures that lead to conclusiosn supporting the prefered explanation

  • procedures are more likely to be significant

Sources of bias in the research process

• problem selection

• sampling design

• funding

• data collection

• data analysis

• reporting of findings

• use of findings

Bias in problem selection

• selecting a problem is depending on researcher

  • what they think is important

• also bias towards quantitative approach

• choices of subject can show the values of the researcher

selection bias

• error due to the systematic differences in characteristics between those who were selected and those who were not selected for a study

  • methods used to select participants

  • factors that influence a persons descision to participate

  • losses to follow-up in a cohort study

Sampling design

• results can be distorted by choosing to study populations with known slants

  • povety, reduced nutritional status etc

• sexism may also be present

Funding descisions

• funding increased for hot topic areas

• SSHRC - main social science funding

• CIHR - funds health research

• CFDR - priority funding areas

Data collection effects

• experimenter effect

• information bias

• expectancy

• demand characteristics

Experimenter effect

• tendency to produce findings that consistent with expeirmenter expectations

Information Bias

• flaw in measuring exposure or outcome data that results in different accuracy between groups

Expectancy effect

• anticipation of particular research results

• leading to distortion of results in the direction of expectations

Demand characteristics

• distortion indroduced during data collection when respondents give answers they believe that the researcher is expecting

Data analysis error

• random error

• systematic error

• data messaging

Random error

• inconsistencies that enter into the coding process but have no pattern

• noise not bias

• measurement error etc

systematic error

• error which distorts the data in one particular direction

data messaging

• practice of playing with the data until the analysis produces the strongest association that you can identify

  • this is portential source of bias

Reporting findings

• reporting findingds that are not significant cannoy

• cannot report only significant results

Minimizing bias

• educating society that most research supports some interest

• distinguish advocacy from pure research

  • note funding sources

• give competing explanations a chance

• invest into discomfirming relationships

• be skeptical

• read literature cautiously

• specify data analysis procedures in advance

  • avoid playing with data or data dredging

• be sensitive to outcome preferences

• do not disclose hypothesis to subjects

  • reduced expectancy bias

• be accepting of all responses

• check for random and systematic errors

Nuremberg code

• legal principles adopted internationally after nazis

what requires ethical review

• human subjects

• human remains

• secondary human data use

• naturalistic observation of humans

• research with animals

• research with biohazards

When do you not need ethics

• when research is using publicly available data

• quality assurance

Ethical issues in social research - conflicting pressures

• persuit of scientific knowledge

  • perfection of research design and importance of precise measurement

• rights of those being studied

  • stresses the rights of the respondents

3 main principles of belmont report

• respect for human dignity

• beneficience

• justice and inclusiveness

core

• respect for persons

• concern for welfare

• justice

coercion

• unethical to force people to participate in a study

equipoise

• balance of two experimental conditions

• conditions have to be equivalent in order to randomly assign participants to them

Minimal risk

• if potential subjects can be reasonably expected to regard the probability and magnitude of possuble harms implied by participation to be no greater than those in every day life

Ethics review boards

• three canadian agencies developed the tri-council policy 2 guidelines for ethics

  • social science and humanities research council

  • natural sciences and engineering research council

  • Canadian institutes of health research

Sponsored research areas of conflict

• sponsors may try to interfer with research

• sponsor could suppress findings

• sponsor may be hidden

Sharing benefits

• if an experimental trail shows that one condition is more beneficial than another, the group who did not receive the condition should be exposed to it after the study

confidentiality

• maintained when anything is learned about participants is held in the strictest confidence

• information may need to be disguised when necessary to protect anonymity

  • minimizaton of how many people see and handle the data

Informed consent - information sheets

• sheet includes

• purpose, design, who can participate, who is researher, what will they have todo?

• possible risks

• possible benefits

• compensation

• confidentiality

• summary and contact sheet

Informed consent forms

• should be read and signed to ensure that they understand risks and are willing to participate

• outline the confidentiality, risks, benefits and that they have the option to leave the study

• parent and child form for under 18 years old