Presentation on theme: "Research Methods 3010 Ethics & Research Variables Chapter 5 & 6."— Presentation transcript:
Research Methods 3010 Ethics & Research Variables Chapter 5 & 6
Ethics Guidelines Principles to assist you in deciding which goals are most important in reconciling conflicting values. For example: Research vs. Individual’s rights
Three Areas of Ethical Guidelines Relationship between society and science Professional issues Treatment of research participants.
Three Areas of Ethical Guidelines Relationship between society and science: Government and communities values have an influence through setting of funding priorities Researchers are shaped by their cultural background (e.g., gender research; health services research). These values can potentially bias research.
Three Areas of Ethical Guidelines Professional Issues: Reporting fraudulent results Failure to execute the design as published Partial publications (e.g., publishing several publications on related aspects of the data set). Publishing same data in different journals to widen the distribution (e.g., different journals likely reach different audiences). Re-using research data for another study (e.g., DNA analysis; reviewing videotapes for new question). Honorary authors
Three Areas of Ethical Guidelines Treatment of research subjects: Little Albert
Three Areas of Ethical Guidelines Treatment of research subjects: Unfortunately some research cannot be done without inflicting harm (e.g., deception, some humiliation). Dilemma – Potential for research gain Vs. costs to the research participant. How do we handle ethical dilemmas?
Handling of Ethical Dilemmas Seek consultation from: Other scientists Students/patients Lay individuals Have the institutional review board (IRB) review your proposal.
Ethic’s Decision Plane Model A B C D High Low High A & D studies easy to decide. C & B studies difficult to decide about. Cost of Doing Benefit of Doing
IRBs Not always correct. McGill study on psychological treatment for depression in post myocardial infarction patients.
American Psychological Association Ethical Guidelines First published in 1953. Developed by a committee Chaired by Nicholas Hobbs. Ten general principles.
American Psychological Association Ethical Guidelines Principle A: If in doubt about ethical issues, seek expert consultation. Most IRBs do not leave the decision up to you. Rather all research must undergo an ethical review. Even with consultation, the ultimate responsibility rests with the investigator.
American Psychological Association Ethical Guidelines Principle B: Determine if participant in your study is at “risk” or “minimal risk.” Examples of minimal risk?
American Psychological Association Ethical Guidelines Principle C: Investigator takes responsibility to ensure high standards of ethics; and Ensures that his/her staff perform in ethical manner. Having more staff involved in your research does not dilute your responsibility but multiplies it.
American Psychological Association Ethical Guidelines Principle D: Research clarifies the obligations and responsibilities of the research team and participants with the participant before subjecting subject to the experiment. Informed consent. Challenge where deception is involved. Challenge with those unable to comprehend issues (e.g., children, mentally ill).
American Psychological Association Ethical Guidelines Principle E, when deception is involved: Determine if deception is really justified. Consider alternatives to deception. Debrief participants as soon after as possible.
Types of Deception Active deception – by commission Deliberately misleads the participant Passive deception – by omission Information is withheld.
APA Guidelines for When Deception is Acceptable 1. Research question is very important and cannot be addressed any other way. 2. Once subjects are debriefed they will suffer no loss of confidence in the integrity of the investigators or of others involved.
APA Guidelines for When Deception is Acceptable 3. Participants are allowed to withdraw at any time without penalty. 4. Researchers take full responsibility for detecting and removing stressful after-effects of the experiment.
Deception When handled well, research suggests that participants usually do not mind having been mislead and understand its value. Reactions to deception less positive when observed behaviours are very private or negative.
American Psychological Association Ethical Guidelines Principle F: Researchers respects and ensures the person’s right to decline participation without penalty. Cannot coerce participation (even with heavy inducements).
American Psychological Association Ethical Guidelines Principle G: Researcher protects participant from psychological or physical discomfort, harm, etc. If risk of harm unavoidable, informed consent necessary. If risk of harm unavoidable, harm from study must be less for participant that potential harm if he/she does not participate (e.g., treatment study). Great benefit likely and participant is fully informed.
American Psychological Association Ethical Guidelines Principle H: Debrief participants after experiment if fully informed consent was not possible. Ensure that there are no lasting ill effects from participating (remember Little Albert). What are some of the challenges to debriefing?
American Psychological Association Ethical Guidelines Principle I: If there are untoward ill effects from participating, the researcher must detect and remove these effects when possible.
American Psychological Association Ethical Guidelines Principle J: Information obtained on the participant is confidential unless otherwise agree upon in advance. What happens if the courts demand this information from you? What are your options? Are there limits to confidentiality?
Animal Research Now very controversial but in many instances, essential for research. Animal laboratories are now kept hidden and secure. Set of ethical guidelines also govern animal research.
Animal Research Need to ensure high quality care. Experiment needs to be highly justified. If animals are used for educational purposes, use as few animals as possible, or better yet, find alternative means of demonstration (e.g., video).
Animal Research Animals should be lawfully purchased from qualified supplier or bred in researcher’s facility. Human care should be taken (e.g., unacceptable to use a paralytic drug or muscle relaxants during surgery without general anesthetic). IRB board reviews animal research.
What is a variable? Any characteristic of the organism, environment, or experimental situation that can vary. Independent variable- the characteristic that the researcher manipulates. Dependent variable – what is measured to assess the influence of the independent variable (e.g., reaction time, anxiety).
Types of Variables Discrete, nominal, mutually exclusive Sick or well; young or old; male or female Continuous Variable forms a continuum, matter of degree. Qualitative Variables that vary in kind (e.g., diagnosis) Quantitative Vary in amount (e.g., degree of anxiety).
Types of Variation Presence vs. absence (e.g., Drug vs. placebo Cancer vs. no cancer Amount of variation Vary dose, intensity, etc. Type of variable under study Labeled students as “trouble-makers” vs. studious and see how teacher later rates these students.
How do you control variation? Just saying you have varied the independent variable does not mean you have achieved the desired aim. Create two conditions one that is characterized as anxiety provoking, the other as calm.
Types of Manipulation Instructional manipulation Event manipulation
Instructional Manipulation Through instructions, participants are lead to expect either one type of condition (e.g., anxiety provoking) or another (e.g., excitement).
Event Manipulation Vary the condition that the participant has (e.g., cancer vs. no cancer) or is exposed to (e.g., drug vs. placebo). Event manipulation is usually better than instructional manipulation because it is more realistic and so less vulnerable to misinterpretation.
Individual Difference Manipulation Different medical conditions Different personality characteristics Limitations in this manipulation are Cohort bias Third variable problem
Operational Definition In some instances, you need to operationalize the independent variable. How would operationalize abstract constructs like an attitude, emotion, or even learning?
Construct Validity of the Independent Variable Extent to which the abstract construct or conceptual variable of interest can be inferred from the operation definition. Does the experimental operations represent the construct (conceptual variable) I had in mind? E.g., Does this anxiety questionnaire actually measure anxiety?
Steps to Establishing Construct Validity 1. Clearly define the construct. 2. Determine if it relates to a predicted outcome in the expected direction (e.g., if anxious, is there avoidance of the feared stimulus, heightened physiological arousal, expression of nervousness (convergent validity)?
Steps to Establishing Construct Validity 3. Demonstrate that the construct is not related to a different construct (e.g., sadness) (discriminate validity). 4. Does the operationalization of the construct (e.g., questionnaire) look “right” (face validity).
Number of Independent Variables Additional variables that help explain (predict) the variation in the dependent variable. All of the likely or known independent variables that explain mortality outcome in cancer patients. You may need to control for some of these variables before testing for an association with the independent variable of primary interest.
CIDS: Baseline Depression and Mortality up to 3.8 Years Follow-up
Number of Independent Variables Variables that interact such that the independent variable has a different influence on the outcome based on different levels of another independent variable. Aspirin may influence cardiac risk differently in men and women.
Psychosocial Predictors of Sudden Cardiac Death (BDI>10) 0 200 400 600 800 1.000.950.900.85 Survival in days Placebo, BDI <10 Placebo, BDI >10 AMIO, BDI <10 AMIO, BDI >10 ProportionSurviving
Dependent Variable Behavioural variable to measure the effect of the variation of the independent variable. The measure needs to be sensitive to the influence of the independent variable. Measures the direction and magnitude of the effect of the independent variable.
Dependent Variable Potential problems with the dependent variable: Insensitive to the influence of the independent variable Not valid or reliable Response bias – participant does not give a true answer (e.g., social desirability; yes- responding; defensive; lie)
Response Bias Measure of the dependent behaviour that requires a commitment on the part of the participant probably best measure for overcoming response most types of response biases: Not just say who you like best but commit to having that person as a roommate for a year.
Dependent Variable: Challenges The construct you want to measure is not directly observable (e.g., shyness) and so your dependent measure is often only approximates the measurement of the whole construct.
Dependent Variable Validity Is the dependent measuring what you think it is measuring? Reliability Consistency or stability of the response over time or over different measures of the same construct.
Reliability Increases when you take more than one measure of the construct and take an average of all of these measures (e.g., blood pressure is most reliable if you assess it 3 times per occasion over 5 occasions.)
Constraints on Reliability Not always possible to get repeated measures over time. Some measures cannot be repeated because by repeating them you change what is really being measured. With respect to the latter, you may have repeat the study on a new sample to see if the phenomenon is reproducible.
Validity Convergent validity Measuring same things as other measures of the constructs (often expressed as correlating highly with these other measures). Divergent validity Variable is not related to conceptually different constructs.
Number of Dependent Variables Often more than one dependent measure of the construct is assessed because they contribute to the understanding of the phenomenon.