Class 9 Bias 2020 Sp1
Get an idea, select a topic or problem, review literature
State the research question, write hypotheses, define variables
Select study design, sample parameters, measurement method
Determine data storage, analysis, budget, gain IRB approval
Conduct pilot study, analyze results, make adjustments
Bias occurs due to limitations in study design or conduct
Uncertainty in results due to potential biases
Difficulty in determining the degree of bias influence
Study planning, execution, and analysis stages
Experimenter bias, design bias, selection/sampling bias
Consideration of variables, operationalization, recruitment
Procedural bias, measurement bias, errors in data collection
Importance of consistent procedures, training, and calibration
Methods/procedures, data storage, statistical tests
Reporting all data, avoiding attrition bias and reporting bias
P-Hacking for statistically significant results
Wansink's study on choice architecture for eating habits
Lack of publication for studies showing no effect
Negative impact on scientific progress and resources
Brian Wansink's controversial food research
Downfall and questions raised about scientific integrity
Brian Wansink: high-profile researcher at Cornell University
World-renowned eating behavior expert
Keynote speaker, author of bestselling books
Engages in questionable research practices
Admitted to these practices himself
Increasing number of papers retracted or corrected
Still a professor at Cornell University and director of the Food and Brand Lab
Brian Wansink had 15 studies retracted
Cautionary tale on bad incentives in science
Media reports may misrepresent study results
Importance of evaluating evidence behind scientific claims
Sensationalized headlines misrepresent research findings
Unrepresentative samples lead to biased conclusions
Conflicts of interest can influence research outcomes
Correlation does not imply causation
Unsupported conclusions and problems with sample size
Tools for assessing bias in scientific studies
Cochrane Collaboration for systematic reviews
Tools like Downs & Black Tool for assessing different types of studies
Designed to assess all types of studies in about 20 minutes
Assesses several types of bias in studies
System for developing and presenting evidence summaries
Domains include Risk of Bias
Get an idea, select a topic or problem, review literature
State the research question, write hypotheses, define variables
Select study design, sample parameters, measurement method
Determine data storage, analysis, budget, gain IRB approval
Conduct pilot study, analyze results, make adjustments
Bias occurs due to limitations in study design or conduct
Uncertainty in results due to potential biases
Difficulty in determining the degree of bias influence
Study planning, execution, and analysis stages
Experimenter bias, design bias, selection/sampling bias
Consideration of variables, operationalization, recruitment
Procedural bias, measurement bias, errors in data collection
Importance of consistent procedures, training, and calibration
Methods/procedures, data storage, statistical tests
Reporting all data, avoiding attrition bias and reporting bias
P-Hacking for statistically significant results
Wansink's study on choice architecture for eating habits
Lack of publication for studies showing no effect
Negative impact on scientific progress and resources
Brian Wansink's controversial food research
Downfall and questions raised about scientific integrity
Brian Wansink: high-profile researcher at Cornell University
World-renowned eating behavior expert
Keynote speaker, author of bestselling books
Engages in questionable research practices
Admitted to these practices himself
Increasing number of papers retracted or corrected
Still a professor at Cornell University and director of the Food and Brand Lab
Brian Wansink had 15 studies retracted
Cautionary tale on bad incentives in science
Media reports may misrepresent study results
Importance of evaluating evidence behind scientific claims
Sensationalized headlines misrepresent research findings
Unrepresentative samples lead to biased conclusions
Conflicts of interest can influence research outcomes
Correlation does not imply causation
Unsupported conclusions and problems with sample size
Tools for assessing bias in scientific studies
Cochrane Collaboration for systematic reviews
Tools like Downs & Black Tool for assessing different types of studies
Designed to assess all types of studies in about 20 minutes
Assesses several types of bias in studies
System for developing and presenting evidence summaries
Domains include Risk of Bias
