This past weekend, we saw modelling for coronavirus fatalities challenged and dismantled as using outdated data.
There are fundamental methodological issues with epidemiological modelling. I am today releasing a report on them by Mike Hearn.
stevebaker.info/?p=18274
There are fundamental methodological issues with epidemiological modelling. I am today releasing a report on them by Mike Hearn.
stevebaker.info/?p=18274
These issues include:
1. A lack of public review: models often contain internally inconsistent and non-replicable numbers.
2. Poor characterisation of statistical uncertainty: uncertainty bounds are either not reported at all or have extremely wide ranges.
1. A lack of public review: models often contain internally inconsistent and non-replicable numbers.
2. Poor characterisation of statistical uncertainty: uncertainty bounds are either not reported at all or have extremely wide ranges.
3. Non-existent or circular model validation: some scientists have argued that few healthcare models can ever be validated against reality, yet they should still be used to make decisions. Research papers can pre-suppose their own conclusions.
4. Misleading press statements.
5. Excessive freedom in choosing input data: Researchers may freely select data and add assumptions without regard to quality.
6. Lack of cost/benefit analysis: quality adjusted life year analysis in epidemiological papers can be rare.
5. Excessive freedom in choosing input data: Researchers may freely select data and add assumptions without regard to quality.
6. Lack of cost/benefit analysis: quality adjusted life year analysis in epidemiological papers can be rare.
7. Silencing of disagreement: models that depart from the "accepted" viewpoint can be rejected for publication. Scientists who disagree can be prevented from speaking at academic conferences or in the media.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
