China
Africa
Explore chapters and articles related to this topic
Basic Instrumentation
Published in Vinayak Bairagi, Mousami V. Munot, Research Methodology, 2019
Pradeep B. Mane, Shobha S. Nikam
In natural experiments, experimenter does not manipulate variables deliberately as the variables change naturally. So, experimenter just measures the effect of something that is already happening. Natural experiments are used in observational studies. Natural experiments are employed as study designs when controlled experimentation is extremely difficult to implement or unethical example evaluating health impact due to particular disease. When controlled experimentation like Laboratory experiment is not possible to study the situation, in such cases, Natural experiments are often performed. Situations like policy changes, weather events and natural calamities create circumstances for natural experiments. Natural experiments are used most commonly in the fields of epidemiology, political science, psychology, and social science.
Adventures in Covariance
Published in Richard McElreath, Statistical Rethinking, 2020
Of course sometimes it won’t be possible to close all of the non-causal paths or rule of unobserved confounds. What can be done in that case? More than nothing. If you are lucky, there are ways to exploit a combination of natural experiments and clever modeling that allow causal inference even when non-causal paths cannot be closed.
Patterns of pre-crash behaviour in railway suicides and the effect of corridor fencing: a natural experiment in New South Wales
Published in International Journal of Injury Control and Safety Promotion, 2019
Shirley Gregor, Gary Beavan, Adrian Culbert, Priscilla Kan John, Nguyen Viet Ngo, Byron Keating, Ruonan Sun, Ibrahim Radwan
Apart from descriptive statistics, the data allowed for a natural experiment where cases from mostly fenced areas could be compared with geographic areas with less fencing. Natural experiments are observational studies and can be used in cases where controlled experiments are difficult to implement or unethical. An early well-known example was the attribution of an outbreak of cholera in Soho in 1854 to the nearest public water pump. Analysis of the incidence of cholera in this area compared with other geographic areas provided convincing evidence that the specific pump was responsible (see Freedman, 2009). Due to the nature of the data, non-parametric statistical tests were used (chi-square tests). A significance value of p < .05 was used unless otherwise noted.