Commercial Motor Vehicle Driver Fatigue, Long-Term Health, and Highway Safety Research Needs (2016) / Chapter Skim
Currently Skimming:
6 Research Methodology and Principles: Assessing Causality
6 Research Methodology and Principles: Assessing Causality
Pages 87-104
The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.
From page 87... ...
The theme of this chapter is that methods from the relatively new subdiscipline of causal inference encompass several design and analysis techniques that are helpful in separating out the impact of fatigue and other causal factors on crash risk and thereby determining the extent to which fatigue is causal. A primary question is the degree to which fatigue is a risk factor for highway crashes.
|
From page 88... ...
To draw proper inferences about crash causality, then, it is important to understand and control the various causal factors in making comparisons or assessments -- including those outside of one's interest, referred
|
From page 89... ...
Included in this section is a description of techniques that can be used at the design and analysis stages to support drawing causal inferences from observational data and extrapolating such inferences to similar population groups. DEFINITION OF CAUSAL EFFECT The definition of a causal effect applied in this chapter is that of Rubin (see Holland, 1986)
|
From page 90... ...
If one simply determines that a factor is associated with an outcome, however, it may be that the specific circumstances produced an apparent relationship that was actually a by-product of confounding factors related to treatment and outcomes. DRAWING INFERENCES AND STANDARDIZING CRASH COUNTS As one example of confounding and the challenges entailed in drawing causal inferences, it is common for those concerned with highway safety to plot crash counts by year to assess whether road safety is improving for some region.
|
From page 91... ...
. This nonlinearity is likely due to traffic density as an additional causal factor.
|
From page 92... ...
Traditional randomized controlled trials also are usually designed to have relatively homogeneous participants so that the treatment effect can more easily be measured. This homogeneity is achieved by having restrictive entry criteria.
|
From page 93... ...
These latter studies will often benefit from methods described in the next section for addressing the potential impacts of confounding factors. OBSERVATIONAL STUDIES Observational studies are basically surveys of what happened in the field (e.g., on the road)
|
From page 94... ...
Here the dependent variable would be the outcome of interest, the treatment indicator would be the primary explanatory variable of interest, and the remaining causal factors would be additional explanatory variables. The problem with this technique is that the assumption that each of the explanatory variables (or a transformation of a variable)
|
From page 95... ...
of interest and for the confounding causal factors. Then, one identifies controls that match a given case for the confounding factors from among drivers in the database who have not been involved in recent crashes.
|
From page 96... ...
(Of course, assessing whether a driver has texted is not always straightforward.) Analysis Methods for Observational Data This section describes some analytic methods that can be used to select subjects for analysis or to weight to achieve balance between a treatment and a control group on confounding factors.
|
From page 97... ...
The basic idea of the marginal structural model is to weight each observation to create a pseudopopulation in which the exposure is independent of the measured confounders. In such a pseudopopulation, one can regress the outcome on the exposure using a conventional regression model that does not include the measured confounders as covariates.
|
From page 98... ...
. In the fatigue alerting example, such an instrumental variable could be the indicator of a health insurance plan that provides free fatigue alerting devices to drivers.
|
From page 99... ...
. The introduction of such instrumental variables can be a useful design, but it can be difficult to identify an appropriate instrumental variable that is related strongly enough to the treatment of interest and does not have a direct effect on the outcome(s)
|
From page 100... ...
A key example in the present context is drawing inferences about commercial motor vehicle drivers when the relevant research is for passenger car drivers. When is it safe to make such an extrapolation?
|
From page 101... ...
The hope is that the findings can be translated to the administration of the same or a closely related treatment for a similar population. Criteria for determining the degree to which a study enables causal inference have been considered for many decades.
|
From page 102... ...
Fisher, proposed an alternative explanation: that there existed a factor that increased both the likelihood a person would use tobacco and the risk of contracting lung cancer, such as a genetic variant that made a person more likely to smoke and more likely to contract lung cancer through independent mechanisms. This alternative hypothesis was placed in doubt by a sensitivity analysis showing that if such a factor existed, it would need to have an association with smoking at least as great as the observed association between smoking and lung cancer, and the proposed factors, such as genetic variants, were unlikely to have such a strong association with smoking.
|
From page 103... ...
. These techniques, and additional ideas described here, have been applied in a number of policy areas and can be used to reduce the opportunity for confounding factors to influence outcomes when a study does not have a randomized controlled design.
|
Key Terms
This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More
information on Chapter Skim is available.