Question

What are the differences between structural and statistical models. Give an example of each that is being used to model the future effects of climate change. Which model is more realistic and why?

Answer 1

• Structural equation modeling (SEM) includes a diverse set of mathematical models, computer algorithms, and statistical methods that fit networks of constructs to data. Structural equation models are often used to assess unobservable latent variables. Latent variables are those, which are not directly observed, but are inferred from other variables that can be observed. Latent variable models are used in many disciplines like  psychology, demography, economics, medicine, physics, machine learning, econometrics, management and social science.

Structural models can be used for modelling the impact of climate changes and its effects, for example on crop yield and overall food security. The structural approach captures key knowledge across multiple areas of expertise. Linking crop and economic models can improves analysis of climate change impacts.

• A statistical model is a mathematical model that has a set of statistical assumptions concerning the generation of some sample data and similar data from a larger population. A statistical model represents, often in considerably idealized form, the data-generating process. In mathematical terms, a statistical model is usually thought of as a pair (S, P), where S is the set of possible observations, i.e. the sample space, andP  is a set of probability distributions on S.

Statistical models can be used to model sensitivities to changes in temperature, precipitation, carbon dioxide (CO2), and ozone, as it has been carried out in a few experiments. Statistical models generally require fewer resources to produce robust estimates, especially when applied to crops beyond the major grains.

• In the structural approach, one approach that can be used is the IMPACT system of models. It links general circulation models (GCMs), crop simulation models, water models, and a global economic model in the International Model for Policy Analysis of Agricultural Commodities and Trade (IMPACT,  Robinson et al., 2015). The  IMPACT system of models can be used to run alternative scenarios for drought and heat tolerant crop varieties under two extreme future climate scenarios. The structural approach is formalized in the design of the IMPACT system of models. Importantly, the link between physiological/biological studies and crop models is made explicit in this system.
• A statistical model named regional climate model (RCM) can be used to predict future ambient ozone, and its impact on health due to climate change. In one experiment, this approach was applied to the 20-county Atlanta metropolitan area using RCM simulations from the North American Regional Climate Change Assessment Program. Future ozone levels and ozone-related excesses in asthma emergency department (ED) visits were examined for the period 2041-2070.
• As discussed earlier, structural approach captures key knowledge across multiple areas of expertise. To better inform decision-making, the structural approach needs to be further developed to improve post-processing of economic information on supply and demand to estimate effects on food security, nutrition, environmental impacts, and welfare. A structural model can still be more realistic, than statistical, as statistical is based on assumptions, and can provide more pessimistic predictions of climate change impact.