Outcomes Pre Intermediate Answers

[Evelyn Normington]

Itis quite clear that one should not control for post-treatment variables / intermediate outcomes when the goal is causal inference, but I wasn't sure if the same advice should hold when one's goal is to build a model for prediction.

Here is some context for my question: I'm trying to build a model that predicts if a college student will earn a bachelor's degree within 6 years of high school graduation using a large observational data set. I have data on students' high school variables (HS GPA, test scores, number activities participated in, etc.), some data on the students' college experiences (delayed enrollment in college, full-time / part-time status, transferred within two years of enrolling), as well as data on the characteristics of the college they attend. In other words, I have student level and institutional level data. I would account for the nesting of students within a particular institution.

Some have told me that the student level data on college experiences are intermediate outcomes and I shouldn't include them in the model. It isn't clear to me if I should / could include the college experience variables (which could be considered intermediate outcomes) in the predictive model, and if I do include them, how they should be treated.

I'd strongly discourage including those intermediate outcomes, for the sake of interpreting the parameters. Let's translate "predicted difference in the response of two individuals that differ by one unit on the regressor in question and that have the same value on all other regressors" into the terms of the study question: if you include the intermediate outcomes in the model, you're only comparing students who delayed enrollment to others who delayed enrollment, and comparing students who went to similar colleges to each other. If students with lower high school GPA are also more likely to delay college enrollment, and students who go to private colleges have higher GPAs, a model that includes both variables will only be useful for predicting a student finishing college based on their GPA if you also already know what kind of college they went to. Is that the prediction you need?

To develop this systematic map, we will search two bibliographic databases (including 11 indices) and 67 organization websites, backward citation chase from 39 existing evidence syntheses, and solicit information from key informants. All searches will be conducted in English and encompass subtropical and tropical terrestrial ecosystems (forests, grasslands, mangroves, agricultural areas). Search results will be screened at title and abstract, and full text levels, recording both the number of excluded articles and reasons for exclusion. Key meta-data from included articles will be coded and reported in a narrative review that will summarize trends in the evidence base, assess gaps in knowledge, and provide insights for policy, practice, and research. The data from this systematic map will be made open access.

Addressing climate change will also require clear and explicit consideration of impacts on human society and natural ecosystems. In general, the impacts of climate change are felt most acutely by individuals, communities, and countries who contribute the least to global emissions and are often the most vulnerable due to their location or history of colonialist rule and exploitation. Thus, NCS that are also targeted at increasing the resilience and adaptive capacities of ecosystems and communities are key to mitigating the ongoing and future impacts of climate change. NCS may provide important co-benefits in the form of adaptation, biodiversity maintenance, and other essential provisioning, regulating and cultural ecosystem services, which are fundamental for achieving the Sustainable Development Goals alongside the goals of the Paris Climate Agreement [9, 13]. For example, climate-smart agricultural practices aim to reduce GHG emissions while also enhancing the adaptive capacity of farmers to cope with the ongoing impacts of climate change [14, 15]. However, NCS, like other types of conservation and natural resource management interventions can have unintended and negative consequences for local communities and ecosystems [16, 17]. For example, some area protection and restoration activities have ignored practices of free, prior, and informed consent and disregarded customary rights of Indigenous Peoples and Local Communities (IPLCs) which has resulted in inequitable and potentially detrimental impacts on human well-being, and ultimately on ecosystems.

To slow down or halt climate change impacts and improve and maintain the well-being of nature and people, we need to understand the ability of different types of actions to achieve climate change mitigation outcomes. Nevertheless, the extent of evidence about how much change in land-based mitigation outcomes (e.g. additionally avoided CO2-equivalent emissions or enhanced carbon sequestration, or their respective land use/land cover proxies) can be attributed to specific interventions (i.e. policies, programs, and projects) is often not rigorously evaluated. Moreover, estimates of potential sequestration of different land-use and interventions from stand-alone modeling or estimation studies have met considerable debate (e.g. [18, 19]). There have been several systematic studies and meta-analyses characterizing the evidence base about the link between various types of NCS interventions and climate mitigation outcomes. An evidence gap map from the International Initiative for Impact Evaluation (3ie) on the impact of land use change and forestry programs on GHG emissions and food security has helped illuminate the state of evidence in this topic, taking into account various conservation interventions (e.g. protected areas, community-based conservation, payments for environmental services) [20]. A recent update of this evidence gap map showed that the evaluation of forest conservation outcomes has considerably grown in recent years, but that the evidence base remains insufficient across most forest conservation intervention types, and notably as it relates to climate mitigation outcomes [21]. Several global meta-analyses comparing the effectiveness of various approaches to ecosystem restoration (e.g. passive restoration, assisted natural regeneration, and active plantation), and other approaches following the principles of Forest Landscape Restoration (FLR)) also highlighted that evidence exists that some interventions have been effective in enhancing carbon sequestration, but require attention to improving the evidence base, notably by more rigorously disentangling the net impact of specific restoration interventions from confounding factors and site location bias (see e.g. [22,23,24]). An evidence gap map looking specifically at the environmental impacts of agroforestry also found that most studies employ study designs that are not well-suited for broader causal inference [25]. As well, a multi-sites pantropical study measured the climate mitigations outcomes resulting from reduced-impact logging in several projects located in 7 tropical countries [26]. Recently, a systematic map highlighted that the evidence base for climate-smart agriculture in East and Southern Africa was skewed towards just a few commodities and geographies [27].

The objective of this systematic map is to identify, map, and describe the evidence base surrounding the impacts of NCS interventions on climate change mitigation outcomes (and/or related LULCC impacts) in tropical and subtropical terrestrial ecosystems.

What are the extent and distribution of reviews and meta-analyses that examine links between NCS interventions and climate change mitigation outcomes, intermediate outcomes, biodiversity/ecosystem, or human well-being outcomes in tropical and subtropical forests, grasslands, and agricultural systems?

What are the extent and distribution of articles that examine climate change mitigation outcomes while also examining co-impacts on intermediate outcomes, biodiversity/ecosystems, and/or human well-being outcomes?

What are the extent and distribution of articles that examine different mechanisms through which NCS operate? (e.g. governance, monetary and non-monetary incentives, capacity development, policies and regulations)?

Lastly, we will compare the distribution of studies focused on the co-impacts of NCS interventions and any of the climate change mitigation, biodiversity/ecosystem, and/or human well-being outcomes. This is intended to help understand and characterize the nature of existing knowledge gaps on co-impacts (see Framework Development section below).

(Tropical and Subtropical Coniferous Forests; Tropical and Subtropical Dry Broadleaf Forests; Tropical and Subtropical Grasslands, Savannas and Shrublands; Tropical and Subtropical Moist Broadleaf Forests; Mangroves)

Climate change mitigation outcomes (in terms of equivalent metric tons of CO2) or environmental outcomes directly related to climate change mitigation (e.g. changes to forest and land cover, avoided land conversion, land use change)

The framework for this systematic map was developed through a series of discussions and meetings involving the synthesis team and advisory group. During this process, the overall scope of the project was determined and refined. The framework used in this study reflects a synthesis of existing conceptual models and causal theories on the links between NCS actions and climate change mitigation targets (e.g. [2, 20, 28, 29]) within a unified, generic theory of change (Fig. 1).

In this mapping work, we have chosen to focus on how these interventions link to changes in GHG emissions, carbon storage, and/or carbon sequestration through changes in land and management practices and changes in LULCC patterns. Within this conceptualization, we consider changes to land and management practices (e.g. adoption) and LULCC as intermediate outcomes along the pathway to climate change mitigation.

3a8082e126