The IPCC

• provides comprehensive, objective, and transparent assesments of the scientific literature on climate change
• reports are published every 6-7 years → serve as trusted sources of information across climate science, impacts and adaptation, and mitigation
• the summaries for policymakers are approved by national governments → vital for international climate negotiations - tools for holding governments accoutnable
• Working groups of the IPCC
  • WGI - physical science basis
  • WII Impacts, Adaptation and Vulnerability
  • WGII Mitigation of Climate Change
• early IPCC reports had 100s and 1000s of scientific studies → more recent assesment report had more than 300, 000 studies published on climate change
  • reports are bigger and involve more authors and comprise more pages and contain more references - not kept pace with the growth in the literature
  • IPCC citing a small proportion of available relevent literature (because there is so much of it) → what do they cite and how do they get selected becomes important ++++

Evidence Synthesis Methods

more resources here → https://www.lshtm.ac.uk/research/centres/centre-evaluation/evidence-synthesis

type of research method that allows researchers to bring together all relevant information on a research question → useful to identify gaps in knowledge and establish an evidence base

important stakeholders: Centre for Reviews and Dissemination, York → do health-related systematic reviews + have high equality systematic reviews and associated economic evaluations, Health Evidence Network → publications for policy makers yay my fave fs, there’s more like africa evidence network (this is cool++) that are using systemic review evidence in public health policy

• formal methods for selecting and assessing evidence on a research question

  • more used in medical domain → important decisions (life or death outcomes) are not often based on single studies or on normal literature reviews (this is where one expert or a lot of them select and summarise a literature with no formal producedures for how evidence is selected or assessed).

  • evidence synthesis methods prioritise comprehensiveness in the evidence base assessed + transparency in how it’s identified selected and assessed

    

• synthesis methods

  • there’s a spectrum from more aggregative (trying to bring together a set of experimental findings) to configurative (trying to bring together diverse evidence to provide an overall picture - like a jigsaw puzzle i think)
  • methods
    • content analysis - coding and categorisation of qualitative / quanttative data
      • knowledge building and theory generating qualative systems review
      • systematic review and content analysis of bullying + cyberbullying strats
    • framework - organized structured data extraction and organisation
      • attitudes to walking and cycling
      • framework based synthesis for conducting reviews of qualative studies
    • etc. (check resource at the top)

proposed idea: machine-learning assisted evidence synthesis

problem: screening potentially relevant studies or coding studies by the type of evidence they have is repetetive + time consuming

as the number of papers grows → larger challenges to the goal of producing comprehensive assesments

research is in written texts → so we can use nlp

uses of ML

• identifying relevant studies - binary classifier to predict whether documents without labels will be relevant or not