Quantitative content analysis
analytic methodCSDDD
Systematic, replicable coding of a document corpus against a pre-specified framework. In our benchmarks the corpus is a population of corporate disclosures and the framework is derived from legal requirements and recognised standards, so scores can be compared across companies, sectors and years.
ApplicationsRegulatory benchmarking at scale; peer comparison; disclosure quality where consistency across companies matters.
LimitsMeasures what companies disclose, not what they do. Scores are only as good as the codebook and the agreement between coders.
Pre-specified codebookKPI-based scoring frameworkDouble coding of a sample
- LkSG HREDD Performance Assessment (2026): 141 practices, 25 effectiveness metrics
- UK Modern Slavery Act: three studies (2018, 2019, 2022)
- Dodd-Frank §1502 filing evaluations, reporting years 2014–2016
- French Duty of Vigilance (134 companies, 42 KPIs); EU NFRD (Germany, Austria, Sweden); California TISC (n=1,504 and n=1,961)
Survey design
designCSDDD
Structured measurement of a defined population: cross-sectional surveys at one point in time, and population-based surveys with probability cluster sampling where a representative prevalence estimate is needed. Repeated waves or a panel move a survey up to the next rung.
ApplicationsPrevalence where no administrative baseline exists; what workers, communities and suppliers experience, know and trust.
LimitsSelf-report and non-response bias; sampling error grows with design effects in clustered samples. Sensitive topics need indirect questioning (see instruments).
Structured questionnaireProbability cluster samplingWeighting for design effects
- Workers in Fairtrade small producer organisations: 300 worker interviews across five countries
- SBTi assessment: parallel surveys of staff (n=74) and Advisory Group members (n=82)
- Cocoa sector oversight in Côte d’Ivoire and Ghana for the US Department of Labor (2008–2011): population-based surveys, 40+ clusters per country, evaluating compliance with the Harkin-Engel ProtocolDI associates
Econometric estimation and modelling
analytic method
Regression, extrapolation and equilibrium models that produce defensible estimates of phenomena that cannot be observed directly, with every assumption documented and confidence intervals reported.
ApplicationsEstimating the scale of hidden phenomena; projecting the effects of policy options.
LimitsEstimates inherit the assumptions of the model and the quality of the input data; they describe, they do not attribute.
Regression analysisExtrapolation from partial dataEquilibrium modelling
- Estimating the number of hired workers in Fairtrade cocoa cooperatives, Côte d’Ivoire
- Europe’s child labour footprint (€50 billion, 2019), using the Basu multiple equilibria model
Cost-benefit and cost-effectiveness analysis
decision-analytic methodCSDDD
What a measure costs per unit of outcome, where reallocating spend would do more good, and the economic costs and benefits of regulatory options. Built to be auditable and contestable.
ApplicationsPrioritising due diligence spend; value-for-money reviews; regulatory impact assessment.
LimitsDepends entirely on the outcome measures it is attached to and on how costs are attributed; results shift with the discount rate and the perspective chosen.
Costing frameworkCost-effectiveness ratiosValue-for-money criteria
- Europe’s child labour import footprint: €50 billion estimate (2019)
- Cost model of Dodd-Frank Section 1502 compliance (2011), submitted to the US Securities and Exchange CommissionDI associates
Value chain analysis
analytic frameworkCSDDD
A framework from the global value chains literature for mapping the actors, relationships, power and governance in a supply chain: its scope, who does what at what price and under what conditions, and how value and risk are distributed. Case studies, interviews and price data are gathered within it.
ApplicationsMapping supply chains beyond tier 1; finding where governance gaps and leverage lie.
LimitsA framework for organising evidence, not a test of it; findings are descriptive unless combined with a comparative or causal design.
Interviews across chain tiersPrice and margin dataContract and audit review
- Non-judicial grievance mechanisms: global mapping of the remedy ecosystem
- Rainforest Alliance HREDD ecosystem study
- Cocoa producer empowerment, western Côte d’Ivoire: 31 cooperatives and a 9-cooperative comparison group
Qualitative content and thematic analysis
analytic method
Systematic coding of media, policy and corporate texts into themes, developed inductively or from a deductive frame, following Braun and Clarke. Corpora are built by explicit rules, and AI-assisted extraction is validated by manual coding.
ApplicationsPublic and stakeholder perceptions; how an organisation or controversy is framed; questions of reputation and legitimacy.
LimitsInterpretive by design: themes are patterns of meaning, not frequencies. Rigour rests on a transparent corpus, a documented coding frame and a second coder, not on statistical inference.
Rule-based corpus constructionThematic codingAI-assisted extraction, manually validatedSentiment analysis
- SBTi assessment: 141 media articles from 58 outlets, with social media sentiment analysis