Impact analysis methodology

The ImpactScan is ValuePolitics' analytical format for decisions, measures and policy options. It can inform a decision, accompany its implementation or assess its effects retrospectively.

We begin by defining the question, geographical scope, time horizon and baseline. The central question is what would probably have happened without the measure under consideration. This provides the basis for attributing effects to a decision.

We assess data sources for provenance, timeliness, geographical and substantive relevance, and their suitability for the case. We give priority to current official data from the area studied. Where these are insufficient, we identify the approximation used and its limitations.

Observed data, assumptions, model calculations and conclusions remain distinct. Where an effect cannot be reliably quantified, we describe it qualitatively or refrain from reporting a monetary figure.

We map the path from a decision to its consequences. This includes direct and indirect, intended and unintended effects, as well as their distribution across affected groups.

Where the data support it, we value impacts in monetary terms. We use established economic and social science approaches such as willingness to pay, revealed preferences and shadow pricing. Money makes different indicators understandable and comparable with financial metrics.

We vary costs, reach, duration and effect size across scenarios. Ranges show how robust a result is to changes in assumptions.

For every ImpactScan, we document sources, data reference dates, assumptions, data preparation, calculations and limitations. Readers should be able to trace results from the findings through the model to the source and subject them to expert scrutiny.

ValuePolitics retains professional responsibility for the methodology, valuation and publication.

For ImpactScans, we often work with data that companies and organisations provide from their own records, such as case numbers, programme data or internal metrics. We treat these data as confidential.

We use artificial intelligence where it makes our work more thorough and faster: researching publicly available sources, developing our analytical models and programming the analyses. For this, AI needs only the structure of your data. We therefore develop the analysis programs using a description of the data structure and specially created sample data.

The actual analysis of your original data then takes place separately in a local computing environment. Confidential data are not transmitted to external AI systems.

This separation also makes our work transparent: we disclose the calculation model, assumptions and uncertainties so that you can examine every result and continue using it independently.

The first published ImpactScan combines georeferenced accident data with the surroundings of schools in Saxony. It identifies spatially concentrated accident hotspots and prioritises them using modelled societal accident costs.

The study assesses a portfolio of locations. It does not replace site-specific planning: measures, feasibility, responsibility for road infrastructure and costs must be examined locally. The results therefore make both potential benefits and uncertainties transparent.