Partnership model

Outsourcing hires an outside expert on contract. The partnership model does something different: it builds a relationship between two teams who both want the evaluation to succeed, even without a formal contract.

The workbook's description: a research team — interested in adding to global evidence and pushing the boundaries of impact evaluation — partners with policy makers and programme implementers who want a good-quality impact evaluation set up in their own project. Both sides bring something the other needs.

A written agreement is always preferred, the workbook says, for clarity on roles and responsibilities. But a partnership can, in principle, work even without a formal contract, provided both sides commit to a clear evaluation plan.

This flexibility carries a real risk. Researchers may push for methods more rigorous than a policy team actually needs, so close alignment of objectives between the two teams is essential.

The workbook's real-world illustration of the partnership model comes from the World Bank (Chapter 7, section 7.2.2), through three named groups that provide funding and technical support to impact evaluations:

• DIME (Development Impact Evaluation)
• SIEF (Strategic Impact Evaluation Fund)
• GIL (Gender Innovation Lab)

Its worked example: in Côte d'Ivoire, a partnership between the World Bank, the Abdul Latif Jameel Poverty Action Lab (J-PAL) and the government evaluated a Youth Employment and Skills Development Project. The research team combined a World Bank team leader, international academics and local experts; the policy team combined specialists from the project implementing unit, the affiliated Ministry, and World Bank staff.

Key success factors, per the workbook: close alignment between the research team's and policy team's objectives, jointly designed evaluation strategies with clear roles, and a laid-down evaluation plan describing operational rules and potential risks — even where a formal contract is absent.

Illustrative, in the spirit of the workbook's Côte d'Ivoire example; figures are made up.

An Indian state government wants to evaluate a skilling scheme for 50,000 unemployed youth. Instead of hiring an external evaluator on contract, it partners with a university research team that wants to study skilling-scheme effectiveness for its own academic purposes.

The two sides agree an evaluation plan: the university gets access to programme data and can publish its findings; the government gets a rigorous impact evaluation using randomised assignment across 20 districts, partly self-funded through a research grant.

The risk surfaces midway: the university wants to extend the evaluation timeline by 6 months to test an additional research question, which the government's own reporting deadline cannot accommodate — exactly the tension the workbook warns a partnership model can create when the two teams' objectives are not tightly aligned.

Chapter 7 (Social Impact Assessment Reporting), section 7.2.2 (Partnership Model), sets out this organisational model for structuring evaluation, with the World Bank's DIME, SIEF and GIL groups, and the Côte d'Ivoire Youth Employment case, as its worked illustration. Expect a question naming the World Bank groups or identifying the partnership model from a description of a research-team/policy-team collaboration.

• The partnership model does not require a formal contract, unlike outsourcing — though the workbook still prefers a written agreement.
• DIME, SIEF and GIL are World Bank groups cited as examples of partnership-model funders.
• The workbook's own risk: a partnership can suffer if the research team's push for rigour is not closely aligned with the policy team's immediate local objectives.
• Do not confuse the partnership model with the DMEO's outsourcing example (a contractual, arm's-length relationship) or the Fully Integrated Model (continuous, embedded assessment).