The most fragile sentence in an advocacy document is often the strongest sounding one: “Research proves that this policy will solve the problem.” It may feel useful in a meeting, a press line, or a funding proposal. It can also make the whole argument easier to dismiss.
Research rarely speaks in campaign slogans. It can show patterns, estimate effects, compare options, reveal mechanisms, and identify risks. It can also be partial, contested, context-bound, or silent on the exact decision in front of you. Using evidence well means keeping that complexity visible without making the message unreadable.
This article stays in the research lane. It does not teach organizing strategy, campaign evaluation, or power mapping. Civic Futures Lab’s article cfl_09 is the better hand-off for judging whether campaign activity is becoming progress. Here, the focus is narrower: how to connect evidence to advocacy claims without overstating what the evidence can carry.
Start with the claim type
Before adding a citation, table, survey finding, or interview quote, identify what kind of claim you are making.
A descriptive claim says what is happening: “Families report long waiting times for appointments.” A comparative claim says how groups or places differ: “People with limited German report more difficulty understanding written notices.” A causal claim says one thing produces another: “This rule causes people to lose access.” A predictive claim says what may happen next: “Changing the opening hours could reduce missed appointments.” A normative claim says what should happen: “The city should fund language access.”
Research supports these claims differently. A community survey may support a descriptive claim about respondents. It may not support a causal claim about why an office behaves as it does. Administrative data may show a pattern over time. It may not explain how people experience that pattern. Interviews may reveal mechanisms and consequences. They do not, by themselves, estimate prevalence.
What is known: advocacy arguments become clearer when claim type and evidence type match. A policy brief can be strong without pretending every claim is causal.
What is contested: advocates sometimes worry that visible limits weaken urgency. Researchers sometimes understate moral urgency because they are trained to protect uncertainty. The better practice is to separate the two: be exact about the evidence and explicit about the values.
What we cannot say: evidence alone cannot decide what a society owes people. Research can inform that judgment. It cannot replace it.
Build an evidence chain, not a pile
A strong advocacy argument often needs a chain:
- A problem exists.
- The problem affects people in concrete ways.
- The current response is insufficient or unequal.
- A proposed change is plausible.
- The proposed change has costs, risks, and implementation conditions.
Different evidence can support each link. Survey data may show how often respondents report a barrier. Interviews may show why the barrier matters. Program evaluations from similar settings may show what kinds of interventions have helped elsewhere. Administrative rules may show where discretion exists. Community testimony may show consequences that are invisible in forms.
Do not make one piece of evidence do every job. A moving quote does not establish scale. A large dataset does not explain meaning by itself. A study from another country does not automatically predict results in a German municipality, but it may suggest mechanisms worth testing.
The practical question is not “Do we have perfect proof?” Most public decisions happen without perfect proof. The better question is “Is the demand proportionate to the evidence we have, and are we honest about what would need to be monitored?”
Use uncertainty as information
Uncertainty is not a defect to hide. It tells readers how confident they should be and where caution belongs.
Useful phrases include:
- “The available evidence suggests…”
- “Among respondents to our survey…”
- “Studies of similar interventions often find…”
- “This does not prove motive, but it shows a pattern that needs explanation.”
- “The evidence is stronger for the problem than for this specific solution.”
- “The likely effects depend on implementation.”
These phrases do not make an argument timid. They make it harder to misrepresent.
Avoid false precision. If your survey was shared through partner mailing lists, do not write “residents believe…” Write “respondents reached through partner networks reported…” If a qualitative project interviewed 20 people, do not write as if it measured prevalence. Write what interviews are good at: showing pathways, meanings, barriers, and consequences.
This answers “How can we make a strong argument without exaggerating?” Strength comes from fit. The claim should fit the method, the sample, the context, and the uncertainty.
Do not cherry-pick the convenient finding
Cherry-picking means selecting only the evidence that supports the preferred position while ignoring relevant evidence that complicates it. It can happen intentionally. It can also happen because teams are busy, angry, or already convinced.
Before publication, run a contrary-evidence check:
- What evidence points in a different direction?
- Are there studies or evaluations showing limited effect?
- Are there groups who might experience the proposal differently?
- Are there implementation failures in similar programs?
- Are there trade-offs that should be named?
Naming contrary evidence does not require giving bad-faith arguments equal weight. It means showing that the advocacy position has passed through reality.
For example, a team may argue for more local advice hours. Evidence might show unmet need, long waits, and positive user experiences. A contrary-evidence check might reveal that extended hours only help if childcare, transport, language access, and staff safety are addressed. The policy demand can become more precise: not simply “more hours,” but “more hours with the conditions that make them usable.”
Separate values from evidence
Many advocacy statements mix two sentences:
“This policy is ineffective.”
“This policy is unjust.”
They are related but not identical. A policy can be unjust even if it is administratively efficient. A policy can produce a measurable benefit and still distribute burdens unfairly. A policy can lack strong evaluation evidence and still be morally necessary as an emergency response.
Research can inform values by showing who is harmed, how often, through which mechanisms, and with what alternatives. But the value judgment should be named honestly.
Instead of writing, “The evidence proves the city must adopt our proposal,” consider:
“The evidence shows a persistent access problem, especially for people who cannot navigate German-only written procedures. Our position is that essential public services should be usable without depending on private translation help. For that reason, we recommend…”
This is both stronger and cleaner. The evidence supports the problem statement. The value principle explains why action is required. The policy demand follows from both.
When research is mixed, say what is mixed
“The evidence is mixed” is often too vague to help. Mixed how?
It may mean studies find different effects in different settings. It may mean one outcome improves while another does not. It may mean short-term effects are clearer than long-term effects. It may mean the evidence is strong for adults but weak for young people. It may mean the intervention works only with enough staff, trust, language access, or follow-up.
The response should match the uncertainty.
If evidence is strong on the problem but weaker on the solution, argue for a pilot with evaluation rather than a permanent system-wide rollout. If evidence is strong in similar settings but not the local one, argue for adaptation and monitoring. If evidence shows benefits and risks, argue for safeguards. If evidence is genuinely too thin, say the proposal is grounded in rights, lived experience, and plausible reasoning, and then ask for learning mechanisms.
What is known: mixed evidence is common in social policy because context matters. Human services are not tablets with the same chemical composition in every setting.
What is contested: decision-makers differ on how much evidence should be required before acting. Waiting for stronger evidence can prevent waste. It can also preserve harm when the current system is already failing.
What we cannot say: a mixed evidence base cannot be turned into certainty by confident wording. It can be turned into a more careful policy demand.
Treat contested research as a conversation, not a trapdoor
Research can be contested for good reasons: weak methods, changing contexts, narrow samples, poor measures, political misuse, or reasonable disagreement about interpretation. It can also be attacked in bad faith because its findings are inconvenient.
Do not respond to every challenge in the same way.
If the challenge is methodological, answer methodologically. Explain the sample, comparison, measure, or limitation. If the challenge is about interpretation, distinguish finding from conclusion. If the challenge is ideological, do not pretend a technical footnote will resolve a value conflict. If the challenge is bad faith, repeat the strongest defensible claim and avoid expanding into claims you cannot support.
In a German/EU context, many advocacy topics involve administrative data, community surveys, qualitative interviews, evaluation studies, and legal frameworks. Each has a different authority. None is a magic shield.
Make the policy ask traceable
A reader should be able to see how you got from evidence to demand.
Use a simple trace:
- Finding: what the evidence shows.
- Interpretation: what you think it means.
- Value principle: why it matters.
- Demand: what should change.
- Test: how to know whether the change helped.
For example:
- Finding: respondents report difficulty understanding official letters.
- Interpretation: written communication is a barrier, not merely an individual comprehension problem.
- Value principle: access to essential services should not depend on private language support.
- Demand: provide plain-language notices and qualified interpretation in defined service pathways.
- Test: monitor comprehension, appointment completion, complaints, and user feedback.
This structure also protects advocacy teams internally. If someone wants a sharper headline, the team can ask: which link in the chain supports it? If no link does, the sentence should change.
End with the strongest honest version
“Wie nutzt man Forschung in politischer Advocacy?” Use it as a disciplined bridge between the world as documented and the world being demanded.
The strongest honest version of an argument may sound like this:
“Our evidence does not prove that this single measure will solve the whole problem. It does show a repeated access barrier across the people we reached, supported by interview accounts and consistent with what service providers report. The proposed change is a proportionate response to that evidence, and it should be implemented with monitoring so unintended effects are visible.”
That is not a weak sentence. It tells the reader what is known, what is still uncertain, and what follows. It leaves room for democratic disagreement without handing opponents an easy factual correction. It also respects the people whose experiences became evidence by not stretching their contribution beyond what it can honestly say.