Measurement that holds up when someone checks it.
We design instruments, run community needs assessments, and evaluate programs for organizations serving families. The work is documented well enough that a skeptical reviewer can follow every decision we made and see why we made it.
We run the programs we study.
Nexus Point's research program is led by Kaylynn Rohde, the foundation's Executive Director. The program is built to grow, and we are actively looking for collaborators and affiliated researchers to build it out.
What we bring is graduate training in research methods, a completed master's thesis, poster presentations at the Midwestern Psychological Association, an open-science workflow built around preregistration on the Open Science Framework, and the uncommon position of being practitioner and researcher at once. We operate a resource directory and a mutual aid board, so we are not studying family resource-seeking from a distance. We are in it, and we measure our own work with the same instruments we would build for you.
Jewell Meagher, our treasurer, is how the work reaches people: recruitment, the community relationships across our counties, and the financial administration any funded project carries. She also reads every participant-facing document before it goes out and marks anything she had to read twice. A researcher already knows what they meant, which makes them the worst judge of whether a question is clear, and an unclear question measures confusion instead of the thing you meant to ask.
Then the collecting. Kaylynn builds those systems herself, front end through database, including our resource directory, a live application holding roughly 69,775 records across 3,163 counties along with the pipelines and deduplication behind it. So if your data has to come out of a system you already run, or arrive in a dashboard your board can actually read, or be collected on a phone in a waiting room, we can build that. That also means we know what it is like to be the organization on the other side of a funder's reporting requirement.
Then the analysis, in R, and in jamovi when a partner wants the same output without reading code. That choice matters more than a tool list suggests. Because the analysis is written as a script rather than clicked through menus, every step is recorded and reproducible. You get the script along with the report, so a new staff member, an incoming board, or a skeptical reviewer can rerun exactly what we ran and get exactly what we got. Menu-driven analysis cannot make that promise. When someone asks two years from now how a particular number was produced, there is no way to retrace it.
You can read the design, its rationale, and the limitation we deliberately accepted on our research page. Publishing our own weaknesses is a reasonable way for you to judge whether you want us handling your numbers.
If you are a researcher rather than a client, the same door is open. We are building the kind of program that brings people in, and early collaborators get to shape what it studies.
What we do, in full.
Click any of these to see the actual method, the deliverables, and what we would need from you.
A conversation
No charge, no proposal unless you ask for one
A conversation
No charge, no proposal unless you ask for one
Thirty to forty-five minutes. We ask what you are trying to find out, what you already collect, what your funder requires, and what has gone wrong the last time you tried to measure something.
What you leave with
- An honest read on whether measurement would actually help, or whether your problem is something else wearing a data costume
- The cheapest useful version of the work, named specifically
- If we are the wrong fit, who we think is a better one
We do this for free because it is how we learn what rural and small organizations actually need, and that shapes our own research agenda. It is not a sales call with a different name on it.
Who you would be talking to: Kaylynn Rohde, Executive Director. The person on the call is the person who would do the work, not a salesperson who hands you off afterward.
Survey and instrument design
Questions that measure the construct you actually care about
Survey and instrument design
Questions that measure the construct you actually care about
Most surveys measure something other than what their author intended. Items get written in an afternoon, response scales get chosen out of habit, and the problem only surfaces months later when the data cannot answer the question that prompted it. By then the collection window has closed.
How we build one
- Construct definition before any item is drafted. We write down precisely what you are measuring and how you would know it had changed. Most bad surveys are bad because this step was skipped.
- Existing measures first. If a validated instrument already exists for your construct, we adapt it rather than invent a rival. Reviewers trust an established scale, and a novel measure has to prove it is not an old one renamed.
- Item drafting with attention to framing. Behavioral wording rather than status wording, because "have you looked for help" and "have you needed help" get different answers from the same person on the same day. We check reading level and strip the jargon that makes people guess.
- Deliberate response scales. Number of points, whether there is a midpoint, and whether "does not apply to me" is distinguishable from a skipped question. These choices decide what analysis is possible later.
- Cognitive pretesting with five to eight people from your actual population, thinking aloud as they answer. This is the step that gets cut and it is the one that matters most. It catches misreadings that no amount of expert review will find.
- Pilot and item analysis. Response distributions, floor and ceiling effects, and the items people abandon partway through.
What you receive
- The final instrument, formatted for how you will actually field it
- A codebook and data dictionary, so the person analyzing it in two years knows what every variable means
- An item-by-item rationale memo, which is what you hand a funder or an IRB when they ask why you asked what you asked
- A written pretest summary, including the items we changed and why
What we need from you
Access to five to eight people from your target population for pretesting, and one person on your side who can answer questions about how the program actually runs.
Why us for this one: instrument design is the work we do on ourselves. We are building our own navigation burden instrument through exactly this process, and the reasoning behind every design decision, including a limitation we knowingly accepted in our preliminary questions, is published on our research page. You can inspect how we think before you buy it, which is not true of most people who will quote you for this. We also build the thing you field it with, so the instrument does not have to be bent to fit a survey platform's constraints.
Community needs assessment
For board planning, grant narratives, and hospital and health department requirements
Community needs assessment
For board planning, grant narratives, and hospital and health department requirements
A needs assessment that ends in a pile of findings has failed. The point is to finish with a ranked, defensible set of priorities that a board can act on and a funder can be shown.
How we run one
- Define the service area and population precisely, including who is deliberately out of scope. Vague boundaries produce numbers nobody can use.
- Secondary data first. American Community Survey, County Health Rankings, and state agency data get compiled before we field anything. We are not going to spend your primary collection budget rediscovering your county's median household income.
- Primary collection to fill the actual gaps. Usually a survey plus key informant interviews with people who see the population daily. Focus groups when the question warrants them and not by reflex.
- A written sampling plan that states honestly what kind of sample it is. If it is a convenience sample, the report says so and says what claims that does and does not support. Assessments that quietly imply representativeness are the ones that fall apart under review.
- Analysis. Descriptive statistics, cross-tabulations by the subgroups that matter to your decisions, and thematic coding for open-ended responses.
- Prioritization. Needs ranked against magnitude, severity, feasibility, and equity, with the criteria stated so anyone can see how a need got to the top.
What you receive
- The full report, written for the audience that has to act on it
- A short board-facing version, because nobody reads the long one in a meeting
- The cleaned, de-identified dataset, which belongs to you
- The analysis script, so any number in the report can be traced back and rerun
If you are a nonprofit hospital
IRS section 501(r)(3) requires a community health needs assessment every three years, adopted by an authorized governing body, made widely available to the public, and informed by input from persons representing the broad interests of the community, including those with public health knowledge. It has to be paired with an implementation strategy. We build to that standard, and we can serve as the community-organization input source for an assessment someone else is running.
Why us for this one: we built and maintain a resource directory spanning 3,163 counties across all 50 states and DC, covering utilities and financial help, childcare, food, diapers, maternal and infant health, healthcare, housing, and mental health. It is a relational database Kaylynn designed, populated, and maintains, with the SQL and Python behind it handling consolidation from multiple sources, fuzzy deduplication of near-identical records, and scoring. That is the same discipline as building a clean dataset for a study, and it is why we can take messy inputs from several agencies and return something analyzable. It also means we arrive in your county already holding a picture of what exists there, and you can go look at it right now rather than take our word for it.
Program evaluation
Process and outcome evaluation over a full program year
Program evaluation
Process and outcome evaluation over a full program year
Funders ask whether your program works. That is really two questions, and conflating them is the most common way evaluations go wrong.
How we structure it
- A logic model first. Inputs, activities, outputs, outcomes, and crucially the assumptions connecting them. Most programs have never written the assumptions down, and that is usually where your own staff quietly disagree with each other.
- Process and outcome evaluation kept separate. Process asks whether the program is being delivered as designed. Outcome asks whether anything changed. A program can fail on outcomes purely because it was never delivered as intended, and if you only measure outcomes you will never find that out.
- Indicators defined operationally. Written so that two different people measuring the same thing six months apart measure it the same way.
- Baseline before changes, which usually means moving faster than feels comfortable at the start.
- Fidelity monitoring through the year, so drift is caught while it can still be corrected.
- Analysis and reporting written into your funder's actual template rather than a generic academic format they then have to translate.
The part we will not soften
Without a comparison group, an evaluation can describe change but cannot attribute cause. If your participants improved, we can show you that they improved. We cannot show you that your program is why, and any evaluator who tells you otherwise is setting you up for a hard conversation at renewal. Where a credible comparison is feasible we will design for one. Where it is not, the report says plainly what the design can and cannot support.
What you receive
- A logic model and a written measurement plan
- A baseline report, an interim report, and a final report
- Your cleaned dataset and documentation, de-identified and yours to keep
- The analysis script, so every figure in the report can be traced back and rerun
Why us for this one: we work in a preregistered, open-science workflow, which means the analysis plan is written down before the data arrives. That is the single strongest protection against an evaluation that quietly reshapes itself to flatter the program, and it is why a funder should trust the result more than a self-reported number.
External evaluator on your grant
The bigger path, and it costs you nothing up front
External evaluator on your grant
The bigger path, and it costs you nothing up front
Most organizations realize they need an evaluator after the proposal is written, when there is no money left to pay one. Bring us in beforehand and the cost sits inside the award instead of coming out of funds you do not have.
How it works
- You tell us what you are applying for and when it is due.
- We help write the evaluation section of the proposal, which is frequently the weakest part of an otherwise strong application.
- We appear in your budget as the named external evaluator, with a scope and a figure.
- If the grant is funded, we do the work. If it is not funded, you owe us nothing.
What your budget can carry
The old rule of thumb put evaluation at 5 to 10 percent of a program budget. Current guidance runs higher, commonly 13 to 16 percent, and reviewers increasingly expect to see it. The numbers below are your grant dollars, not our fee:
- A $25,000 program leaves room for roughly $3,300 to $4,000 of evaluation
- A $50,000 program, roughly $6,500 to $8,000
- A $150,000 program, roughly $19,500 to $24,000
Even a small grant has room in it. That is the part most organizations do not discover until the proposal is already submitted and it is too late to add.
Why a funder cares that the evaluator is external
Self-reported outcomes get discounted, and reviewers know that an organization grading its own homework has every incentive to grade generously. A named outside evaluator with a documented method is worth points on the application before a single number is collected.
Scope of what we can hold: we serve as external evaluator or as a subawardee on someone else's award. We are not currently positioned to be principal investigator on a federal grant, and we will tell you that at the start rather than after you have built an application around us.
We price each job after we understand it.
The honest number depends on sample size, how much of the collection your own staff can absorb, and how fast you need it. Publishing a single figure would only mean quoting you for work we had not yet defined.
Tell us your budget in the first conversation and we will tell you straight whether it covers the work. If it does not, we will describe what a smaller version would still answer, and if we are simply the wrong fit for what you need, we will say that too. We work with organizations of very different sizes, so do not rule yourself out on an assumption.
What we do not do.
An evaluator who claims to do everything is not worth hiring. Here is where we are not the right call:
- We do not run randomized trials or make causal impact claims. If your funder requires proof that your program caused an outcome, you need a different shop, and we will say so early rather than late.
- We are not positioned to serve as principal investigator on a federal grant. As a subawardee or named external evaluator, yes.
- We will not write an evaluation that flatters the program. If the numbers are unflattering, that is what the report says, and you should want an evaluator who works that way.
- We have not been doing this for twenty years, and we price accordingly.
Tell us what you need to measure.
A couple of sentences about what you are trying to find out and when you need it is enough to start. We will come back with a scope, a price, and an honest read on whether it is worth doing at all.
Request a scope of work Or email the director directly