# Lesson 2: Needs Assessment and Planning Models

*Companion-podcast transcript, Sarah and Kiffer*

---

**Sarah:** Welcome back to Office Hours. I'm Sarah.

**Kiffer:** And I'm Kiffer. This is the second episode for Health Sciences eight twenty-six, Program Planning and Evaluation. Last time we talked about what evaluation is and introduced the program we'll follow all term. This week we go back to the moment before a program exists.

**Sarah:** Needs assessment and planning models. Why does an evaluation course spend a whole week on planning?

**Kiffer:** Because an evaluation judges a program against what it was meant to do and for whom, and if nobody wrote that down, the evaluator has to reconstruct it. Rossi, Lipsey and Henry treat the assessment of need as the first question an evaluator can ask. A program that runs beautifully for people who didn't need it has limited worth.

**Sarah:** So the lesson moves from need to priorities to design to objectives.

**Kiffer:** That's the order. Section one is about needs and assets, section two is about setting priorities, section three compares three planning frameworks, and section four is about goals, objectives and the program plan.

**Sarah:** Let's start with the word itself. What is a need?

**Kiffer:** The definition I use comes from Belle Ruth Witkin and James Altschuld. A need is a measurable gap between what is and what should be. The current state of some group of people, compared with a desired or acceptable state.

**Sarah:** That sounds simple enough.

**Kiffer:** It hides two important things. First, somebody has to say what should be, so every needs assessment contains a value judgement. Second, the gap is in the condition of people. Planners often write things like the region needs more community connectors. That sentence names a solution before anyone has described the problem.

**Sarah:** And the problem in that case would be what?

**Kiffer:** Something like a quarter of older adults in the region are lonely, and lonely people have no clear route from their family doctor to community support. Connectors are one possible response to that.

**Sarah:** The reading also separates need from want and demand.

**Kiffer:** Right. A want is something people would like, whether or not it changes their health. Demand is how much of a service people seek, and that depends on whether they know about it, whether they can afford it and whether they can get there. Service use depends on supply as well. Health care planners in the United Kingdom define need for care as the capacity to benefit from it, which ties need to the existence of something that works.

**Sarah:** Then we get to Bradshaw.

**Kiffer:** Jonathan Bradshaw, writing about social services in Britain in nineteen seventy-two, noticed that people used the word need in at least four different senses and rarely said which. Normative need is need defined by experts against a standard. Felt need is what people themselves say they lack. Expressed need is felt need that has turned into action, like a referral or a name on a waiting list. And comparative need appears when one group lacks a service that a similar group receives.

**Sarah:** Can you put those on Cedar Valley?

**Kiffer:** Sure. As a reminder, Cedar Valley is our fictional community connector program for lonely and isolated older adults, run by a fictional health authority in British Columbia. A regional survey of one thousand six hundred older adults found that three hundred and ninety-two, or twenty-four point five percent, scored six or higher on the three-item University of California, Los Angeles Loneliness Scale, which we usually call the UCLA scale. The cut-off of six is a standard the planners chose, so that's normative need. When the survey asked people directly whether they often felt lonely, fewer said yes, and that's felt need. The three hundred and twelve referrals in the first six months are expressed need. And the fact that rural older adults have higher rates and fewer groups nearby is comparative need.

**Sarah:** Why would fewer people say they're lonely when asked directly?

**Kiffer:** Stigma, mostly. Many people are reluctant to describe themselves as lonely.

**Sarah:** Bradshaw's types often disagree, then. Is that a problem?

**Kiffer:** The disagreement is useful. Bradshaw suggested that a need recognized under several definitions at once gives you a stronger case. When they disagree, the pattern tells you where to look. High normative need with few referrals suggests people don't know about the service, don't want to ask, or can't reach it. High demand with low normative need can mean the service is easy to use and attracts people outside the intended target.

**Sarah:** The section also talks about different populations.

**Kiffer:** Rossi and colleagues distinguish three. The population at risk has a meaningful chance of developing the problem, the population in need has it now, and the target population is the group the program intends to serve, defined by eligibility rules staff can apply.

**Sarah:** And eligibility rules can go wrong in two directions.

**Kiffer:** Over-inclusion lets in people who don't have the need, which spends money where it does little. Under-inclusion misses people who do. If you've taken epidemiology, you'll recognize that as specificity and sensitivity. Cedar Valley accepts a score of six or higher, or a clinician's judgement, and that second route buys sensitivity at some cost in specificity.

**Sarah:** Let's do the arithmetic. How many lonely older adults are there in Cedar Valley?

**Kiffer:** The region has about forty-six thousand adults aged sixty-five and older. Twenty-four point five percent of forty-six thousand is about eleven thousand two hundred and seventy. But a survey proportion has sampling error, so we report a range. The standard error is the square root of the proportion times one minus the proportion, divided by the sample size, which comes to about one point one percentage points. The ninety-five percent confidence interval runs from about twenty-two point four to twenty-six point six percent, so the number in need is somewhere between roughly ten thousand three hundred and twelve thousand two hundred and forty.

**Sarah:** And that range covers everything that could go wrong?

**Kiffer:** No, and that's the point I'd stress. The interval covers sampling error only. If lonely people were less likely to answer the survey, which is plausible, the estimate is too low. A household survey also misses people in long-term care. A good needs assessment states those assumptions alongside the number.

**Sarah:** The first-wave clinics referred three hundred and twelve people in six months. How does that compare?

**Kiffer:** The first-wave clinics serve about twenty-three thousand older adults, so roughly five thousand six hundred and thirty-five of their patients would meet the standard. Three hundred and twelve referrals in six months is about six hundred and twenty-four a year, which is about eleven percent of that estimate.

**Sarah:** So the program is missing eighty-nine percent of lonely patients.

**Kiffer:** That's the tempting conclusion, and the reading warns against it. Some people who meet the standard don't want a connector, some are already well connected to services, and some are lonely for reasons a connector can't address. The gap tells you to go and assess felt need and access. It's a poor basis for a referral target on its own.

**Sarah:** How do planners actually gather this evidence?

**Kiffer:** The reading covers four main methods. Epidemiological and social indicators come from existing data, like census counts of older adults living alone or emergency department visit rates. They're fast and comparable, but loneliness doesn't show up in hospital records. Surveys measure the condition and felt need directly. Key informants such as clinicians and Elders explain how the problem shows up in practice. And community forums let residents speak about their own priorities.

**Sarah:** Forums sound like the most democratic of the four.

**Kiffer:** They're open in principle, but attendance is self-selected. The people with the greatest need are often the least able to come to an evening meeting in town. Structured methods help. In the nominal group technique, people write ideas silently, share them in turn without debate, discuss them for clarity and then vote privately, which keeps the loudest person in the room from setting the agenda.

**Sarah:** And then the section turns to assets.

**Kiffer:** Yes. John Kretzmann and John McKnight argued in nineteen ninety-three that needs assessments describe communities mainly through their deficits. Residents start to see themselves as clients, and the capacities a community already has go unnoticed. Asset-based community development starts instead by mapping the skills of individuals, the associations people form and the institutions present, and then builds programs by connecting them.

**Sarah:** Is that a replacement for needs assessment?

**Kiffer:** The two answer different questions, and a plan needs both. The needs assessment tells you how big the problem is and where. The asset map tells you what you can build on. Lynne Friedli has also cautioned that asset language can be used to justify withdrawing public services, especially where poverty is driving the problem. So you report both.

**Sarah:** What did the Cedar Valley asset map show?

**Kiffer:** The regional centre had plenty of groups and a busy seniors' centre, but long waits for volunteer drivers. The rural areas had fewer formal groups but active faith communities, a library branch with weekly social programs and strong informal networks. The partner First Nation identified Elders' gatherings and land-based activities as assets. The practical lesson was that a connector needs something to connect people to, which is why the plan includes partner grants to start groups and a transport fund.

**Sarah:** Section two. A needs assessment finds more problems than anyone can fix. How do you choose?

**Kiffer:** Explicitly, ideally. Every priority decision defers someone's need. Norman Daniels and James Sabin argued that because reasonable people disagree about the principles, the legitimacy of the decision depends on a fair process. Their framework is called accountability for reasonableness. The reasons must be public, they must be relevant to fair-minded people, there must be a way to appeal and revise, and the process must be enforced.

**Sarah:** And the criteria themselves?

**Kiffer:** Most frameworks use five. Burden is the size and seriousness of the problem. Modifiability is whether something available can change it. Equity is how the problem is distributed and whether acting on it narrows unfair differences. Feasibility is whether the organization can deliver a response with the money, staff and partners it has. Acceptability is whether the people affected, the providers and the funders think the response is appropriate.

**Sarah:** Some of those describe the problem and some describe the response.

**Kiffer:** Exactly. Burden and equity describe the problem. The other three describe a response. So the reading recommends scoring each problem paired with a specific candidate program, such as loneliness paired with a connector program.

**Sarah:** Walk me through the scoring example.

**Kiffer:** The Cedar Valley planning committee compared three problems among older adults: loneliness with a connector program, falls with a community exercise program, and medication harm with pharmacist medication reviews. Each was scored from one to five on each criterion. The committee agreed weights before scoring: zero point two five each for burden and equity, zero point two for modifiability, and zero point one five each for feasibility and acceptability. You multiply each score by its weight and add them up. Loneliness came out at four point three five, falls at four point zero five, and medication harm at three point two.

**Sarah:** So loneliness wins.

**Kiffer:** Under those weights. Then the committee tried a different set, giving burden and modifiability zero point three five each and the others zero point one. Falls jumped to four point five and loneliness dropped to three point nine five. Falls has stronger evidence of effective interventions, so it does better when modifiability counts more.

**Sarah:** Doesn't that make the whole exercise arbitrary?

**Kiffer:** It makes the value judgement visible, which is the point. The ranking depends on how much the committee cares about equity and feasibility relative to burden and modifiability. That's a legitimate debate, and the sensitivity analysis shows the board exactly where the debate lies. In this case the committee also noted that the region already funds a falls program, which shrinks the gap a new one would fill.

**Sarah:** The reading mentions the Hanlon method too.

**Kiffer:** The Hanlon method rates size, seriousness and the effectiveness of interventions, with seriousness counting double, so a problem with no effective intervention scores zero. It's usually paired with the PEARL screen, five yes-or-no questions about propriety, economics, acceptability, resources and legality.

**Sarah:** Then the Health Equity Impact Assessment.

**Kiffer:** That's a tool developed by the Ontario Ministry of Health and Long-Term Care. Where priority setting compares problems, the equity assessment takes one proposal and asks who it might affect differently. It has five steps: scoping, potential impacts, mitigation, monitoring and dissemination.

**Sarah:** What happened when Cedar Valley ran it?

**Kiffer:** The committee looked at the referral rule, which only allows referrals from clinicians in participating clinics. The most important finding was that older adults without a regular primary care provider can't be referred at all, and those may be some of the most isolated people in the region. Screening in a short visit could also miss people with hearing loss or limited English, and rural participants face long trips. The mitigation included a referral route from libraries and community agencies, interpreters for screening, and using the transport fund first for rural participants.

**Sarah:** And monitoring?

**Kiffer:** The analyst will report referrals and first meetings each quarter by local area, age group, gender and language. Reporting about First Nations participants follows the partnership agreement, because those data are governed by the Nations. Monitoring is where the equity assessment hands off to the evaluation, since those breakdowns become indicators later.

**Sarah:** Let's talk about local data. What's available in British Columbia?

**Kiffer:** Health authority analysts produce population profiles and service-use data. BC Stats, the provincial statistics agency, publishes population estimates that give you denominators, and the census counts older adults living alone or in low-income households. Population Data BC facilitates access to linked administrative data, such as physician billing, hospital discharges and prescriptions, for approved research.

**Sarah:** That last one sounds powerful.

**Kiffer:** It is, and it's slow. You need an approved research proposal, ethics review and approval from each data steward, and the analysis happens in a secure research environment. Approvals often take many months, so if an evaluation will depend on linked data, the request should start early.

**Sarah:** The reading compares Cedar Valley's four local areas. What did that show?

**Kiffer:** It showed the difference between rates and counts. In Kestrel Lake, a rural area near the partner First Nation, about thirty-six percent of respondents scored six or higher, with a confidence interval from about twenty-eight to forty-four percent. In Cedar City, the regional centre, it was twenty-two percent, with an interval from nineteen to twenty-five. The intervals don't overlap, so the difference probably isn't sampling noise. But Cedar City has far more lonely older adults in absolute terms, about four thousand eight hundred, compared with about one thousand three hundred and sixty in Kestrel Lake.

**Sarah:** So which area gets the resources?

**Kiffer:** If you go by numbers, the city. If you go by rates, the smallest rural area. Michael Marmot's idea of proportionate universalism suggests a middle path: serve every area, with more intensive support where rates and barriers are highest. In Cedar Valley that means transport funds and partner grants weighted toward the rural areas.

**Sarah:** And age standardization comes in where?

**Kiffer:** Rural areas tend to be older, and most health problems become more common with age. The reading compares emergency department visit rates among older adults. Kestrel Lake's crude rate was about fifty-seven visits per thousand higher than Cedar City's. But Kestrel Lake has more people aged eighty-five and older. Once you apply both areas' age-specific rates to the same standard population, the gap shrinks to about twenty-three per thousand. More than half the crude difference was age structure.

**Sarah:** Section three. We know the priority. Now what does the program actually do?

**Kiffer:** That's what planning frameworks are for. They give you a structured path from a health problem to its behavioural and environmental causes, from those causes to the factors a program can change, and from those factors to program components. The result is a plan with an explicit account of why it should work, which is the program theory we'll build into a logic model next week.

**Sarah:** The first framework is Precede-Proceed.

**Kiffer:** Lawrence Green and Marshall Kreuter developed the model, and its core idea is to start with the result and work backward. You ask what quality of life the community wants, which health problems stand in the way, which behaviours and environmental conditions cause those problems, what drives those behaviours, and what program and policy resources you need to change those drivers. Then implementation and evaluation run forward through the same chain.

**Sarah:** And the drivers come in three kinds.

**Kiffer:** Predisposing factors are knowledge, attitudes and beliefs that motivate a behaviour. Enabling factors are the skills, resources and conditions that make it possible. Reinforcing factors are the rewards and feedback that follow it. For a Cedar Valley participant joining a walking group, a belief that loneliness is a private matter is predisposing. Having no ride is enabling, or rather the lack of an enabling factor. A warm welcome from the group and a call from the connector afterwards are reinforcing.

**Sarah:** What's the weakness of Precede-Proceed?

**Kiffer:** It gives you a strong overall architecture and builds evaluation in from the start, but it says relatively little about how to pick specific intervention methods once you know the factors. The other two frameworks add that detail.

**Sarah:** Intervention Mapping is next.

**Kiffer:** Intervention Mapping was developed by Kay Bartholomew, Guy Parcel and Gerjo Kok. It's a six-step protocol that runs from a logic model of the problem, through objectives, program design, production and implementation planning, to an evaluation plan.

**Sarah:** The matrix is the part people remember.

**Kiffer:** It's the distinctive product. You take a behaviour and break it into performance objectives, the specific actions a person has to take. For Cedar Valley, that might be choosing an activity, arranging transport and attending the first session. Then you cross those with determinants such as knowledge, self-efficacy and outcome expectations. Each cell states a change objective. For example, the participant expresses confidence in introducing themselves to the group.

**Sarah:** And then the program has to change that.

**Kiffer:** Right, and step three matches each change objective to a method. Low self-efficacy for a first visit might call for modelling, like a peer describing their own first visit, or guided practice, like the connector going along the first time. The matrices also hand the evaluation its short-term indicators. The cost is time. A full Intervention Mapping process can take many months.

**Sarah:** And the Behaviour Change Wheel.

**Kiffer:** Susan Michie, Maartje van Stralen and Robert West built it in twenty eleven by synthesizing nineteen earlier frameworks. At the centre is the COM-B model, which stands for capability, opportunity, motivation and behaviour. A behaviour happens when a person has the capability, the opportunity and the motivation to do it. Capability is physical or psychological. Opportunity is physical or social. Motivation is reflective, meaning plans and beliefs, or automatic, meaning habits and emotions.

**Sarah:** And around the centre?

**Kiffer:** Nine intervention functions, such as education, persuasion, training, environmental restructuring, modelling and enablement, and around those, seven policy categories, such as guidelines and service provision. When you choose among options, the guide asks you to check affordability, practicability, effectiveness, acceptability, side-effects and equity.

**Sarah:** The reading applies the model to clinicians.

**Kiffer:** Yes, because programs depend on provider behaviour too. Referral rates varied a lot across the twelve first-wave clinics, so the team coded clinician interviews with COM-B. The biggest barrier was physical opportunity: short visits and no prompt in the electronic medical record. Some clinicians had low reflective motivation because they doubted a referral led anywhere, and some felt uncomfortable raising loneliness.

**Sarah:** So what did they choose?

**Kiffer:** A prompt in the record and screening by medical office assistants at check-in, which is environmental restructuring. Training. A one-step electronic referral, which is enablement. And quarterly feedback to each clinic about what happened to the patients it referred, which is persuasion. They considered paying for referrals and rejected it as unaffordable and likely to encourage over-referral.

**Sarah:** How should students choose among the three frameworks?

**Kiffer:** Precede-Proceed suits planning that starts from a population and a broad concern. Intervention Mapping suits a new program or a major redesign where you have time to involve the priority population and ground every component in theory. The Behaviour Change Wheel suits problems where a clearly specified behaviour is the main lever. Many good plans combine them, as long as they say which framework is doing which job.

**Sarah:** Section four. Goals and objectives.

**Kiffer:** A goal is a broad statement of the long-term result you're contributing to, such as older adults in the region having the social connections they want. It sets direction, and it usually can't be measured directly. An objective is a specific, measurable result that the program commits to achieving in a stated period.

**Sarah:** And objectives should be smart.

**Kiffer:** The acronym goes back to George Doran in nineteen eighty-one, writing about management. Public health versions usually say specific, measurable, achievable, relevant and time-bound. In practice I ask students to check for five parts: who, what will change or be delivered, how much, by when, and how you'll know. That last part, the data source, is what makes the objective testable.

**Sarah:** Then there are three levels.

**Kiffer:** Process objectives say what the program will deliver and to whom. Impact objectives describe short and intermediate changes in the factors, behaviours and environments we talked about in Precede-Proceed. Outcome objectives describe longer-term changes in health and quality of life.

**Sarah:** Wait. In Lesson one, impact evaluation meant estimating what a program caused.

**Kiffer:** Good catch, and that's a genuine terminology trap. In evaluation, impact usually means attributable change. In the planning tradition, an impact objective means an intermediate change. Some development agencies use impact for the very longest-term results. The advice in the reading is to define the term once in your plan and use it consistently.

**Sarah:** Give me the Cedar Valley objectives.

**Kiffer:** The process objective is that in the twelve months after the second wave starts, at least eighty percent of referred adults in all twenty-four clinics attend a first meeting with a connector. The first-wave baseline was seventy-seven point two percent. The impact objective is that by twelve weeks, at least sixty percent of participants report taking part in a community activity every week. The outcome objective is that by twelve weeks, at least forty percent of participants who scored six or higher at intake score five or lower.

**Sarah:** Why not set a target for the whole region?

**Kiffer:** That's the target-setting example in the reading. Suppose the committee wants regional prevalence to fall from twenty-four point five to twenty-two percent in two years. That's about one thousand one hundred and fifty fewer lonely people. With all twenty-four clinics referring at the first-wave rate, about one thousand two hundred and forty-eight people a year are referred and about nine hundred and sixty-four attend a first meeting. Even if forty percent of them moved below the cut-off and stayed there, that's about three hundred and eighty-six a year, or seven hundred and seventy-one over two years.

**Sarah:** Which is about one point seven percentage points.

**Kiffer:** Right, and that's before you account for new cases of loneliness and for people who would have improved anyway. So the regional target is implausible for the program alone. It can stay as a population indicator the program contributes to, while the program commits to objectives for its participants.

**Sarah:** You said people who would have improved anyway. Is that regression to the mean?

**Kiffer:** It is. Cedar Valley enrols people because they score high, and people selected for high scores on a measure that varies day to day will tend to score lower next time, even with no program. In the first six months, the mean score among participants with both measurements fell from seven point one to six point three. Some of that fall would have happened regardless. So a pre-post objective is fine for guiding practice, but the plan should say plainly that attribution is a question for the evaluation design, which we cover in Lessons six to eight.

**Sarah:** Let's talk about the needs statement.

**Kiffer:** A needs statement is the part of a plan or proposal that establishes the problem. It defines the problem and who is affected, gives the size and distribution with local data and sources, describes consequences and causes, names assets and current responses, and explains the gap and why action is needed now. It describes the problem before it describes the program, and it reports uncertainty honestly.

**Sarah:** And the rest of the program plan?

**Kiffer:** A work plan lists each activity with who's responsible, when it happens, what it needs and what output marks completion, and every activity should trace to an objective. A Gantt chart shows dependencies, like the fact that referrals can't open until connectors are hired.

**Sarah:** And the budget?

**Kiffer:** At this stage it's an outline that links cost categories to activities. In Cedar Valley, personnel is almost ninety percent of the eight hundred and forty thousand dollar budget, so the number of connectors sets the program's capacity. Lesson ten covers costing in detail.

**Sarah:** So how do the needs statement and the objectives fit together on the page?

**Kiffer:** The needs statement comes first, and it's built on local data with sources. Then come three to five objectives that cover at least the process and outcome levels, in a table that gives each objective's type, its baseline or how the baseline will be measured, and its data source.

**Sarah:** What separates a strong needs statement from a weak one?

**Kiffer:** Real local data with sources and stated uncertainty. More than one type of need. Assets as well as deficits. And targets that make sense given the baseline, the evidence and the size of the program. The Cedar Valley example at the end of section four shows one way to do it.

**Sarah:** Anything students should hold on to from this lesson?

**Kiffer:** State the problem before the solution, show your reasoning when you choose among needs, and write objectives that someone else could check. Those habits make the evaluation much easier later in the term.

**Sarah:** Thanks, Kiffer. Next time, program theory, logic models and theories of change.

**Kiffer:** Thanks, Sarah. See you then.