Playa Rincón, Samaná, Dominican Republic

Higher, broader, and deeper level results

Thomas Aston


I’ve written in the past on the difference between outputs and outcomes, but I’ve rarely reflected on the issues of breadth or depth of results. For example, is self-efficacy an intermediate level result or a higher-level result? This was a question Marina Apgar and I discussed recently, and where we somewhat disagreed, and it offered me the opportunity to rethink my position. It depends where you’re coming from and what you’re trying to achieve. Self-efficacy is a common realist mechanism of change, and thus it connects contextual factors with outcomes. But, self-efficacy could also be argued to be a change in someone’s “mental model” — it’s people’s beliefs in their capabilities to produce desired effects by their actions. There is, indeed, a line of systems thinking which sees it purely in epistemic and not ontological terms (i.e., “there is no system out there, only ways of understanding how things in the world relate to each other”).

Empowerment is another such change much debated. Is empowerment an intermediate step — similar to self-efficacy (i.e., power within) which leads to results such as accountability or responsiveness, or is empowerment the real change you’re looking for? It rather depends which aspects of empowerment we’re talking about; whether these are personal, relational, or environmental, and whether power is actually exercised or not. Though let’s not go too far down that rabbit hole. I’ve written in the past about how there can be trade offs between the higher level outcomes you prioritize. In a nutshell, all desired results don’t necessarily go hand-in-hand.

Nowadays, I’d argue it’s more common to talk about “systems change” than “empowerment.” Donnella Meadows’ work on leverage points and places to intervene in a system was a landmark piece of work in how we understand systems change. Leverage points are presented in supposed increasing order of effectiveness (i.e., the further up you go, the more transformative the change). David Ehrlichman made a nice graphic which represents these leverage points:

Ehrlichman, 2018

In short, Meadows suggests that cognitive paradigms supersede, institutions, institutions supersede processes, and processes supersede material changes. This reflects a particular cognitive paradigm for understanding what is important and how what is quantifiable is not necessarily what is most important. I share some sympathies for this perspective. But, I think there are limits to this conceptual paradigm, as I’ll explain below.

Building on this, and the work of Peter Senge, Kania et al. developed the six conditions of Systems Change framework, illustrated below:

Kania et al. 2018

This is a model now widely embraced, but I’ve yet to see it receive much critical scrutiny. Marcus Jenal provides a useful discussion of Kania et al’s six conditions of systems change. I agree with him that not only changes in mental models are transformative. I think he’s also right that we need to map constraints and understand how robust or permeable these are. As a political economy analysis aficionado, this is not wholly new information to me, but it’s nonetheless an important point about what drives and impedes change.

I personally find the inverted pyramid problematic because it clashes with my own prior (biased, partial, limited) experience of using Knowledge, Attitude, and Practice (KAP) Surveys. While at CARE International over a decade ago, I can recall working on assessing gender based violence in Bolivian schools and how, for instance, some teachers might increase their knowledge, and even change their attitudes, but did not in the end change their practices. Attitudes are not wholly equivalent to mental models, but are similarly a cognitive state and domain of change. So, why is it that an attitudinal change would find itself low in a results chain and a change in mental model might be considered the deepest level of change? It seems like a contradiction. I have a feeling that the reason for this because most logic models and theories of change actually flow in the opposite direction — from knowledge and attitudes to material changes in people’s lives (i.e., impact).

Rather than suggesting that one mental model (cognitive or material) is superior to the other, I believe there’s a problem with hierarchical schematic representation in general. The suggestion that there are six “conditions” actually fits with a configurational logic (i.e., necessary and/or sufficient conditions, or more often INUS conditions) for system change to be achieved. They are not sequential or hierarchical steps (as the diagram implies).

Views of systems change are often intertwined with notions of either sustainability or scaling up. There isn’t one singluar understanding of “scale.” There are a wide range of different interpretations. In recent years, we talk not only of scaling up, but scaling out, and even scaling deep. Much of the scaling up literature speaks to how your small intervention, innovation, or idea can grow. An evidence-based policy lens suggests “replication” of a model with high implementation fidelity is what we’re looking for. At CARE International, our main aim was reaching and impacting as many people as possible (e.g., reduced malnutrition or maternal mortality). Elsewhere, we find that aim is for other farmers or companies to adopt an innovation.

A couple of years ago, Tatiana Fraser wrote an insightful blog Scaling Deep: Where it came from and more to go followed by an useful paper on The Art of Scaling Deep which questions the notion that bigger is always better.

UNDP, 2021

Growth isn’t always good. And this is partly why some, such as the Griffith Centre for Systems Innovation, now talk about “right scaling.” Scale needs to be seen relative to context and thus we need to move between big, wide, and deep. The Griffith Centre have a helpful background paper on everyday patterns of shifting systems. I believe they’re right that everyday patterns may look small, but can make a significant difference (it’s not “linear”). They have their own iceberg which situates events at the top and values at the bottom (systems icebergs are a twist on other hierarchies).

Scaling deep is similar to the aims of changing hearts and minds. As the Griffith Centre put it, ‘scaling deep is imperative for grappling with and shifting the mental models and assumptions that underpin current patterns in the system.’ Changing hearts is arguably a deeper change than changing minds. It sits at the axiological level of values. Thus, I agree with them that a shift in normative values is probably deeper than mindsets.

Nonetheless, there are serious limitations of thinking of systems hierarchically (i.e., deepest) rather than as a configuration of factors in a network and a set of inter-related outcomes. Looking at systems change as if it were a matter of root causes seems itself to be the wrong paradigm, as Jenal argues. In my view, we need both material conditions AND values to change for there to be more robust and durable (rather than brittle) systems change — one is not necessarily more important than the other. Both are likely necessary but insufficient conditions. I can’t help but think back to an anecdote in James Ferguson’s great book Teach a Man to Fish when one participant in a multi-day human rights workshop said: “All I have heard about today is that I have the right to a house […] But the problem is I don’t want the right to a house […] I want a house.”

Thus, in my view, we should instead be thinking in terms of how, for instance, changes like empowerment relate to responsiveness (i.e., interaction effects), and recognise the positive and negative feedback loops there may be between “progress” in different domains (dimensions or areas) of change related to the system we’re focused on. Consider the photo above. It’s of Rincón beach in the Dominican Republic where a river (Caño Frio) feeds into Rincón bay. If we’re talking about ecological sustainability, for example, we need to look at both the river and the bay, and we need to look at the inter-relationship between the two. Or as Chris Corrigan puts it, systems thinking is a set of “ways of understanding how things in the world relate to each other.”

So, in my view, rather than icebergs and results hierarchies, at lease once we get beyond outputs, it’s time for some more horizontal, configurational, and network thinking about systems change.



Thomas Aston

I'm an independent consultant specialising in theory-based and participatory evaluation methods.