What is a Framework and how do they help find the story in the data?
Before Information Technology a framework was the skeleton of a building or object, something around which the finished product could be built. But, over the last 30 years the term framework has expanded into a much wider range of situations, embracing conceptual structures and procedural models. It is this wider concept of frameworks that is of value in helping find the story in the data.
Examples of well-established frameworks include:
- Software framework, a collection of programs and rules to allow new software to be added to an existing system, or for a system to be created from libraries and established practices.
- Legal framework, a set of laws and procedures.
- Local planning framework, an overview of how a geographic area is going to be developed, including the aims, rules, and presumptions.
- Conceptual framework, a set of theories that within the framework are accepted and allow the field to be developed.
It is also worth mentioning a couple of specific methods that include the terms Framework:
- Framework Analysis. A method of analysing qualitative data, i.e. an alternative to Grounded Theory.
- Logical Framework Analysis. A method/tool for planning and implementing projects.
Why do we need a framework approach to finding the story?
Some people appear to be naturally good at finding the story in the data. They look at the numbers, read the comments, rearrange a few elements and seem to intuitively find the story. However, in most cases these people are not good at teaching other people how to find the story in the data.
What teams need, and what most individuals need, is a systematic way to approach a data set that will efficiently and reliably find the story within the data. Especially when the data comprise disparate streams of information, such as survey data, social media discourses, transactional data, and qualitative data.
A framework approach to finding insight creates a system that can be standardised within a team or organisation, that can facilitate collaboration (which is hard with intuitive approaches), and which can help avoid issues such as silo thinking. However, it whilst a framework approach can be efficient and reliable, it will rarely be as fast as a good intuitive researcher working on their own – but it is more replicable and less prone to error.
The framework you develop for your organisation will be unique in its detail, but will probably share some key characteristics. Below I have set out a simplified basic framework.
- Define the problem. Most problems that can be clearly defined are already on the way to being solved. This involves more than just asking somebody what the problem is. Try asking questions like, what would success look like, what actions do you want to take once you have solved this problem, and what is holding you back from taking those actions now?
- What is already known? Too often a research project seems to assume that the problem can be solved with one data set and should be addressed anew. However, in many cases some or even all the answers already exist. In this part of the framework the key elements are to find out what is already known, who holds the information, and what its strengths and limitations are.
- What is believed? The assumptions and hypotheses that exist are a subset of what is already known. The answer to the problem has to confirm or refute what is already believed, which means that these factors need to be explicitly addressed by the project.
- Organise the data. The data include what is already known and any new information gathered for the project. It is likely that there will be incompatibilities in the data, for example different scales, different time periods, qual and quant, these need to be addressed. It is also likely that there will be a range of challenges in using the data such as missing data, quality issues, and validity issues; these must also be addressed.
- Identifying the findings. With the question being used as a prism, the organised data are reviewed for constructs, messages, themes, groups, and ideas. This process answers questions such as ‘What do most people believe?’ and ‘Are their key groups or sub-groups?’
- Create the story. The findings are reviewed and linked, expanding and organising the themes and constructs found in step 4. The framework will have a target structure for the story, such as: One big idea, supported by three themes, with each theme having three clear components or examples. Anything that can’t be linked to the story is not part of the story. Anything that is not strong enough to be in the story structure (e.g. the main idea, 3 themes, an 9 components) is not part of the main story.
- The parameters of the story. The level of confidence that the audience should have in the story needs to be identified and communicated. The story needs to communicate what the audience should think, feel, and do. Part of the ‘think’ part is understanding any risks or limitations, part of the ‘feel’ part relates to risk acceptance or rejection, part of the ‘do’ part relates to how to monitor the outcome.
- Dealing with the rest of the findings. There are usually many findings that do not make it into the story. The framework should give some guidance as to how these should be documented and dealt with. For example, these findings may be used over time, or they may be parcelled up into packages of information and sent to relevant recipients.
Note, although I have set out this framework in a linear form, it should be approached more flexibly. When answers start to appear it may be necessary to re-frame the original problem, for example making it narrower or wider, depending on the circumstances. During the process of identifying the finding it may be necessary to source more information or re-visit the organisation of the data. During the creation of the story it is usually important to challenge the story, this may require the researcher to go back through all the steps, from problem definition to story creation.
More?
This post is based on a series of workshops I have run with different organisations. Over the next few months I am booked to run several more workshops and presentations and I am working on a new book (which will be published late 2016 or early 2017). Therefore, I will be publishing a series of posts and I’d really appreciate feed on the points raised, the methods suggested, and the detail of what I have said.
If you have any case studies you’d like to share, or examples of problems, or references to interesting articles and books, please let me know – either as a comment on this page or via email.
4 thoughts on “What is a Framework and how do they help find the story in the data?”
Comments are closed.
All right Ray. In my opinion when you define a “framework approach” and stress the need to have one before starting any enquiry or research you are unveiling the reason why Big Data are giving disappointing outcome, so far. Precisely for that: a lack of appropriate “framework” (i.e. lack of a genuine effort to understand the process below). Isn’t it?
Quite right, too often the question put to big data is too vague. When big data has been successful it is usually because a very small, very precise question has been asked and the data have been properly filtered.
I like the emphasis on “What is known?” This is a question too rarely asked up front, and is key to understanding the type of insights. See Robert Moran’s Taxonomy of Insights, which takes into account prior beliefs about findings.
I typically present results in an Associated Press pyramid, from most important to least important. The one exception I make is when the results contradict beliefs. In that case, I find that the results are better received if we go through the methodology and accumulate subpoints that all lead to the major finding that the core belief has been contradicted. Presenting it as most important points first typically leads to disbelief and rejection of the findings.
Hi Ray – Your point that “some people appear to be naturally good at finding the story in the data” is interesting and speaks to a weakness in the research industry – the over reliance on ‘intuition’ and ‘tacit knowledge’ in data collection and interpretation.
It’s easy to understand why this happens – many researchers don’t have the time, budget, or academic training to apply theoretical frameworks to their studies. None-the-less, conducting research without a theoretical framework is risky – leaving findings open to subjective bias, transference and spurious relationships.