‘How to’ resource
Introduction
Data and analysis help inform good policy. Local policymakers in the UK are able to access a range of data on the performance of their local economy. This briefing provides guidance on how this data can be analysed.
This briefing should be used alongside What Works Growth’s ‘how to’ guides on understanding local economic performance. Each briefing focuses on a specific growth topic (such as skills) and provides details of the data available and a framework for understanding what it shows.
Defining the question
Start by clearly defining the question to be answered through data analysis. Good questions should be:
- Specific – clearly stating what will be investigated.
- Focused – relate to a single issue or topic.
- Researchable – it should be possible to answer it through data analysis.
- Feasible – given the resources available (including time).
It is common to have multiple questions and for initial analysis to generate additional questions.
Selecting data
Finding relevant data
Establish what datasets are available that might help answer the question. Sources of data include the Office for National Statistics (ONS), Nomis, ONS Local, UK and devolved government departments and agencies, and other organisations.
Most datasets include multiple variables and may help you answer multiple questions. Whilst some may only include a few interconnected variables, others cover multiple topics.
Assessing the data
Assess whether the dataset meets your needs. Key questions include:
- Will the data help answer the research question?
- What geographies are available?
- What time periods are available?
- What is the frequency and the timeliness of the data?
Analysing data
Most economic data relates to either:
- Stocks – which measure the quantity that exists at a specific point in time. Examples include the amount of labour, skills, machinery or infrastructure available in a local economy or the overall size of the local economy.
- Flows – which measure changes over time. Examples include investment or numbers gaining qualifications in a specific topic.
Data can be analysed in multiple ways.
Most analysis starts by looking at the current position. It is possible to look at the current position for both stocks and flows.
Comparison across areas allows benchmarking. It is important to select comparison areas carefully to ensure benchmarking is useful.
Comparison over time looks at how things are changing. It is good practice to pick a neutral time period (such as five or 10 years) but in some cases analysis will look at performance since a notable event (such as recession or election). It is possible to compare over time for both stocks and flows.
Analysing breakdowns can provide additional insights. For example, comparing different geographic areas or different groups within a population.
Combining data can provide additional insights. For example, combining data on productivity and income helps illustrate whether local people are benefiting from high productivity.
A clearly defined research question helps set the parameters (for example, by indicating that comparison across areas or time is required).
Presenting data
Choice of chart
Different types of charts provide different insights.
- Bar charts can be used to show levels, change over time (specifically between two points in times), or benchmarking. They can use either horizontal bars or columns. When showing levels or change over time, it can often be useful to organise from highest to lowest (or vice-versa).
- Stacked bar charts can be used to compare breakdowns across categories.
- Line charts can be used to show change over time, including indexing. To ensure it is easy to interpret, limit the number of lines (generally no more than five).
All charts should have a title and source. Chart titles should include the measure being shown, units (if appropriate), geography, and date. If there is additional information that the reader should be aware of when reviewing the chart, this should be given in a note.
Other design choices
- Use colours to differentiate between data series. For example, you may wish to present the area of interest in a different colour to the comparator areas.
- Consider whether axes should start at zero or another level. This will depend on what you are trying to illustrate and how much variation there is between different groups.
