Analysing your dementia research data

"Analysis" is the process of looking for patterns in the data which answer the big questions you wanted to explore.

Patterns might include: differences, similarities, relationships, trends and/or themes.

The way you analyse may be more or less sophisticated depending on the type of data you have and what you want to find out.

What might the results of your data analysis tell you?

They might tell you about:

• how well services are working currently
• current gaps and barriers to using, and benefitting from, services
• future needs
• opportunities for better outcomes or lower cost  (or both)
• how easy or not it might be to implement ideas for solutions to issue.

How to analyse quantitative data

You might look at counts, averages, trends over time, differences between subgroups, or correlations. Graphs help bring data to life.

Be aware:

• The larger the sample size (number of people who provided your feedback/ideas), the smaller the margin of error.
• Consider the size difference between groups who provided you feedback/ideas. Small differences in the data from each group could just come down to chance.
• Beware spurious correlations- in other words, avoid seeing connections in your data that are not really there. 

How to analyse qualitative data

When you collect qualitative data, your responses may be recorded in various forms such as:

• notes (especially as 'quotes') of meetings
• notes about observed behaviours 
• transcripts of interviews
• video diaries
• checklist observation and ideas notes from 'walk throughs' checking how dementia-friendly particular places were for people living with dementia.

For observational type methods you may be looking for responses to your overall research question (or challenge you want ideas for how to solve) from records, the media, images, or searching words used in online forum conversations.

Breakdown all that information into meaningful groups: 'coding'

• Look for patterns - group or 'code' comments.
  • You might like to use an existing framework - the Dementia Statements or your service outcomes, for example.
• Draw out key themes
  • representing similarities and differences emerging from the data.
  • even something said only once could be a key theme - it is not dependent upon how many times something has been said.
  • pick out quotes to support and evidence each theme you’ve identified
  • Example of themes identified from group discussions. This file is a pdf made from what in the original is a filtered spreadsheet enabling views of just selected themes, or of all the responses. The pdf doesn't have the filters function but you can still see the themes from the colour coding of the theme words. 
• Computer technology can be helpful here
  • you could do this in table format in Word.
  • If you use Excel you could use filters to enable you to search for all the responses in the group you've identified. 
  • you might like to use computer software such as NVivo computer software for qualitative data analysis
• Check how wide a range of people have responded - breakdown your data further by grouping people's responses by shared characteristics such as age, gender, living in their own home or in a care setting, completed the survey themselves or had a carer/staff complete the survey for them. 

Tips:

• Allow yourself plenty of time for analysis.
  • If you have a survey that included qualitative questions, you may potentially have thousands of responses to analyse. 
  • Qualitative research is not a numbers game, so balance the number of participants with the time and resource you have to collect and analyse their data.
  • See our section on sample groups and recruiting people affected by dementia to be involved.
• look at the words used - and, where possible, beyond the words used.  
  • Dementia may change how a person uses words, or incline a person to feel more emotional in situations than they would in the past.  So look at the words used, and bear in mind how their dementia affects the individual's communication - look for what they are really trying to communicate, even if they get their words muddled or use very strong language.
  • Consider what you can see as well as what you hear: for example, someone may have tried to please you by saying their experience of using your experience was OK, but you were observing them use it. You saw how long it took and their body language indicating intense concentration, uncertainty and that it was a struggle for them. 
  • Consider the context in which the person shared their story and how this may have impacted on what they said, what might have been embellished or left out, and so on. For example, it's often observed that there may be marked differences between what people say in everyday conversation with people they know well and what they may say during a formal, recorded, interview.
     

Myth: It's only one person's story so I can't use it, because I don't know that it's representative of the wider population's experience.

While it may be challenging or unwise to produce an action plan for improvement based on one or two comments, in the end it depends (quality over quantity).

• Example: given a diagnosis of dementia on the day before Christmas.
  
  When a person in the East of England told her local commissioners and providers about receiving her diagnosis of dementia the day before Christmas, and the devastating impact that had on her family and herself over the holidays period when lots of services were closed, the organisations who heard her decided to reflect on that experience: they decide to look into why it happened - and whether it could be avoided for other people.