Can you analyse our existing data?

Yes. We can check and analyse data you already hold, such as customer records, complaint data or earlier studies.

What does “statistically significant” mean?

That a difference between groups is very unlikely to have occurred by chance. Our reports clearly mark significant differences.

Data analysis

Data gains value when it is analysed properly. We secure data quality before analysis and use core and advanced statistical methods to connect findings to your business question.

Send a brief

4principles: accurate, reliable, useful, actionable
5core analyses: frequency, difference, association, correlation, regression
95%confidence level for significance tests

Before analysis: quality data

Before analysing the data collected, PRAGMA makes sure responses are coded correctly and meaningfully and runs pre-analysis checks. Open-ended answers are classified with a code frame, and inconsistent or incomplete records are removed.

We test the independence of answer options, whether responses follow normal and random distributions, and the reliability and validity of the data. Where needed, the sample is weighted to match the population. Analysis then starts from quality data.

We carry out every analysis with a holistic, scientific approach; our aim is results that are accurate, reliable, useful and actionable.

Core analyses

The backbone of every study.

Frequency analysisBuilds the overall profile of the data and shows how often each answer is given.
Difference testsChecks whether differences between groups such as gender, age, region or socio-economic status are statistically significant.
Association testsExamines links between variables, for example whether a rise in market share can be explained by positive consumer attitudes.
Correlation analysisMeasures the direction and strength of relationships between variables.
Regression analysisMeasures the effect of one variable on another and builds predictive models.

Advanced analyses

Some of the methods we apply as needed.

SegmentationCluster analysis to group customers meaningfully by needs and behaviour.
Driver analysisShows which factors drive satisfaction, preference or loyalty most.
Factor analysisSummarises many statements into a few underlying dimensions.
Perceptual mapsVisually compares brands’ positions against competitors.
Qualitative analysisThematic analysis of transcripts, with insights in the audience’s own words.

The analysis process

1CodingWe classify open-ended answers with a code frame and prepare the data.
2ChecksWe run consistency, distribution, reliability and validity tests.
3AnalysisWe combine core and advanced methods to fit your business question.
4InterpretationWe turn findings into insight and insight into actionable recommendations.

What we pay attention to

We judge every finding by one question: what does this tell the decision maker?

Data cleaning and consistency checks before analysis
Weighting to the population where needed
Significance testing of differences between groups
Cautious interpretation of small subgroups
Participants’ own words in qualitative findings
Prioritised recommendations tied to the business question

Frequently asked questions

By the purpose of the study and the decision you need to make. We agree an analysis plan at the design stage and build the questionnaire around it.

Stage 3Data deliveryThe value of research lies in results that are clear and usable. Once analysis is complete, we deliver findings in a simple, visual and decision-focused way.

Let’s turn your data into decisions

Briefly describe your research need or the data you have; we will come back with a suitable analysis plan.