Whatever type of information you have collected, analysis and discussion of your
findings will be time-consuming. This can be daunting for those new to inquiry, but
once begun can reveal fascinating discoveries.
Data needs analysing – it does not 'speak for itself'. Different people interpret
evidence differently. Your analysis will support your interpretation of your data.
Analysis is about organising the information you have collected, drawing out themes
and trends, and interpreting its significance within the context of the study.
How you analyse your data will be determined by the type of information you have
collected and should be part of your research design. The format of the data (questionnaire
replies, test scores, audio or video recordings, written accounts, etc.) will determine
the resources required.
There are many long-established techniques that can be applied although sometimes
you will need to invent or adapt the analysis to fit your research question.
Although free software packages designed for specific data analysis are available,
high-specification applications can be expensive. Most practitioner researchers rely
on the traditional methods unless their project is part of a course of study and they
have access to college or university computer facilities.
The process of preparing data for analysis should begin as it is collected. Raw
data needs to be sorted, coded and indexed in a systemised way.
In some strategies (eg grounded theory), preliminary data analysis informs continuing
data collection.
Analysis should result in a balanced interpretation of what the data reveals, including
acknowledgement of any limitations that will impact on your conclusions.
Coding is the process of packaging information. It can be built into the data collection
strategy. For example, a question might read: 'Is the young person aged 5-7, 8-11,
12-14, 15-16, 16+?'. The coding for these groups may then be 5-7 = 1, 8-11= 2, and
so on. This approach is known as deductive coding and can really
save analysis time. Despite this advantage, it requires thorough planning to cover
all possible responses.
Inductive coding supports exploratory inquiry, since responses do
not need to fit pre-determined constraints. The coding is derived from categories
suggested by the data. For example, answers to: 'Where do you go on holiday?' may
be best categorised by town, region, country, etc., depending on the range of responses
given.
Computer applications save time and allow for more complex analysis. However, the
availability of such hardware and software should not sway your method of analysis.
The primary consideration should always be to use the method that is most appropriate
for the research question and data type.
Many experienced researchers, particularly in the quantitative field, argue that though
undertaking the process yourself adds time to the analysis phase, it can have a beneficial
impact through the interpretations you form and modify as you work with your data.
This example illustrates that it is how the data is understood that influences
the conclusions.
This is a prime example of a situation where analysing observational data alongside
other sources of data (triangulation) may produce a more rounded interpretation.
It also makes a case for involving other people at key points during your data analysis.
How will others interpret a sample of your material? It may be that they offer a different
or better perspective from your own. If several people interpret the information one
way, this may be the more valid interpretation. This is known as interrater reliability.
Blaxter, J, Hughes, C. and Tight, M. (2001) How to Research, Buckingham: Open University Press.
Cohen, L., Manion, L. and Morrison K. (2011) Research Methods in Education. 7th edition. London Routledge.