Secondary analysis

1
Secondary analysis

• Jillian Ireland

2
Secondary analysis definition

• Secondary analysis involves the utilisation of
  existing data, collected for the purposes of a
  prior study, in order to pursue a research
  interest which is distinct from that of the
  original work
• (Heaton 1998)

3
What is secondary analysis?

• Secondary analysis uses existing data, e.g.
  collected for a prior study or for policy-making
  reasons, in order to address a new research
  question or an alternative perspective on the
  original question.

4
Plenty of data around

• Great mass of data is routinely collected by
  governments, businesses, schools, and other
  organisations.  Much of this info is stored in
  electronic databases that can be accessed and
  analyzed. 
• Many research projects store data in electronic
  form in computer archives so that others can also
  analyze the data.

5
Re-using data sets

• Secondary analysis is efficient.  It makes use of
  data already collected (Research equivalent of
  recycling). 
• Allows you to extend the scope of your study
  considerably.  E.g. in many research projects it
  is impossible to consider taking a national
  sample because of costs. 

6
Advantages I

• Limited opportunities for conducting primary
  research and costs of quantitative and
  qualitative work have prompted researchers to
  consider maximising use of existing data
  available to them.

7
Advantages II

• In other words, you can conduct a much bigger
  study (i.e. larger sample), then if you had to
  conduct a primary study. A larger sample allows
  you to do a more detailed (useful) statistical
  analysis.

8
Advantages III

• Many archived databases are already national in
  scope. Using these (perhaps linking different
  databases), you can conduct a much broader study
  than if you collected the data yourself with a
  relatively small budget.

9
Advantages IV

• No need to get people to participate in new
  research.
• Unobtrusive measurement presumably reduces biases
  that result from the intrusion of the researcher
  (e.g. interviewer bias) or measurement
  instrument. 

10
Possible issues I

• Sample comparability
• In any attempt to aggregate or compare the data
  from different official data files, significant
  differences might be found in recording or coding
  specific data.

11
Comparing countries

• Even within the UK in both health and education
  there are many differences in coverage, time
  periods, classifications, coding practices,
  definitions, and other issues, between the
  official statistics produced by Scotland, England
  Wales, and Northern Ireland.

12
Possible issues II

• Original data for secondary analysis are
  collected for a different reason, e.g. patient
  data for service provision, marketing purposes
  (e.g. viewing data for TV), crime statistics for
  policy-making, etc.

13
Possible Issues III

• Using someone elses data greatly reduces the
  degree of control the researcher has over the
  type of data collected.

14
Possible Issues IV

• Using data sets collected by other researchers or
  organisations you often don't know what problems
  occurred during the data collection.

15
Example Quantitative

• Analysis of demographic details of season tickets
  holders of a club, e.g.
• Are women become season ticket holders at a later
  age than men?
• How likely is it that more than one season ticket
  holder lives at same address?

16
Example MSc project

• Original study Postnatal Care in Grampian
  n1249
• Secondary analysis Does attendance at antenatal
  classes have an effect on the postnatal outcomes
  of first-time mothers?

17
Advantages

• As slides 6, 7, 8, 9
• Plus, my supervisor was the author of the
  original study (N.B. this could also be a
  disadvantage choose carefully!)
• Recoding was easy e.g. type of delivery was
  changed from 5 codes to 2 normal or other

18
More advantages

• More time to spend on analysis (students often
  leave too little time)
• 3 methods descriptive statistics, multiple
  logistic regression and matched pairs.

19
Issues

• Could not add questions (although, because the
  original study was very thorough, this was
  academic)
• I would recommend secondary analysis!

20
Data Protection

• But 1998 Data Protection Act makes it difficult
  to get access to the data.
• For example, the University of Aberdeen cant
  just give me as a researcher personal details
  about you as a student for research purposes.
  This includes your address!

21
Secondary analysis Qual. data

• Can involve the use of single or multiple
  qualitative data sets
• Can mix qualitative quantitative data sets
• Researchers could re-use own data or qualitative
  data collected by others.

22
Example Qualitative

• Szabo Strang (1997) used secondary analysis
  of their previous study on informal carers of
  relatives with dementia to consider how carers'
  perceived 'control' enabled them to manage their
  care giving experience.
• Szabo V. Strang VR. (1997) 'Secondary
  analysis of qualitative data', Advanc Nurs Sci,
  20 66-74.

23
Qualitative limitations I

• Some question whether secondary analysis of
  qualitative studies as it involves an
  inter-subjective relationship between researcher
  and interviewee / focus group participants.

24
Qualitative limitations II

• Where does primary analysis stop and secondary
  analysis start? Qualitative research is an
  iterative on-going process and often the
  questions undergo reformulation and refinement
  over time.

25
Some useful references

• Corti, L Thompson P (2004) Secondary analysis
  of archived data, In Qualitative Research
  Practice by Seale, C et al. (Eds), Sage London.
• Heaton J (1998) Secondary analysis of qualitative
  data, Social Research Update 22 (web address
  http//www.soc.surrey.ac.uk/sru/SRU22.html)
• Hinds PS, et al. (1997) The possibilities and
  pitfalls of doing a secondary analysis of a
  qualitative data set, Qualitative Health
  Research, 7 408-24
• Corbetta, P. (2003) Social Research Theory,
  Methods Techniques, Sage, pp.150-152