Skip to Content
Interactive Textbook on Clinical Symptom Research Logo


Home Button

Secondary Analysis of Large Survey Database
Author Bio
Why Conduct Secondary Anaylsis
Advantages of Survey Data
Avoiding the Pitfalls
Currently Selected Section: Start with the Research Question
Determine Variables of Interest
Identify and Evaluate the Data Source
Get the Data
Survey Design
Sampling Frame
Telephone Surveys
Followback Surveys
Multistage Cluster Samples
What is a Panel Design
Mode of Survey Administration
Survey Instruments
CodeBooks
Online Exploratory Analysis
Potential Sources of Error
Cultural Nonequivalence
Analysis of Survey Data
Cluster and Stratified Samples
Using Sample Weights
Missing Data
Power Calculations
Linking Data Sources
Multiple Comparisons
Getting Help
Giving Feedback
Conclusion
Chapter 20: Secondary Analysis of Large Survey Database: Start with the Research Question
          

Once the investigator has defined the purpose of an analysis, he or she must identify the most appropriate data source available to achieve the study's aims.

Descriptive analyses provide new insights into relationships that have previously been insufficiently explored, and can generate hypotheses that can subsequently be tested empirically. Analytic studies using survey data, like any other scientific investigation, begin with well-defined research questions and testable hypotheses.

As stated before, some investigators may be tempted to find an interesting data source and then explore it for associations of interest (data-dredging). Findings using this strategy are problematic for reasons that include:

  • Spurious associations related to large sample sizes and large numbers of variables are commonplace;
  • Conflicting or unexpected findings may be difficult to interpret; and
  • Evidence from post hoc analyses is considerably weaker than findings from research based upon conceptual models and testable hypotheses.

Numerous survey databases containing health care information are currently available, providing the capacity to answer a wide array of questions. Data limitations may require modifications to the study questions or scope of the study. Balancing feasibility and limitations to specify the research question and to identify the best data source requires time, but increases the likelihood that the research will yield useful insights.

 

Page 5 of 30
      Previous Section