Larger, More Complex Studies: Secondary Analysis, Big Data, and Meta-Analyses

Secondary Analysis

  • Analysis of data collected from another study for a new purpose.

Big Data

  • Definition: Collection and analysis of overwhelmingly massive amounts of information.
  • Characteristics:
    • Huge volume.
    • Great variety.
    • High velocity.
  • Research challenges:
    • Rigor of the study design.
    • Reliability/validity of theory.
    • Protection of participant identity.
  • Steps in the Development of a Big Data Study:
    • Research question.
    • Data management.
    • Data processing.
    • Data analysis.
  • Challenges Related to the Quality of Big Data:
    • Data entry errors.
    • Transforming data into analyzable variables.
    • Discarding nurses’ notes.

Meta-Analysis

  • Analysis of results of multiple studies on a single topic.

  • May be qualitative or quantitative.

    • Qualitative: metasynthesis or metasummary.
    • Quantitative: meta-analysis.

  • Steps in Beginning a Meta-Analysis

    • Define study aims and state a research question
    • Set boundaries on search for relevant studies
    • Conduct search
    • Select relevant studies
    • Evaluate quality of selected studies

  • Quantitative Meta-Analysis

    • Most commonly uses effect size
    • Use of correlations and odd ratios
    • Refinement of studies
      • Quality
      • Effects of certain characteristics
    • Alternative approach
      • Metadata banks

  • Qualitative Meta-Analysis

    • Initial steps similar to those of quantitative meta-analysis
    • Type of available data determines whether you conduct a:
      • Metasynthesis
      • Metasummary

  • Steps in Metasynthesis

    • Read and reread each study
    • Identify “key” metaphors
    • Standardize findings using common codes
    • Identify relationships among findings by showing how key metaphors relate
    • Juxtapose the studies’ findings
    • Develop a line of argument

  • Questions About Metasynthesis

    • Can the findings of studies from such a diverse base be combined?
    • Should they be clustered by the method used?
    • If the levels of abstraction differ widely across studies, does this undermine the ability to synthesize the findings?
    • Should only peer-reviewed articles be included?
    • Are the findings generalizable?

  • Metasummary

    • Used when data are summarized rather than synthesized
    • Results or findings are extracted from the articles and grouped by topic