How to Verify Accuracy of Secondary Data for Research

1. Evaluate Source Credibility

The first and foremost step to verify the accuracy of secondary data is to evaluate the source credibility.
Not all sources are the same, so you need to consider:

• The credibility of the publisher: is it an official data source, an academic source, or an authority in the field?
• The origin of the data: do you know where the figures came from?
• The transparency of the source about its method
• The fact that the data is taken from an authoritative source and not from some anonymous or unbiased data source without any backing information

Data from third party sources such as Statista and IBISWorld is credible but it may be useful to see whether their figures are based on primary data or government census data or other secondary sources.

2. Cross-Verification and Data Triangulation

No matter how reliable a single database is, it should be used sparingly. Comparing the numbers found in different databases is one of the best ways to ensure secondary data accuracy.

• Compare figures from no less than two or three different independent databases
• Detect inconsistencies and explain why they occurred (they may differ in period, methodology, definition etc.)
• Apply triangulation technique before including the figure in a market-sizing or competitive intelligence report [4]

3. Examine Research Methodology

Knowing how the data was collected gives you a lot of information on how you can validate your secondary datasets. Consider:

• Sample size and its statistical adequacy based on the claims made
• Response rate especially if the data was collected using surveys
• Probability sampling or non-probability sampling used in the study
• Methods of data collection such as online panels, field surveys or administrative data
• Limitations and biases associated with the research design that were stated [3]

4. Check Data Currency and Relevance

However, even an analytically sound dataset may be misleading when it is not anymore fresh enough. In evaluating data freshness:

• Look for dates of publication or last updates
• Beware of old data re-used in fresh publications without proper notification
• Make sure that the dataset is a representation of up-to-date market information, particularly in dynamic industries such as IT or healthcare
• Evaluate the connection between the dataset and the goals of the research: an absolutely accurate but irrelevant dataset will not help much
• Compare trends in the past to detect inconsistencies

5. Apply Data Cleaning and Validation Procedures

Prior to any statistical analysis, the following are essential steps that should be taken regarding the secondary datasets:

• Delete duplicated cases
• Address missing data cases properly without simply ignoring them
• Analyze the outliers
• Make sure there is standardization between the variables in the combined dataset [3]

This step is vital in any data validation list and is the basis of data quality assurance.

6. Use Statistical Validation Techniques

Statistics serve to validate your findings in a quantitative sense:

• Conducting descriptive statistics to test plausibility
• Detecting any outliers
• Trying out the correlation between the variables
• Comparing summary statistics with benchmarks or industrial research

7. Run Consistency and Logic Checks

The simple logic test identifies mistakes which statistics might fail to recognize:

• It recognizes impossible numbers (negative age or percentage greater than 100)
• Internal consistency of related variables is checked
• The relation between data points is logical

8. Review Metadata and Documentation

Metadata offers contextual information which cannot be gleaned from figures alone. A good review would be considered:

• Definitions and units of variables
• Assumptions and limitations
• History of revisions, if any, after its original publication.

Complete review of documents will help in verifying facts and prevent misinterpretations down the line [4].