How To Solve a Research Problem


Elizabeth Shown Mills


It’s doable. Really. All it takes are two nevers and three always. 

1. Never rely on a single document for any point, even when it tells you exactly what you want to know. Records often err and we won’t know it until and unless we test their assertions against other records. With any piece of research our goal should be to use all relevant records and to make certain that anything we use to “corroborate” something else has totally independent origins.

2. Never assume any type of record is reliable. Record types are like people. Some are totally dependable. Some make us cringe or roll our eyeballs. Most are a mixture. We love what’s good about them. We’re wise to look warily at the flaws.

3. Always test each finding—no matter how small it may be—against the Evidence Analysis Process Map.  The ultimate “fact” in all projects is this:  Our conclusion can never be better than the evidence on which we base it.

4. Always use each finding to lead you to something else. No fact is an absolute. No record is an end to itself.  Facts and records are always part of something much bigger. To understand the fragment we’ve found and interpret it correctly, we need the rest of what went on.

5. Always contrast, compare, and challenge. Each new piece of information needs to be scrutinized rigorously. We study not just the details in a record but the construction of the record and the record set. We watch for anomalies in the records and discrepancies between them.  

We study the behavior of the people who created each record. We study the laws that governed the record—and the degree to which participants and their communities complied with those laws. We constantly look for ways to measure differences and similarities—and ways to quantify actions. We watch for behavior that is aberrant to community norms—which means we may also have to study the community and its records deeply enough to define those norms.

Yes. It’s doable. I didn’t say it’s quick 'n easy. But it’s doable. The issue now is this: How badly do you want to solve that research problem?

PHOTO CREDITS: "5 Star Solution," CanStockPhoto ( : downloaded 7 April 2015), image csp16887728, uploaded by drcmarx, 20 November 2013; used under license.


Posted 8 April 2015