Research shows that patients who say they have memory problems are more likely to develop full-blown dementia.

But a new study from Pennsylvania State University finds that many of the questions researchers commonly ask older people about memory are confusing or anxiety-provoking — flaws that could weaken research findings. The study revealed "wide variation" in how people interpreted the questions and decided how to respond, said Nikki Hill, an assistant professor of nursing who led the project. It was published in the Gerontologist.

Questions about self-reported memory problems can flag people at higher risk for dementia well before commonly used cognitive tests would, she said.  Those tests tend to pick up people who already have fairly serious problems, especially if they were very high functioning to begin with.  What's more, the tests are given in quiet environments where it's easier to concentrate, while "real life is really messy and complicated," Hill said.

She thinks that better questions could help define which people are at highest risk.

Even if people aren't on the road to dementia, perceived memory problems can be a sign of depression or anxiety, Hill said.  And, if people are worried, for example, that they're forgetting names, they may withdraw from social situations.  Isolation can also be harmful.

>>READ MORE: Many people with dementia don't know it.  Here is who's most at risk.

At this point, Hill is focusing on questions about memory that are often used in research, but the results could also apply to discussions that doctors have with patients in their offices.  Her "end goal" is to develop a short questionnaire that doctors can use to identify patients at risk.

In Hill's study, 49 older adults from central Pennsylvania who did not have dementia were asked to answer 20 questions that are often used to screen for memory problems.  Hill's team then asked participants what they were thinking when they answered the questions.

People complained that all of the questions were vague. They said that many weren't clear about time periods, types of memory, and whether the problem was consistent or episodic. They couldn't know the answers to some questions, such as how they compared to others.  Many questions made them anxious.

This question was considered particularly weak:  Compared to other people your age, how would you rate your memory on a one to five scale with the anchors "excellent" and "poor"?

And participants found it troubling to be asked to compare themselves to others.  Making people feel threatened might change their answers, Hill said.

Many of the study participants craved more precise language, Hill said.  Let's say they were asked, "Have you had memory problems in the last year?"  They wanted to know what kind of memory problems, and whether the aim was to identify problems that were unusual for them.  One participant assumed researchers would want to know only about "serious" problems.  What if they'd always had trouble remembering names?  If asked how much their memories had changed in 10 years, many were aware that retirement or a move to a retirement community had made their lives much easier, so they had less to remember.

Hill said she didn't see a pattern in whether people were more likely to lean toward saying they did or didn't have memory problems if they considered a question vague.

Time frames that were shorter and more specific worked better.  It also helped to specify if researchers were interested in memory problems that affected important daily activities.  Hill thinks perception of a decline is likely important.

This is a question that got relatively little criticism in the study:  In the last month, how often did memory problems interfere with your daily activities: every day, most days, some days, rarely, or never?

Hill said she's not yet ready to recommend an ideal question.  "There's a lot we don't know yet," she said.