BOSTON—The rule of thumb is that you’ll need to replace 70 percent of your preretirement income on average once you retire, but evidence continues to mount that this assumption by many professionals and retirement savers is way off base.

Now, a new study by two professors casts further doubt on the idea that the widely used replacement- rate figure is a sound basis for building a retirement plan.

“The rule of thumb that replacement rates should be above 70 percent to maintain living standards in retirement is conceptually flawed,” wrote John Karl Scholz and Ananth Seshadri, two University of Wisconsin-Madison professors, in their paper “What Replace Rates Should Households Use?”

In fact, no more than 15 percent of the population Scholz and Seshadri studied need to replace 65 percent to 90 percent of their preretirement income. And almost 50 percent of the population needed to replace less than 65 percent of their preretirement income.

In short, the authors said: More refined guidance is needed to serve households well.

Target replacement rates are less than 100 percent for three main reasons, according to the study published by the Michigan Retirement Research Center.

“First, upon retirement, households typically face lower taxes than they face during their working years, if for no other reason than Social Security is more lightly taxed than wages and salaries. Second, households typically save less in retirement than they do during their working years, so saving is a smaller claim on available income. Third, work-related expenses generally fall in retirement.”

Still, that ignores a whole host of issues related to coming up with the right replacement rate.

For instance, consider what effect children likely have on your expenses prior to and in retirement, the authors wrote. Most calculators use the same replacement rate regardless of the number of children in a household, the authors said. But the number of children you have matters when it comes to calculating your replacement rate. In fact, all things being equal, a household with lots of children will have a smaller replacement rate than a household with no children, because the couple with kids, once retired, will face far lower child-rearing costs than they did while working. (Of course, the kids’ ages at the time of retirement will affect that calculation.)

What is needed for your real number is not back-of-the-napkin calculations but something the authors refer to as the life-cycle model. To be fair, the author’s study did note that replacement rates—even when using the life-cycle model—did confirm some commonly held beliefs. Specifically, “replacement rates of low-income individuals and families would need to be higher than replacement rates for high-income individuals and families.”

But even then you still need to take into account the effect of federal taxes, medical expenses, education, and what the authors call earnings shocks or—in laymen’s terms—layoffs and big salary increases.

When all is said and done, the authors suggest that optimal replacement rates could range anywhere from 23 percent (for single parents with several children and a negative late-in-career earnings shock) to 240 percent (for low-income, married households with a few children and a substantial positive late-in-career earnings shock).

In other words, “conventional advice may overstate optimal targets by a factor of two, or understate retirement consumption needs by a factor of three depending on the idiosyncratic experiences of households,” Scholz and Seshadri said in their study. That’s especially the case when it comes to online calculators, the authors said.

With the life-cycle model, the replacement rate depends on factors often ignored by online calculators. “The savings requirements of two households with the same earnings profile, retirement age and life expectancy would be given an equivalent target by the online planning tools regardless of whether one household raised five children and other had none,” the authors said. They said the optimal replacement rate for married couples is 75 percent, but just 55 percent for singles.

Put another way, if you’re using an online calculator to plan your retirement, you might be undersaving or oversaving by a wide margin, though the consequence of oversaving might not be as bad as undersaving.

Experts, meanwhile, seemed to agree with the conclusions reached by Scholz and Seshadri.

“The use of replacement rates to form financial plans does not meet a reasonable fiduciary standard,” said Larry Kotlikoff, a Boston University professor.

“Rules of thumb are, quite simply, rules of dumb,” he said. “Their use violates the financial planner’s Hippocratic oath: First do no harm.”

Kotlikoff also said the model used in Scholz and Seshadri’s study is not without its warts. “But it’s fine for its purpose, which is comparing conventional financial planning with economics-based planning.”

Rick Miller, a certified financial planner with Sensible Financial Planning, cautioned against using any rules of thumb. “Using rules of thumb can be very dangerous, if they significantly understate the requirement, or can risk significant regret, if they overstate the requirement.”

As appealing as rules of thumb are, when it comes to replacement ratios there’s just no getting around it: You simply have to crunch the numbers.