Gentrification and displacement have become so intimately linked in how we talk about certain neighborhoods that they've begun to fuse together, the one word silently implying the other, "displacement" giving "gentrification" its sinister tone (and, presumably, the premise for questions like this one).

Decades of research on revitalizing neighborhoods, however, have left a messy and conflicting record on this front. A thorough recent paper published by the Federal Reserve Bank of San Francisco, written by academics at UCLA and Berkeley, corrals everything we've learned about who — and how many people — gentrification may harm into a single unsatisfying lump. Its conclusion:

Previous studies have failed to build a cumulative understanding of displacement because they have utilized different definitions, compared different populations, and adopted a relatively short timeframe; there is not even agreement on what constitutes a significant effect.

To wit: A study sponsored by HUD that followed families in a revitalizing San Francisco neighborhood in the late 1970s found that a quarter of the residents who'd moved out or within the neighborhood had done so involuntarily — or been displaced.

Around the same time, a nationwide study by Sandra Newman and Michael Owen concluded that about 1 percent of families every year in the U.S. had been displaced, due to reasons like eviction or changing neighborhood conditions. That was the equivalent of about 5 percent of all moves, or 8 percent of moves by urban families.

frequently cited 2004 study, by Lance Freeman and Frank Braconi, found that poor households in New York's gentrifying neighborhoods were actually less likely to move than poor households in low-income communities that weren't gentrifying. That study provocatively suggested there might be benefits to the poor when wealthier families move in around them. The following year, a national study by Freeman concluded that a poor family in a gentrifying neighborhood had a 1.3 percent chance of being displaced, a figure Freeman considered very small.

Then came another study in response, by Kathe Newman and Elvin Wyly. They concluded that between 6-10 percent of all moves in New York City from 1989-2002 were due to displacement. They looked at families who moved because their housing became too expensive, their landlords were harassing them, or they were forced out by actions like condo conversions (all of these studies define "displacement" and even "gentrification" in slightly different ways shaped by theory or limited by data).

Other studies, along with these, have routinely found that the people moving into such changing neighborhoods have been wealthier, whiter and more educated than the people moving out. But that still leaves us without a good sense of what happens to the people who do leave, and how different their paths would be if their neighborhoods didn't change at all.

This last question — the counterfactual — is one of the hardest to answer. Poor families tend to move a lot. So how do we single out the forces of gentrification from all the other reasons why a family might otherwise leave a poor neighborhood where wealthier whites aren't moving in. Maybe they can no longer double up with family, or they want to leave behind a high crime rate.

It's certainly inaccurate to suggest that all moves out of a changing neighborhood should be counted as "displacement." And part of the difficulty here also lies in the blurry line between what people choose to do, and what's forced on them. If a family chooses to sell its home to profit off rising property values, is it displaced? What about a family that can still afford the rent but leaves because the bar scene has become too rowdy?

Large datasets rarely reflect these nuances. And researchers are often trying to understand a group that by definition isn't around to tell us why they moved, or whether they felt pushed out. Measuring displacement, Rowland Atkinson wrote in 2000, is like trying to measure the invisible. The people we want to know about have scattered from the place we're trying to understand.

A final complication: Neighborhood change takes place at an incredibly slow pace (despite how it sometimes feels when new restaurants are sprouting on, say, Washington's 14th Street). That means a study that runs from 1970-1977, or even 1989-2002, doesn't capture at lot of what's going on. And because the pace of change is gradual, the Fed paper points out, it's hard to contrast a "before" and "after" picture of a single neighborhood. There's no bright line separating the two.

One takeaway from all of this is that gentrification — especially its relationship to displacement — is hard to study. But it's also true amid all this uncertainty that researchers have left some space for the possibility that revitalization may benefit long-time residents, too. And this record hardly supports the idea that "gentrification" and "displacement" are synonyms for the same phenomenon.