In recent weeks, Democratic critics have hammered President Bush's economic policies, noting statistics that show nearly 2.1 million payroll jobs have been lost since he took office and saying that the nation is stuck in a "jobless" recovery from the 2001 recession.

But other government figures paint a different picture. A monthly survey of households indicates that the net job loss since early 2001 has been only about 600,000. That survey, adjusted for population changes introduced this year, found that after 1.4 million people lost jobs in the first year of the recession, about 800,000 found work over the past year.

Those recent additions to the workforce help explain why the nation's unemployment rate has remained steady at about 6 percent or less for the past 17 months.

Neither survey suggests a strongly rising demand for workers. "The best you can say at this point is that it does not seem to be getting worse," Wayne M. Ayers, chief economist at Fleet Boston Financial Corp., said of the job market. "There are no huge gains and no huge losses. By the payroll numbers, it is a jobless recovery. By the household numbers, it is not exactly a jobless recovery, but it is not a robust labor market either."

Most analysts agree that the unemployment rate will not fall until economic growth picks up. The economy grew 2.9 percent last year, and most economists expect growth in the first half of this year to be lower than that. Thus, many analysts predict the unemployment rate could rise further.

One reason that last year's growth failed to generate more jobs was that labor productivity -- the amount of goods and services produced for each hour worked -- rose significantly. That is good news in the long run because it leads to higher incomes for workers and higher profits for businesses, but in the short run it reduces the demand for workers.

"We have got to see growth above 3 percent for some time for the unemployment rate to come down," Ayers said.

The figure of 2.1 million payroll jobs lost is a focus because it comes from a monthly Labor Department survey of more than 300,000 businesses that employ 37 million of the country's 130 million workers. Because of the survey's size, most economists regard it as more reliable over shorter periods than the companion survey, which covers only about 60,000 American households.

But some analysts, including Ayers, said that once a recovery begins, the household survey results may be more reliable.

"The payroll survey is generally considered to be better, but coming out of a recession the payroll survey won't pick up all the jobs being created," Ayers said. "Over the past 12 months, the household survey is showing much stronger growth in employment, and that's not unusual." Because new firms are not immediately included, "the payroll survey usually underestimates the increase in new hires and in self-employment," he said.

The self-employed are not included in the payroll survey but are counted as working in the household survey. Their ranks have grown by about 400,000 since March of last year after declining by 656,000 in the 12 months before that.

This is not the first time the stories told by the two surveys have been so different, though no one has ever been able to fully explain why. The surveys always have later converged -- though the experts can't explain that either.

"When the gap opened up in the past, it tended to narrow later," said Alan B. Krueger, an economics professor at Princeton University and a former chief economist at the Labor Department. "Maybe it's just a catch-up, but we don't know why that happens."

Economist L. Douglas Lee of Economics From Washington Inc., a consulting firm, recalled that a similar divergence developed in the early 1990s, and in that instance the household survey turned out to have been closer to the mark.

"Most of the time we believe the payroll data are the more accurate, but in that case the payroll data got revised upward," Lee said. "The speculation was that at a turning point in the business cycle, when growth resumes after a decline during a recession, it takes one or two years to capture the firms that are being created in the payroll survey. On the other hand, the stuff that goes out of business and the jobs that are lost show up immediately and it's picked up in the survey."

Some analysts -- and some of Bush's critics -- have also suggested that the current 5.8 percent unemployment rate underestimates the degree of stress in the labor pool because many workers have become so discouraged that they dropped out of the labor force and aren't counted as officially unemployed. Others have pointed to the rising number of people forced to work part time when they want full-time paychecks as another sign of distress.

Those figures all come from the household survey, which last month showed 145.8 million people in the labor force -- 137.3 million who had a job and 8.4 million who were actively seeking one. Over the past 12 months, the labor force, employment and unemployment have all increased at the same rate, so last month's jobless rate was the same 5.8 percent as it was in March 2002.

Over that same span, the working-age population, 16 and over, has increased at a faster rate than the share of the population in the labor force. So the proportion of the population with a job, the "participation rate," has declined to 66.2 percent, from 67.1 percent when the recession began.

Some analysts regard that decline as a measure of the degree to which workers are discouraged about their job prospects. On the other hand, the participation rate has always declined during recessions when jobs become less plentiful. Had the participation rate not declined, the jobless rate would be about 6.2 percent.

Of the people not in the labor force, the Labor Department regards only 474,000 as "discouraged." They are workers who have looked for a job sometime in the past year and are not currently looking because they think no job is available for them. That figure is 125,000 higher than at the beginning of the recession.

The household survey shows that 4.7 million people who were seeking full-time employment were working part time last month. That group represented 3.4 percent of the workforce -- compared with 5.4 percent at a comparable point after the 1990-91 recession.