1,400 New Englanders decided what you watched last night. Who are they?

And why is there growing concern that their views don't add up?

Every day and every night, 600 households in New England decide which TV shows the rest of us are watching.

With a click of their remotes, these 1,400 panelists - known as Nielsen families - tune into their preferred programs. A small box called a "people meter" on their TV set logs their age, gender, and the programs they watch. That information provides stations with numbers to court advertisers. These select few families operate in secrecy.

Yet there's growing controversy over whether the number of these families are too small, how they are recruited, and whether the meters accurately log what they watch.

As station managers pore over their ratings, they evaluate which programs perform well. And sometimes questions arise when the numbers don't look right.

The TV industry pays incredibly close attention to these ratings. A slip in ratings can cost the stations ad dollars and bragging rights, so they are at the mercy of the Nielsen families, who represent 2.4 million households in the Boston TV market, which includes most of Massachusetts, part of New Hampshire, and one county in Vermont.

"By and large, I think they get most things right," said Bill Fine, president and general manager of WCVB-TV (Channel 5). "But on the fringes, there are enough areas of concern for us that it makes a big difference. If the margin between winning and losing a time period is one tenth of a point, you want it to be on your own efforts and not on errors with the ratings system. That is where our confidence level has been shaken a bit."

Fine's faith in the ratings system has been rattled in the past few years. He questioned a drastic drop in viewership for three New England Patriots home games, which typically generate huge ratings. WCVB investigated and learned that a technical problem credited the majority of its viewership to ESPN, which had simulcast the games on WCVB. Last year Nielsen recalculated the ratings and restored the missing viewers to WCVB. For one game in particular, the new numbers credited WCVB with three times the ratings points.

"Nielsen ratings are like currency," said Geoff Klapisch, media and advertising professor at Boston University. TV stations, he said, "absolutely do live and die by them. The Nielsen methodology is constantly under scrutiny."

Gary Holmes, a spokesman for the Nielsen Co., said the system - and the people meters - are accurate despite sporadic hiccups. "We sometimes do get questions on particular shows," he said, "and if that's the case we always investigate to make sure there is no sample issue or technological issue."

Nielsen is funded by television networks and local TV stations, which subscribe to its service and pay for its research. Nielsen recruits people to serve as panelists, and their participation is kept confidential. The company would not help the Globe find current or former Nielsen families unless the paper promised not to print their full names or towns.

"It's very important for people that no one in the industry know who the panelists are," Holmes said. "There would be an incentive there to target them somehow."

Nielsen representatives use census information to pick panelists randomly. Age, race, and ethnicity are not factors. "None of the characteristics of that household come into play," said Michael Link, a chief methodologist for Nielsen. "If you do the random selection properly, they should fall out as a natural selection process."

To find families, field representatives descend into neighborhoods in the 16 counties that constitute the Boston TV market, looking for volunteers. When a household agrees to serve, workers affix the "people meters" to its TV sets. Families are paid a small stipend for their work. Nielsen would not say how much, but one former panelist who asked not to be named said she earned $300 for one year. Panelists can participate for up to two years.

Each member of the household is assigned a code that represents them by age and gender. Whenever they turn on a TV or enter a room in which a TV is already on, they are supposed to punch in their code, which tells the meter who is watching and which program is on. If the viewer leaves the room, he or she must again punch in a code. If a family has guests at the residence, those people are also given codes so their viewing habits can be measured.

Some media buyers, who use ratings to buy TV airtime for clients, find the methodology limiting because the numbers depend on how active the panelists are with their viewing habits inside their homes and how accurate Nielsen is in tracking them.

"It's an imperfect science," said Tracie Manna Chinetti, a senior buyer at Blitz Media in Needham. "If you are off by even 1 percent of the audience, it can have a dramatic effect on the station's spot costs and their revenue . . . We are still relying on people to tell the system what they are doing. You get a phone call and you leave the room - did you remember to click yourself out? There's a lot of viewing that is not being measured."

Lisa Gomes of Stoughton can relate. She participated in 2000, when the Boston market still used paper diaries to track viewing. She wrote down every show she, her husband, and two young sons watched on three TV sets, mostly at night.

"I had to do it all myself because no one else would do it," said Gomes, 45. "It was too much work. I also worked full time, and I had to do all that. It was crazy."

Greg Angland, 38, of Mansfield was a Nielsen panelist last year for the Providence TV market, which still uses diaries. "You have some say in the success or failure of a TV program," said Angland, who has a wife and two young children.

"The challenge of it was trying to keep it accurate while having a very difficult busy lifestyle," said Angland, who also works at Blitz Media. "It was hard. I have kids in one room and my wife in another. I watched 'Curious George' a lot because I was in the room when my kids were watching it."

Another problem that some TV managers have with the system is that only homes are counted. Many programs - sports in particular - are viewed in public places such as gyms, bars, restaurants, and college dorms, but Nielsen does not sample those viewers. Nielsen began exploring the idea of adding meters to businesses but says it suspended plans because of the recession.

"Across the country, we would love to have the service extended to the transient nature of the viewing, which would be in hotels, bars, and dormitories," said Ed Piette, president and general manager of WBZ-TV (Channel 4) and WSBK-TV (Channel 38).

It is these kinds of issues that can make station managers wary of the ratings figures. WHDH-TV (Channel 7) noticed that ratings for last January's Golden Globes Awards came in unusually low compared to previous years. WHDH officials, who declined to comment for this story, also complained to Nielsen. The ratings company reprocessed the numbers and awarded higher ratings to WHDH, according to a letter from Nielsen.

Officials at WFXT-TV (Channel 25) and WBZ say they haven't had major problems with Nielsen. "They are as accurate as you can find," Piette said. "There have been some technical issues that have come up, and those little anomalies are worked out."

Johnny Diaz can be reached at jodiaz@globe.com