American Research Group has just released new polls of Iowa, New Hampshire and South Carolina .

Click through if you want to read about the results from the other states. The new ARG Iowa numbers are:

Edwards 27 percent

Clinton 23 percent

Obama 19 percent

undecided 16 percent

Biden 6 percent

Richardson 5 percent

Dodd 2 percent

Kucinich 2 percent

Gravel and Clark, less than 1 percent

You may recall that in December and January ARG released polls showing Clinton way ahead in Iowa. No other polling firm ever found Clinton ahead of Edwards in Iowa.

Now ARG’s results are more in line with the other firms. I don’t know if ARG changed its likely voter screen or if Edwards has just gained a lot of ground on Clinton since January.

ARG notes that Edwards leads Clinton among women in Iowa, 32 percent to 26 percent. This is not at all surprising to me.

The undecided figure still “feels” low to me. I talk to a lot of Democrats, and I think way more than 1 in 6 tell me they don’t know yet who they will caucus for.

Click through if you want all the Republican numbers for Iowa, but the highlights are:

McCain 26 percent

Giuliani 19 percent

Romney 14 percent

Fred Thompson 13 percent

undecided 12 percent

Seems like Romney’s numbers are not moving in Iowa, despite his early tv ads and the DVD he mailed out to 70,000 Iowa Republicans. Click the link if you want to read the Cyclone Conservative analysis of that DVD.

UPDATE: Check out rob’s diary at the right for more info about these polls. One figure jumped out at me. ARG’s sample in Iowa was taken from “600 completed telephone interviews among a random sample of likely Democratic caucus goers living in Iowa (475 Democrats and 125 no party (independent) voters).”

Now, we had a few independents change their registration on caucus night 2004 so that they could participate. But does anyone really think that 1 in 5 people attending the Democratic side of the Iowa caucuses next January will be registered independents?