The Arkansas Congressional Map
Following the census, Arkansas stayed at 4 districts and merely had to accommodate population disparities between the districts. The 2nd district and heavily Republican 3rd had to shed population while the 1st and 4th needed to gain it. Thus, Dems moved the 1st 0.7% to the left by having it grab a few small Delta counties. The 2nd shed Yell County to otherwise stand pat. The 4th, having to gain the most population, added heavily conservative parts of the Ozarks in the northwest of the state, shifting about 1.5% towards Republicans.
The result is that, when looking at the partisanship of the districts below the presidential level, we were left with this:
Note, the partisan average consists of all D vs R statewide races below president since 2006. I did this to be able to make an apples to apples comparison between the districts since the races are all the same. Additionally, the average is calculated without county splits, so the 3rd and 4th might be marginally different, but the presidential numbers are accurate. All population figures are Voting Age Population only and all partisan numbers are two-party share only.
As you can see, all three of the 1st, 2nd, and 4th are more Democratic than the state while the 3rd is quite clearly packed. However, none of the districts is significantly more Democratic than the state at large and Romney won all of them comfortably. With Rep. Ross retiring and Democrats running three weak, underfunded candidates, Republicans were able to win every district.
The Art of the Dummymander
What exactly is a Dummymander? Well it's a gerrymander that, instead of benefiting the party that drew it compared to a non-partisan map, actually hurts the party compared to that same baseline and results in the party winning a much lower number of seats than expected. This can be caused by spreading one's partisans too thin, or by not taking into account partisan trends and incumbency. This happened famously with the 1992 Georgia Dem gerrymander (pdf) that turned a 9D-1R advantage in 1991 into a 3D-8R deficit by 2001.
With that in mind, the next step then is to determine what a non-partisan map of the state might look like. The criteria I had in mind were similar to that of the California citizens' commission-drawn map that valued compactness and communities of interest and ignored partisanship. Clearly this is a subjective exercise and there are multiple ways to draw the map, but here is one version I feel is quite realistic that a court or commission might draw:
Arkansas' regions pretty neatly fit into four districts with the Delta comprising the 1st, the Little Rock area in the 2nd, the Ozarks in the 3rd, and South Arkansas in the 4th. In contrast to the actual map, there's a clear Democratic district, the 1st.
Compared to the real map, the 1st drops parts of the Ozarks in north-central Arkansas and adds Pine Bluff and the rest of the Delta. It gets 3% more Democratic and instead of just 6% of the district being new, it's now nearly 29% new to incumbent Rep. Rick Crawford (R). Despite Obama getting just under 42% there this year, it's by far the most Democratic of the four. John Kerry won nearly 51% there in 2004 and Sen. Blanche Lincoln won 48.4% in 2010 while getting just 39% statewide. No other Democrat lost the district; even in 2010 the Dem house nominees won 51%. Tallying up the actual election results from 2012, the 1st district nominee, prosecutor Scott Ellington (D), and 4th district nominee, state Sen. Gene Jeffress (D), won a combined 44.4% to Rick Crawford's and Tom Cotton's 55.6%. However, a sizable proportion of that figure comes from the 4th district where Rep.-Elect Tom Cotton (R) was a much stronger opponent than Rick Crawford and Jeffress ran a very 19th century campaign with little advertising. With a stronger, more well funded challenger such as 2010 nominee (and former chief of staff to then Rep. Marion Berry) Chad Causey, Democrats likely could have gained the 5.6% vote share needed for 50%+1.
Consequently, the 2nd district slides the other way, though it remains more Democratic than the state. The district gets 1.6% more Republican and would have been fairly secure for Rep. Tim Griffin. The Dem house nominees, Jeffress and attorney and former state Rep. Herb Rule won 42.2% there this year, which certainly could have been improved upon with a well funded Blue Dog, but unlike in the 1st Griffin is a stronger incumbent and clearing 50% would have been difficult. If this district were open it would be competitive, but Griffin would have won this year.
The third district gets slightly less Republican, but is still dark red and Rep. Steve Womack (R), who was unopposed this year, would have certainly won reelection.
The 4th district, like the 2nd, gets more Republican under this non-partisan map thanks to the loss of Pine Bluff and the addition of Fort Smith. Had Mike Ross not retired Dems might have held this seat, but without him Tom Cotton would have seen an even larger win over Gene Jeffress. This seat isn't completely out of reach though, as Democrats have a strong bench with state Sen. Larry Teague, who holds down a conservative district, and University of Arkansas Community College chancellor Chris Thomason coming to mind. Still, Cotton would have likely gotten even more entrenched than he will in his actual district, so Democrats probably wouldn't win this one period.
Thus, by drawing the map they did, Democrats actually received 1 seat less than they would have under a non-partisan map, making theirs a Dummymander.
The Map Democrats Should Have Drawn
With their poor performance at winning three of four seats in 2010 and 2012 in mind, I set about to draw a strong 2D-2R map that Democrats could and should have drawn. With Mike Ross retiring, the 4th district needed to be shored up in a way that didn't entice Tim Griffin into running there. This was a bit tricky since some of the most Democratic parts of the state in Little Rock were in the 2nd. Next, I aimed to defeat Rick Crawford in the 1st. Finally, since I was using county-level data and Arkansas has historically not split counties during redistricting, I had to make sure all districts were within +/-1% of the target population. With a budget for gathering precinct-level data, Arkansas Dems could have drawn an even more effective map than this one by splitting counties.
Under this map, the 1st gets 2.6% more Democratic for a total of 8% more Democratic than the state and it's 24% new to Rick Crawford. Though Obama only received 40.3% here in 2012, John Kerry barely lost it by just 731 votes in 2004 and Blanch Lincoln won 47.2% in 2010. The Dem House candidates won 45.4% and that should have easily gone past 50% with a well-funded candidate like Chad Causey. This seat should have been Lean D to Likely D, with the end result probably being around a 54%-46% Dem win.
The 2nd district is the key to this map since it drops heavily Dem Little Rock but still entices Griffin into running there. The district is 55% new to Griffin, but it's over 9% more Republican and significantly safer than the neighboring 4th. Griffin would have had this seat as long as he wanted it if he made it past the primary.
The 3rd is almost entirely the same as was drawn with only 6% being new, meaning Womack would have been unopposed again and like Griffin would have had the seat as long as he wanted it.
The 4th district here looks radically different than the real version thanks to avoiding the Ozarks and the addition of Little Rock. Tom Cotton still lives here, but he and other ambitious Republicans might have run in the much safer 2nd district instead. It is made 6.7% more Democratic for a total of 8.4% more Dem than the state and Obama actually won it this year by 1004 votes. John Kerry won 54% here and Blanche Lincoln was the only Democrat to lose it since 2006 and that was by just 439 votes. Combining Herb Rule's and Gene Jeffress' vote shares gets Democrats to 49.3% and with any competently run campaign they would have held onto it easily making it Safe D.
Now that Democrats lost the state legislature following the 2012 elections, passing a map like this would be impossible. However, it seems abundantly clear that without any help from Republicans, Arkansas Democrats cost the party two congressional seats.