Querying the impossible once again….

As readers of this blog know (and I am grateful to each of you, especially as the one-year anniversary of this blog arrives tomorrow), I am writing a book tentatively titled Interrogating Memory: Film Noir Spurs a Deep Dive Into My Family History…and My Own. The impetus for the book came from a career-related conversation with my ever-supportive wife.

The notion of “interrogating memory” emerged when I began to research and write this book. My initial plan was simply to trace my path to becoming a film noir aficionado: from the still-hazy circumstances of my adoption through being a precocious child reader of mysteries through my discovery of the 20th Century Fox Charlie Chan films through my widened exposure to films and hardboiled detective fiction at Yale through the perpetual lure of “night and the city” through my ongoing embrace of the Film Noir Foundation and its annual NOIR CITY film festival (about which I have written here).

But as I began to think more critically about relevant childhood memories—stories I had refined to a high gloss after decades of retelling—I realized they did not always neatly align with verifiable facts from independent sources (e.g., newspaper accounts, contemporaneous records, diaries, photographs).

For example, the story of my in utero adoption always included my Colombian genetic father and Native-American great-grandparent (or was it great-great-?). Only this past summer did I learn through 23andMe genetic testing that I am, at best, 0.5% Iberian and not at all Native American.

Sometimes meticulous investigation raises more questions than answers, as when the ages listed on an old photograph of my sister and me could not possibly both be correct.

And in still other instances, I discovered that memories I had convinced myself were false (or, at best, completely mangled through the mnemonic equivalent of a game of Telephone) turned out to be almost entirely true, like the story of the early childhood friend I never saw again after he severely burned himself.

I still adhere to the last sentence in that post: “It is remarkable what you can learn (good and bad) when you interrogate your memory.”

Thus, “Why do I love film noir so much?” morphed quickly into “Who am I?” (driven externally by ongoing conversations with my psychotherapist) leading inexorably to the epistemological exercise of memory interrogation—and the resultant traveling of unexpected and unusual research paths: from an influential Masonic lodge to a mid-20th-century “Crime Prevention Squad” to a seaside motel that was demolished in 1978.


The latest bit of memory interrogation is the direct result of setting the historic stage (in what I expect will be Chapter 7: “Chinatown”) for that Saturday night in July 1976 when a bored nine-year-old version of me stumbled across a Sherlock Holmes/Charlie Chan double-feature on Philadelphia’s Channel 48.

If it had been either of the previous two summers, I would have missed these movies entirely, because my mother, father and I would have been at the Strand Motel in Atlantic City, New Jersey (seedily nestled in the square lot bordered to the east and west by the Boardwalk and Pacific Avenue and to the north and south by Boston and Providence Avenues)…or strolling the Boardwalk…or visiting my maternal grandfather at the Warwick Apartments on Raleigh Avenue (which look exactly the same as they did 40 years ago).


I am still interrogating memories of why our family finances did not allow us to spend the summer of 1976 in Atlantic City (with my father driving back-and-forth the 84 or so miles every weekend), as we had the previous two summers.

But those two blissful summers on the Jersey shore were particularly liberating for reasons that had nothing to do with finances.


My legal parents married in January 1960. Over the next four years, as they sought to have children, my mother had two miscarriages before being diagnosed with cervical cancer (or uterine cancer, depending on the report[1]) in 1964, when she was just 26 years old. As a result, she had a full hysterectomy, limiting her natural childbearing to a girl—who she and my father named Mindy Joy—born in March 1962.

Unfortunately, Mindy’s birth followed an extremely painful, 18-hour-long labor. In the process, Mindy’s head kept emerging in and out of my mother’s vagina, possibly restricting oxygen flow to Mindy’s still-developing brain.

Perhaps this is why Mindy had delayed developmental milestones, leading to a 1960s-vintage diagnosis of “Severe Mental Retardation” (along with Seizure Disorder—although she has not had a seizure since June 1993), which is now “Depressive Disorder due to another medical condition w/Mixed Features and Pervasive Developmental Disorder.” This disorder may (though medications now greatly reduce their frequency) result in “[u]npredictable changes in mood states, which can lead to tantrum behaviors. Mindy can scratch, bite, hit, pinch and pull others’ hair.”

My parents, of course, loved my older sister absolutely (my father was the one person who could adequately control Mindy—who was inordinately impulsive and strong—when he was around to do so).

Still, in the late summer/early fall of 1966, they arranged through a private attorney they knew to adopt a second child. This is how a four-day-old version of me got driven away from Pennsylvania Hospital to a three-bedroom home on a quiet Havertown street in October 1966.

But that is a story for another day.

As difficult and uncommunicative as she could be, I generally got along well with my older sister, as this 1971 photograph shows.

Mindy and I 1971

It helped the impish younger version of me that Mindy often exhibits echolalia, parroting back words and phrases just spoken to her. Attempting to play with her, I would sit Mindy down and have her repeat words like “Czechoslovakia” and “Yugoslavia.” This would only last a short time, until she began to get agitated. That was my cue to move on to a less potentially destructive activity.

Younger siblings may be a bit cruel at times, but they ain’t stupid.


As of last year, Mindy’s annual progress reports are now being prepared by a non-profit advocacy agency called The Arc Alliance.

These reports have incorporated more detail about a process I vividly recall from my early childhood: the ceaseless search for a long-term care and education facility that would accommodate Mindy for more than a few months.

Between 1970 and 1974, alone, Mindy attended the…

  • Elwyn Institute (Media, PA; 1970-71),
  • Melmark School (Berwyn, PA; February 1971-73),
  • Martha Lloyd Residence (Troy, PA—a 3-4 hour drive north, nearly to the New York state line; July-August 1973),
  • Crozer-Chester Medical Center Intermediate Unit Program (Chester, PA; fall 1973)
  • Van Hook-Walsh School (Middletown, Delaware—an hour-plus drive south; February-June 1974) and
  • NHS School Woodhaven (Philadelphia, PA; since December 3, 1974)

It was not atypical for Mindy to have been “terminated” from the Van Hook-Walsh School after “the neighbors complained.”

Mindy’s requirement for 24-hour care and supervision made it next-to-impossible to do much in the summertime (or in the evenings, as babysitters who could accommodate Mindy were scarce, to say the least).

So when I look at that June 1974 termination date in Mindy’s Arc Alliance annual report, I feel like my memory is playing tricks on me.

There is simply no way we could have spent one night in Atlantic City, let alone an entire summer, if Mindy were not in a residential care facility in July and August 1974.

But if Mindy really was “terminated” from the now-defunct Van Hook-Walsh School in June 1974 and did not move into the brand new, Temple-University-operated Woodhaven campus[2] (where she has been a resident for 43 years and counting) until that December…then where was she that summer?


Just bear with me while I tell a quick story (which I may never be able to interrogate):

By the fall of 1974 my mother had finally tired of finding a permanent program for Mindy, and she wanted to get her enrolled into the newly-opened Woodhaven facility as quickly as possible. However, she was getting nothing but delays and “be patient” from Woodhaven administrators. One afternoon, she and Mindy were in “the offices” [I have no idea where these would have been], and she was getting the same “be patient” message. My mother finally snapped…and she stopped trying to keep Mindy from being disruptive. Mindy promptly ran around the office throwing papers, yelling and generally wreaking havoc. My mother then vowed she would bring Mindy back there every afternoon until she was enrolled. Within a few days, Mindy was accepted into Woodhaven, where she remains to this day.

I am certain there is more than a kernel of truth to this story (I would not have invented it, and my mother certainly had her badass moments), but I may never know how much.


As you would expect, I will now interrogate the memory that my mother and I spent the summer of 1974 living at the Strand Motel in Atlantic City.

As evidence, I submit three photographs with the same developer’s code stamped on the back of each one, two of which have also have “Aug 1974” written on their backs in my mother’s handwriting.

Luvey in Atlantic City August 1974 2

Luvey in Atlantic City August 1974

Strand motel August 1974

The first two photographs are of Luvey, the keeshond we acquired in January 1973. In the second photograph, he is clearly sitting in the doorway that led to the patio we shared with the “B” penthouse (we stayed in the “A” penthouse—a slightly larger motel room with a walk-in closet) overlooking the outdoor pool and, across the adjacent Boardwalk, the Atlantic Ocean.

Incidentally, the resident of the “B” penthouse both years was this interesting man. I used to walk his beautiful golden retriever Whiskey with Luvey, and I once asked him (he would have been 32 or 33 years old) what he wanted to be when he grew up.

I do not remember his answer.

The third photograph shows my mother sitting on the motel room patio of family friends (cropped out to protect privacy) who also spent that summer at the Strand Motel. The second photo of Luvey was taken in their motel room.

If my mother looks, umm, blissed out in this photograph, well…she used to buy her grass from the Strand Motel’s handsome young male lifeguards.

(My mother once told that I was not allowed to start smoking weed until I was 32, because that was how old she was when she started. No comment on whether or not I heeded her advice.)

The bottom line, though is that my memory is correct: my mother and I spent the summer of 1974 staying at the Strand Motel in Atlantic City.

So where was Mindy?

I see two possibilities:

  1. Mindy left Van Hook-Walsh in June 1974 then went into a different facility for the summer, and whoever provided Arc Alliance the list of schools forgot to list it (or someone forgot to include it).
  2. The “6/74” written in the 2017 report is simply a typo. Perhaps somebody inverted a “9” into a “6?

The second possibility makes the most sense to me given the scramble to get Mindy into Woodhaven a month or two later.

When my mother died in March 2004, I acquired all of her paperwork relating to Mindy. It is sitting in a folder in the filing cabinet just to my left as I type.

The answer may lie somewhere in those papers.

Now THAT would be a fascinating interrogation of memory.

For now, though, I leave you with this photograph taken in December 1979, almost three years after my parents separated (they would divorce two years later, seven months before my father’s untimely death at 46), the only photograph I have showing all four of us together.


Until next time…

[1] Much of the information in this and ensuing paragraphs is taken from Mindy’s Individualized Support Plan (ISP), which I receive annually as her plenary legal guardian. I was made co-legal guardian in 2002, when my mother was first diagnosed with terminal ovarian cancer, in what may have been the shortest legal proceeding in history. In my memory, the presiding judge from the Orphans Court of Delaware County took one look at Mindy and said “You’re her legal guardian. Next!”

[2] According to a June 27, 1974 (pg. 10) Philadelphia Inquirer story titled “Woodhaven program is working,” the facility was new as of April 1974.

A less-wicked-early look at the 2018 U.S. Senate Elections

In this May 2017 post, I took a “wicked early” look at the 2018 U.S. Senate (Senate) races. Depending on whether Democrat Doug Jones or Republican Roy Moore wins the December 12 special Senate election in Alabama, Republicans will control 51 or 52 Senate seats heading into the 2018 midterm elections. Democrats (46 or 47 plus two Independents who caucus with them, Bernie Sanders of Vermont and Angus King of Maine) would thus need to win a net gain of either two or three Senate seats to regain control of the Senate following the 2018 elections. At the same time, Republicans would need to net eight seats to have 60, a filibuster-proof majority.

Democrats (including Sanders and King) will defend 26 seats—10 in states won by 2016 Republican presidential nominee Donald Trump (five by double-digits)—while Republicans will only defend eight seats (only one won by 2016 Democratic presidential nominee Hillary Clinton). It had been 25 seats, but the impending resignation of Minnesota Democratic Senator Al Franken means that seat will also be up for election in 2018. I do not know whether whomever Minnesota’s Democratic governor, Mark Dayton, appoints to fill that seat (Lieutenant Governor Tina Smith is the frontrunner) will seek reelection.

Table 1: U.S. Senate Seats up for Election in 2018

Current Senator State Run 2018? 3W-RDM 2012 Margin Ave. Pre-2012 Margin (# Races)
Mazie Hirono HI Yes 34.3 25.2% N/A (0)
Bernie Sanders VT Yes 27.7 46.2% 33.1% (1)
Dianne Feinstein CA Yes 23.2 25.0% 15.5% (4)
Ben Cardin MD Yes 22.6 29.7% 10.0% (1)
Elizabeth Warren MA Yes 22.1 7.6% N/A (0)
Kirsten Gillibrand NY Yes 21.6 45.9% 27.8% (1)
Sheldon Whitehouse RI Yes 18.0 29.8% 7.0% (1)
Chris Murphy CT Yes 12.8 11.8% N/A (0)
Tom Carper DE Yes 12.5 37.5% 25.7% (2)
Maria Cantwell WA Yes 12.1 20.9% 8.5% (2)
Bob Menendez NJ Yes 12.0 19.5% 9.0% (1)
Martin Heinrich NM Yes 6.5 5.7% N/A (0)
Angus King ME Yes 5.9 22.2% N/A (0)
Debbie Stabenow MI Yes 2.2 20.8% 8.6% (2)
Amy Klobuchar MN Yes 1.5 34.7% 20.1% (1)
TBD MN ??? 1.5 N/A N/A
Tim Kaine VA Yes 1.5 5.9% N/A (0)
Tammy Baldwin WI Yes 0.7 5.6% N/A (0)
Bob Casey PA Yes -0.4 9.1% 17.4% (1)
Bill Nelson FL Yes -3.4 13.0% 13.5% (2)
Sherrod Brown OH Yes -5.8 6.0% 13.3% (1)
Claire McCaskill MO Yes -15.9 15.7% 2.3% (1)
Joe Donnelly IN Yes -16.3 5.7% N/A (0)
Jon Tester MT Yes -18.6 3.7% 0.9% (1)
Heidi Heitkamp ND Yes -29.4 0.9% N/A (0)
Joe Manchin WV Yes -35.5 24.1% 10.1% (1)


Dean Heller NV Yes 2.0 1.2% Appointed 2011
Jeff Flake AZ No -9.7 3.0% N/A (0)
Ted Cruz TX Yes -15.3 15.8% N/A (0)
Roger Wicker MS Yes -18.5 16.6% 9.9% (1*)
Deb Fischer NE Yes -25.8 15.5% N/A (0)
Bob Corker TN No -25.8 34.5% 2.7% (1)
Orrin Hatch UT ??? -33.1 35.3% 27.8% (6)
John Barrasso WY Yes -45.7 54.0% 46.8% (1)

   * Wicker won a special election in 2010 to fill Republican Trent Lott’s seat after he

     was forced to resign.

   Barrasso won a special election in 2008 after the death of Republican Craig Thomas

As you can see from the italicized-boldfaced elements of Table 1, there have been significant changes other than the Franken resignation.

  • Delaware Democratic Senator Tom Carper announced his reelection bid (and Wisconsin Democratic Senator Tammy Baldwin was already seeking reelection—my mistake), meaning every Democratic incumbent elected in 2012 will be on the ballot in 2018. This should help the Democrats significantly, as I explain below.
  • Two Republican Senators, Jeff Flake of Arizona and Bob Corker of Tennessee, announced they would not seek reelection. This could also help Democrats, as Republicans will no longer have the incumbency advantage in these races.
  • Republican Utah Senator Orrin Hatch may run for reelection in 2018 after all. This will not materially impact Democrats’ chances next year, as any Republican would be heavily favored to win this seat.


In my most recent post, I assessed the validity of 3W-RDM[1], my measure  of how much more or less Democratic a state is than the nation, by comparing it to two publicly-available measures, the Cook Partisan Voting Index (Cook PVI) and the fivethirtyeight.com “partisan lean” (538 PL).

Compared to Cook PVI (5.0) and 538 PL (4.8), 3W-RDM (5.4) had a slightly higher average absolute difference between projected and actual vote margins. In other words, if the Democratic presidential candidate in 2020 were to win nationally by 1 percentage point, I would expect Pennsylvania (3W-RDM=-0.4) to vote for that candidate by 1.0 – 0.4 = 0.6 percentage points. On average, however 3W-RDM misses by 5.4 percentage points in either direction, meaning the Democrat could win by 6.0 percentage points or lose by 4.8 percentage points.

In the rest of this post, I will use a combination of 3W-RDM, proposed 2018 partisan environments, and an estimated incumbency effect to examine the 2018 Senate elections more closely.


Just bear with me while I briefly explain how I estimated the electoral advantage of incumbency using data from the 2012, 2014 and 2016 Senate races (n=101; every state has ≥two Senate elections in the analysis)

Unless otherwise noted, electoral data are from Dave Leip’s indispensable Atlas of U.S. Presidential Elections.

I first calculated the difference between total Democratic and total Republican votes cast nationwide for Senate in each year, requiring some vote allocation decisions.

  • In 2012, there were 33 Senate races, including victories by Sanders and King. Because they caucus with Senate Democrats, I included their votes in the Democratic total, bumping up the Democratic margin in the total national Senate vote from 11.6 to 12.1 percentage points.
  • In 2014, there were 35 Senate races; both Oklahoma and South Carolina had an extra Senate election due to retirements (Republicans Tom Coburn and Jim DeMint, respectively). In the Alabama Senate race, Republican Jeff Sessions ran unopposed, receiving 795,506 votes; I excluded this election from the analysis. Minus those votes, Republicans won the total national Senate vote by 5.8 percentage points (down from 7.5 with Sessions’ votes).
  • In 2016, there were 34 Senate races. In California, every candidate to succeed retiring Democratic Senator Barbara Boxer ran in a “jungle” primary held on June 7, 2016; the top two vote getters, Democrats Kamala Harris and Loretta Sanchez, faced off in the general election (combining for nearly 12 million votes; Harris won with 61.6%). Rather than inflate the national Democratic margin by including superfluous Democratic votes, I used the sum of the primary votes cast for each party’s candidates (7 Democrats, 12 Republicans, 15 “other”) as the “Democratic,” “Republican” and “other” Senate votes from California. Louisiana also uses a “jungle” primary system, with every candidate running in the general election; if no candidate wins more than 50% of the total vote cast, the top two vote-getters compete in a runoff election. On November 8, 2016, Republican John Kennedy led the field with 25.0% of the vote, followed by Democrat Foster Campbell at 17.5%. In the December 10 runoff election, Kennedy won 60.7% of the vote, just shy of the 61.3% won by all eight Republicans on November 8. For simplicity, I used the runoff votes in my national Senate vote totals. These adjustments dropped the Democratic margin from 10.8 to 0.9 percentage points in 2016.

Next, I calculated a “projected” Democratic margin for all 101 Senate races by adding each state’s 3W-RDM[2] to either 12.1 (2012), -5.8 (2014) or 0.9 (2016), yielding an estimate of what the Democratic margin would be in a Senate matchup between a generic Democrat and a generic Republican.

Finally, I subtracted the actual margin from the projected margin. I would expect incumbents to fare better than projected, on average; of 76 incumbents seeking reelection (38 from each party), 56 (73.7%) beat their projected margins.

For example, in 2014, Virginia Democratic Senator Mark Warner ran for reelection. Virginia’s 3W-RDM was D-1.2, and the Democrats lost nationally by 5.8 percentage points. Adding -5.8 to -1.2 yields -7.0, meaning you would have expected a generic Democrat to lose by 7.0 percentage points (+/-5.4). Warner actually won by 0.8 percentage points, meaning he did 7.8 percentage points better than projected.

Table 2: Average Differences Between Actual and “Projected” Democratic Margins in Senate Races, 2012-2016, by Party and Incumbency[3]

Year DEM Incumbent GOP Incumbent No Incumbent
2012 D+5.1 (n=16) D-19.9 (n=6) D-3.4 (n=11)
2014 D+10.7 (n=15) D-0.2 (n=10) D+0.9 (n=9)
2016 D+10.4 (n=7) D-7.5 (n=22) D+3.9 (n=5)
Overall D+8.3 (n=38) D-7.5 (n=38) D-0.4 (n=25)

As Table 2 shows, while incumbency advantage varied by year (incumbent Republican Senators did an average of 19.9 percentage points better than expected in 2012, but only 0.2 percentage points better in 2014), incumbent Democratic and Republican Senators beat their projected margins by 8.3 and 7.5 percentage points, respectively, overall. By comparison, the actual and projected margins differed by an average of only 0.4 percentage points in the 25 Senate races with no incumbent (15 held by a Democrat, 10 by a Republican), an encouraging result given that these are the races that come closest to featuring a generic Democrat and a generic Republican[4].


Table 3 below displays estimated Democratic margins (+/-5.4; these are NOT formal confidence intervals) for all 34 2018 Senate elections under three conditions:

  1. Democrats win the total national Senate vote by 8.1 percentage points, the current “generic ballot” polling average calculated by fivethirtyeight.com.
  2. Democrats and Republicans split the national vote evenly.
  3. Democrats win the total national Senate vote by 11.6 percentage points, the median margin by which Democrats have won the total national Senate vote in the last seven elections in which they had a “wave” (which I arbitrarily define as ≥7.0 percentage points): 2008 (7.4), 1982, 2006, 2012, 1976, 1970, 1974 (15.7).

Table 3: Projected Democratic Margins in 2018 U.S. Senate Elections Under 3 Scenarios (+/- average 5.4 percentage point absolute value prediction error)

Current Senator State 3W-RDM Dems +8.1 Dems even Dems +11.6
Mazie Hirono HI 34.3 D+50.7

(+45.3, +56.1)


(+36.7, +47.5)


(+48.8, +59.6)

Bernie Sanders VT 27.7 D+44.1

(+38.7, +49.5)


(+30.1, +40.9)


(+42.2, +53.0)

Dianne Feinstein CA 23.2 D+39.6

(+34.2, +45.0)


(+25.6, +36.4)


(+37.7, +48.5)

Ben Cardin MD 22.6 D+39.0

(+33.6, +44.4)


(+25.0, +35.8)


(+37.1, +47.9)

Elizabeth Warren MA 22.1 D+38.5

(+33.1, +43.9)


(+24.5, +35.3)


(+36.6, +47.4)

Kirsten Gillibrand NY 21.6 D+38.0

(+32.6, +43.4)


(+24.0, +34.8)


(+36.1, +46.9)

Sheldon Whitehouse RI 18.0 D+34.4

(+29.0, +39.8)


(+20.4, +31.2)


(+32.5, +43.3)

Chris Murphy CT 12.8 D+29.2

(+23.8, +34.6)


(+15.2, +26.0)


(+27.3, +38.1)

Tom Carper DE 12.5 D+28.9

(+23.5, +34.3)


(+14.9, +25.7)


(+27.0, +37.8)

Maria Cantwell WA 12.1 D+28.5

(+23.1, +33.9)


(+14.5, +25.3)


(+26.6, +37.4)

Bob Menendez NJ 12.0 D+28.4

(+23.0, +33.8)


(+14.4, +25.2)


(+26.5, +37.3)

Martin Heinrich NM 6.5 D+22.9

(+17.5, +28.3)


(+8.9, +19.7)


(+21.0, +31.8)

Angus King ME 5.9 D+22.3

(+16.9, +27.7)


(+8.3, +19.1)


(+20.4, +31.2)

Debbie Stabenow MI 2.2 D+18.6

(+13.2, +24.0)


(+4.6, +15.4)


(+16.7, +27.5)

Amy Klobuchar MN 1.5 D+17.9

(+12.5, +23.3)


(+3.9, +14.7)


(+16.0, +26.8)

TBD MN 1.5 D+9.7

(+4.3, +15.1)


(-4.3, +6.5)


(+7.8, +18.6)

Tim Kaine VA 1.5 D+17.9

(+12.5, +23.3)


(+3.9, +14.7)


(+16.0, +26.8)

Tammy Baldwin WI 0.7 D+17.1

(+11.7, +22.5)


(+3.1, +13.9)


(+15.2, +26.0)

Bob Casey PA -0.4 D+16.0

(+10.6, +21.4)


(+2.0, +12.8)


(+14.1, +24.9)

Bill Nelson FL -3.4 D+13.0

(+7.6, +18.4)


(-1.0, +9.8)


(+11.1, +21,9)

Sherrod Brown OH -5.8 D+10.6

(+5.2, +16.0)


(-3.4, +7.4)


(+8.5, +19.5)

Claire McCaskill MO -15.9 D+0.5

(-4.9, +5.9)


(-13.5, -2.7)


(-1.4, +9.4)

Joe Donnelly IN -16.3 D+0.1

(-5.3, +5.5)


(-13.9, -3.1)


(-1.8, +9.0)

Jon Tester MT -18.6 D-2.2

(-7.6, +3.2)


(-16.2, -5.4)


(-4.1, +6.7)

Heidi Heitkamp ND -29.4 D-13.0

(-18.4, -7.6)


(-27.0, -16.2)


(-14.9, -4.1)

Joe Manchin WV -35.5 D-19.1

(-24.5, -13.7)


(-33.1, -22.3)


(-21.0, -10.2)



Dean Heller NV 2.0 D+2.6

(-2.8, +8.0)


(-10.9, -0.1)


(+0.7, +11.5)

Jeff Flake AZ -9.7 D-1.5

(-6.9, +3.9)


(-15.0, -4.2)


(-3.4, +7.4)

Ted Cruz TX -15.3 D-14.7

(-20.1, -9.3)


(-28.2, -17.4)


(-16.6, -5.8)

Roger Wicker MS -18.5 D-17.9

(-23.3, -12.5)


(-31.4, -21.6)


(-19.8, -9.0)

Deb Fischer NE -25.8 D-25.2

(-30.6, -19.8)


(-38.7, -27.9)


(-27.1, -16.3)

Bob Corker TN -25.8 D-17.6

(-23.0, -12.2)


(-31.1, -20.3)


(-19.5, -8.7)

Orrin Hatch UT -33.1 D-32.5

(-37.9, -27.1)


(-45.8, -35.2)


(-34.4, -23.6)

John Barrasso WY -45.7 D-45.1

(-50.5, -39.7)


(-58.6, -47.8)


(-47.0, -36.2)

Let me be very clear: the margins presented in Table 3 are NOT predictions[5], they are estimates of what the Democratic margin would be in a Senate race given the state’s partisan lean, the national political environment and whether an incumbent is running; candidate quality and campaigns can (and will) result in different margins on Election Day 2018. These estimates serve as helpful context for anticipating what may happen in these elections in 2018, which I address next.

Safe incumbents. In the worst-case scenario for Democrats (even with Republicans nationally; every Democrat loses 5.4 percentage points), nine Democrats (plus Sanders) would still be projected to win by at least 10.0 percentage points: Mazie Hirono, Dianne Feinstein, Ben Cardin, Elizabeth Warren, Kirsten Gillibrand, Sheldon Whitehouse, Chris Murphy, Tom Carper and Maria Cantwell. New Mexico’s Martin Heinrich and King should also be considered safe, projected to win by 8.9 and 8.3 percentage points, respectively, in this scenario[6].

New Jersey Democratic Senator Bob Menendez would be in this safe category but for the fact that he has been indicted for bribery, although his trial ended in a hung jury. One estimate of the impact of scandal is that it shaves 12.6 percentage points off an incumbent’s reelection margin. In the worst-case scenario, though, Menendez would still be projected to eke out a 1.8 percentage point victory. Absent a top-notch Republican challenger, I would consider Menendez safe…for now[7].

In the worst-case scenario for Republicans (lose by 11.6 percentage points nationally; every Democrat gains 5.4 percentage points), three Republicans (plus whoever the Republicans nominate in Utah) would still be projected to win by at least 9.0 percentage points: Roger Wicker, Deb Fischer and John Barrasso.

Let’s consider these 16 races Safe.

Less-safe, though still-favored incumbents. Five Democrats would still be projected to win by at least 2.0 percentage points in their worst-case scenario: Debbie Stabenow, Amy Klobuchar, Tim Kaine, Tammy Baldwin and Bob Casey. Stabenow, Baldwin and Casey represent the three states (Michigan, Wisconsin, Pennsylvania) Trump won by a combined 77,744 votes, allowing him to win the Electoral College. Klobuchar and Kaine, the 2016 Democratic vice-presidential nominee, represent states (Minnesota, Virginia) that are only slightly more Democratic (D+1.5) than the nation as a whole. Still, even with only a modest Democratic performance in 2018, these five Democrats should win reelection fairly easily.

Texas Republican Senator Ted Cruz would still be projected to win by 5.8 percentage points in his party’s worst-case scenario, making him likely safe. It is worth pointing out, however, that the only public poll of a match-up between Cruz and his likely Democratic opponent, U.S. Representative Beto O’Rourke, showed each receiving only 30%, with an astonishing 37% undecided. I have argued elsewhere that Democrats would be wise to look to southwestern states like Texas and Arizona for future electoral success. If Democrats are having a historically good night next November, Cruz could be in trouble.

Overall, that is an additional six seats that are Likely.

Bill Nelson and Sherrod Brown (plus Al Franken’s seat). As incumbent Democratic Senators representing Republican-leaning states (Florida, Ohio) in a strong Democratic year, my best estimate (Democrats win by 8.1 percentage points nationally, no error in 3W-RDM estimates, Democratic incumbency advantage of 8.3 percentage points) is that each would beat a generic Republican opponent by 10-13 percentage points (+/-5.4). Any Democrat Senate nominee in Minnesota would be expected to do similarly well.

However, should Republicans do far better than expected in 2018, Ohio continues to trend sharply Republican and/or strong potential challengers like Florida governor Rick Scott or former Minnesota governor Tim Pawlenty run, these three Democratic seats could be vulnerable. I would also note Nelson and Brown appeared to have no incumbency advantage in 2012, when their actual margins differed from projected margins by an average of only -0.1%. And in the worst-case Democratic scenario, a generic Republican would win these races by low-single-digit margins.

At the same time, Hurricanes Irma and Maria have displaced thousands of Puerto Ricans—who tend to vote Democratic—to Florida, which could boost Nelson in 2018.

These races, then, are “canaries in the coal mine” for 2018. If polls in the spring and summer of 2018 show these three races tied, or Republicans ahead, that would be a bad sign for Democrats. But if Democrats open up substantial leads in these three races, Democrats could be on the verge of an historic wave.

For now, let’s call them Lean Democratic.

Five Trump-state Democrats. Claire McCaskill, Joe Donnelly, Jon Tester, Heidi Heitkamp and Joe Manchin—representing five states averaging 23.1 points more Republican than the nation—should be in deep trouble. Even in the best-case scenario for Democrats, the latter two would be projected to lose by an average of 7.2 percentage points.

My best estimate is that McCaskill, Donnelly and Tester would be in toss-up races against a generic Republican, while Heitkamp and Manchin would lose by an average of 16.0 percentage points.

And this may well be how Election Day 2018 plays out for Democrats—a loss of two seats from these five.

It is noteworthy, however, that in 2012 (an excellent year for Democrats), these five Democrats beat their projected margins by an average of 11.6 percentage points. These five Democratic Senators clearly understand how to win in very Republican states.

Manchin did especially well, beating his expected margin by 26.8 percentage points! Indeed, as West Virginia was trending sharply Republican (D+9.2 to D-35.5 in just six elections), Manchin won four statewide elections for governor (2004, 2008) and Senator (2010, 2012) by an average of 27.0 percentage points. It is telling that Jim Justice, elected governor of West Virginia as a Democrat in 2016 only to become a Republican in 2017, endorsed Manchin’s reelection.

Based solely on the best estimate projections, though, I will call Indiana, Missouri and Montana toss-ups and North Dakota and West Virginia Lean Republican, though it is plausible that Democrats hold all of them.

Bob Corker’s seat. Tennessee’s strong Republican lean (D-25.8) means that under the best-case scenario for Democrats, a generic Democrat would lose to a generic Republican by 8.7 percentage points, with a “best guess” loss of 17.6 percentage points. This would seem to be a seat Republicans would keep fairly easily.

However, the announcement that former Democratic governor Phil Bredesen would run could alter that assessment. Bredesen won his first election in 2002 by only 3.1 percentage points before winning reelection in 2006 by an astonishing 38.9 percentage points.

Notes of caution, however, come from the experience of other Democrats—Bob Kerrey (Nebraska) in 2012; Russ Feingold (Wisconsin), Evan Bayh (Indiana) and Ted Strickland (Ohio) in 2016—who ran for the Senate at least four years after last holding statewide office. Each lost, and by an average of 12.4 percentage points.

For now, I would put Tennessee in the Lean Republican category.

Dean Heller and Jeff Flake’s seat. Excluding the special Senate election in Alabama, these are the two seats Democrats are most likely to flip.

Heller is the only Republican Senator seeking reelection in 2018 in a state Clinton won in 2016. That my best estimate has Heller losing by “only” 2.6 percentage points is due solely to his incumbency: a generic Republican (say, Danny Tarkanian, who is mounting a serious primary challenge to Heller). In the worst-case scenario for Democrats, Heller would be projected to win by 10.9 percentage points.

Democrats are also boosted by a strong likely nominee: first-term U.S. Representative Jacky Rosen. The Democratic lean of the state in a Democratic year, even with Heller’s incumbency advantage, suggests this race is more likely than not to flip to the Democrats.

I will call it Lean Democrat.

Flake, finally, was already facing a strong challenge from Democratic U.S. Representative Kyrsten Sinema (and a strong primary challenge from Kelli Ward) although as an incumbent, my best estimate would have projected a 9.0 percentage point win.

As an open seat, however, this looks like a close race (best estimate: Republican wins by 1.5 percentage points, +/-5.4), and I thus consider this race a Toss-up.

Conclusion? As this map (courtesy of 270towin.com) shows, the partisan landscape for 2018 shows Democrats favored in Nevada, Republicans perhaps favored in North Dakota and West Virginia, and the outcome uncertain in Alabama, Arizona, Indiana, Missouri and Montana.

2018 projected Senate map 12-11-2017

There is a plausible path for Democrats to net three (or more) seats, and there is a (less) plausible path for Republicans to net eight seats.

The likeliest outcomes are between Democrats net losing five and gaining three, with a midpoint of Democrats net losing one (lose two of the five Trump-state Democratic seats plus Alabama and Arizona, win Nevada, hold Franken’s seat).

But 18 of the 34 races remain less than safe for either party…and that is exciting.

Until next time…

[1] It is the weighted average (1-2-3) of the difference between a state’s presidential margin (Democratic % of total vote minus Republican % of total vote) and the national presidential margin over the last three presidential elections.

[2] For the 2012 Senate elections, 3W-RDM was calculated using presidential election data from 2000, 2004 and 2008, and for the 2014 and 2016 elections, data from 2004, 2008 and 2012 were used.

[3] For Sanders and King in 2012, the margin is Independent% – GOP% (Sanders had no Democratic opponent; King beat Republican Charles Summers 52.9- 30.7%, with Democrat Cynthia Gill finishing 3rd with 13.3%). In the 2014 South Dakota Senate race, Independent Larry Pressler (a former Republican Senator) won 17.1% of the vote, possibly cutting into the margin (20.9 percentage points) by which Republican Mike Rounds beat Democrat Rick Weiland. In the 2016 Alaska Senate race, incumbent Republican Lisa Murkowski won with 44.4% of the vote, ahead of two Independents (Joe Miller [29.2%) and Margaret Stock [13.2%], with Democrat Ray Metcalfe finishing 4th with 11.6%. I used the margin between Murkowski and Metcalfe (-32.7 percentage points); using the margin over Miller (a Tea Party Republican) seemed inappropriate and would not have markedly changed the results.

[4] I considered using -0.4 as the baseline for my “incumbency advantage” average, rather than 0.0, but I decided it was close enough to 0.0 to be statistical noise.

[5] I don’t really expect Mazie Hirono to win by 51 percentage points, though she could. Ditto John Barrasso winning by 45 percentage points.

[6] Were Republican governor Paul LePage to run against King that could make this race closer, though that is likely if LePage’s current 42% approval rating (against 52% disapproval) holds.

[7] Of course, if Menendez’ polling numbers started tanking in 2018, he could well resign (or announce he was not seeking reelection). A Democratic replacement appointed by incoming Democratic governor Phil Murphy—or a Democratic nominee—would face a worst-case projected margin of D+6.1.