So, as a result of the presidential election, I am curious if anyone else is wondering about the future of GOTV? The short version of this is: the vaunted GOTV capabilities of the Democratic Party, based in meticulous and scientific tracking of voting records was defeated by a Trump organization that effectively had no on-the- ground GOTV effort except in a couple of places. What does this say about current and sacrosanct Dem party theories of voter mobilization, locally, statewide and nationwide?
I would make one observation: it seems like the current model of GOTV is based on the idea that our voters aren’t particularly motivated, so you have to work really hard to get your voters to actually go vote by individually tracking and targeting them through endless email, door knocking, phone calls, etc. This entails an enormous effort involving many thousands of activists And huge hauls of cash. But the Trump model seems to be the opposite: give your voters the proper motivation to vote and they will go vote. On their own. Without being begged or badgered. Without the coaxing of, might we say, a Nanny Party?
So, has the Dem GOTV model evolved in its supposed sophistication out of “facts on the ground,” in other words, uninspiring politics and candidates, that have led to a demotivated base of voters? Have we unintentionally normalized the bland, safe, uninspiring agenda not relevant to a large swath of voters? We lionize all of the wonderful people who travel weekend after weekend to do GOTV in NH, etc. but what if that is all just busywork focused on the wrong targets? What if instead of targeting voters, we should have been targeting our own thinking and what we stand for? If so, then the solution, as some of the commentary on the election suggests, is to stand for something and run candidates that voters really want to turn out and vote for. Like Trump did. And by the way, it can be argued that not long ago, Scott Brown did that in defeating Martha Coakley . . . Etc.
As a town committee chair for a decade, I was involved in endless debates about the effectiveness of GOTV efforts before, during, and after every election. But, perhaps we should have been focused on a different target altogether.
Thoughts?
Rather than the Trump campaign. There’s a number of other nationally funded GOTV organizations as well, including in MA.
Conventional wisdom says it’s worth 2-3%. Someone wasn’t paying attention in WI, PA, MI though. Would have helped there.
http://www.alternet.org/election-2016/trumps-get-out-vote-effort-not-even-close-clintons-experts-say
Is that why you posted it?
Because her campaign trusted the algorithms over the observed reality on the ground. Of course GOTV doesn’t work in that situation. What this does suggest is that we should stick to traditional, human GOTV and not let the computers second guess the volunteers/
It’s a triage model wherein voters fall into one of the following categories:
Always vote, always vote Dem – leave them alone
Always vote, sometimes vote Dem – try to persuade them your way
Always vote, never vote Dem – don’t bother
Sometimes vote, always vote Dem – highest GOTV priority
Sometimes vote, sometimes vote Dem – need persuasion and GOTV
Sometimes vote, never vote Dem – don’t bother
Votebuilder will give voting history and ID calls/knocks will give you a sense of persuadability. You want to GOTV your likely voters except don’t waste your time on superDems who will take care of themselves. Depending on the kind of candidate you may want to create new voters as well, but that takes much more advance planning and effort. It will require registration drives and possibly education on basic voting mechanics.
Trump said early on that he ‘didn’t want anybody getting rich off this campaign”. Kushner said, “We played Moneyball. We researched what states would provide the best ROI for electoral votes.” Traditional GOTV centers on connections, activists, lists, etc. Trump did none of that but went after EC votes rather than gross totals, just like a proxy fight or hostile takeover.
This kind of surgical precision violates a lot of baby kissing tenets, but worked.
Parenthetically, it’s what makes the popular vote angst so futile. If that is what won the election,an entirely different strategy would have emerged.
…so if you want to win certain states it stands to reason you would have solid GOTV operations at least in those states.
There weren’t a whole lot of Clinton GOTV operations in Massachusetts or in Oklahoma. How is what you just described any different than traditional POTUS campaigning and GOTV efforts?
On Georgia, Arizona and other states while he did invest in OH, MI, PA, and WI. I said early on those four states could be peeled off if he invested all his time and energy there and he did so, while retaining Republican states like NC, AZ, and GA and barely holding onto Florida where he spent more time than Clinton. It turns out time mattered more than money in these particular states.
Besides, you have to admit the temptation was great to send a clear and convincing message to ourselves and the world that America would absolutely shut down such a Dangerous Unqualified Misogynistic Bigot.
A lot has been made of her failure to go to Wisconsin at all. She didn’t spend any time in red states during this period. Perhaps surrogates did, and perhaps they spent money in such places, but she didn’t go there.
Clearly with hindsight not going to Wisconsin at all and not going to Michigan until the last minute was not wise. The problem was of all the places she visited down the stretch, she lost all but NH and NV. That’s not good. It’s hard to argue that an increased physical presence in the upper Midwest would have helped her there, where it didn’t seem to help anywhere else.
To summarize:
6 trips to Florida
4 trips to NC
4 trips to PA
3 trips to OH
2 trips to MI
2 trips to NH
1 trip to each of NV, AZ and IA
Here is her campaign appearance schedule down the stretch of the campaign (i.e., after the last debate on 10/19):
10/20 Nothing
10/21 Cleveland
10/22 Pittsburgh and Philly
10/23 Multiple stops in NC
10/24 Manchester (w Senator Warren)
10/25 Broward County FL
10/26 West Palm Beach and Tampa FL
10/27 Winston Salem and Greensboro NC
10/28 Cedar Rapids and Des Moines IA
10/29 Daytona Beach and Miami
10/30 Broward County FL
10/31 Kent St and Cincinnati
11/1 Multiple stops in FL
11/2 Las Vegas and Tempe AZ
11/3 Multiple stops in NC
11/4 Pittsburgh and Detroit
11/5 Philly and Broward County FL
11/6 Philly, Cleveland, Manchester
11/7 PA, MI and NC
I know she came to Massachusetts in August but she stayed far, far away from Framingham, Franklin, and Fall River…nope. She rubbed elbows with CHER in P-Town at a big fundraiser.
Hindsight it 20/20 but if she had visited the aforementioned towns instead of the glitz and glamour, she’d be picking out new drapes for the Oval Office.
Are you seriously saying if she visited more MA communities she would have won? You do understand how the Electoral College works and know she already got MA to the tune of 61%, right? I share pbrane’s skepticism about how helpful physical presence is, but she did hold FREE rallies in many states as he listed above.
Your implication is that HRC’s team did. Not 538, not kos, but HRC. What evidence is there of that? A feign into Utah, Arizona, and Georgia? What percent of her efforts went there? 1 percent? Less? You make it sound like she spend 3 days a week in Oklahoma.
In sports as in politics, when you win, every decision was genius. When you lose, everything you did was wrong. Except that’s nonsense.
Jonsax writes:
As much as I am thankful for all the money raised by the Pan Mass Challenge and other such activities, it seems inefficient. Take a few thousand people on bikes from here to there….and that raised money for cancer research? Wouldn’t it be more appropriate to ask all those people to work with cancer patients, cook them meals, do simple home repairs and cleaning for them while they are getting chemo treatments etc? And at the same time, announce their work and collect $$ from friends? Is this about cancer or is this about the bike ride? And before anyone jumps on me, I donate to several riders each year and I’ve participated in similar rides over the years.
Is all the time and effort and expense of canvassing well spent? Is there a better way? Phone banking looks to be a dying art with the extinction of the home phone. A couple of years ago, I personally canvassed my entire district for a candidate. I spent evenings and weekends hunting down doors to knock on. I think I may have actually spoken to maybe a dozen people.
Yeah, I hear that. It’s also hard to motivate canvassers when candidates won’t even visit your town, or supply a lot of support. The canvassers feel taken for granted, then ignored until needed next time.
GOTV is vital for low-information campaigns. It’s what won on Question 2.
But nobody needed to be told who the candidates were for president, or when Election Day was.
I believe that high-information campaigns move the needle maybe 5 points. The rest is already baked in.
GOTV Is as important than message, with the following proviso: Regarding the latter, the messenger is as important (if not more so) than he message.
Trump had a comprehensive neighbor-to-neighbor ground game, and Clinton didn’t have squat, credibility-wise.
The last thing any campaign needs are canvassers – or phone bankers – that aren’t from the immediate (repeat immediate) areas as their contacts. Motivating and organizing these folk can be done, but there has to be a social and civic component Trump’s people supplied that.
Democrats have to face the fact that every aspect of Clinton’s operation was objectively sub-par, in the face of an extremely well-planned and executed Trump campaign, particularly on the ground.
It need not be the case that everything Team Trump did was superior than everything Team Clinton did for Mr. Trump to win the election.
Because your last sentence is clearly nonsense:
Nonsense, and
nonsense.
Team Trump won. I’m not arguing that Team Clinton was perfect or that Team Trump was an unmitigated disaster. I am arguing that both Team Clinton and Team Trump did some things well and other things poorly.
But, to Paul’s point, I think Trump’s message overwhelmed any GOTV efforts, and his message was (usually) relentless.
…Trump had people on the ground training and organizing their neighbors on his behalf. Here is an example from Florida.
Hidden in plain sight was arguably the most comprehensive grassroots GOTV training effort in American electoral history.
As in Florida, so went Pennsylvania, Ohio, Wisconsin…
As I’ve been saying for months, in much of the contested States, the ONLY effective grassroots field operated on Trump’s behalf.
All I kept hearing was that Trump had virtually no staff and no offices in key states and that Clinton had him beaten by a mile by these metrics, to the point where some wondered whether Trump were taking this at all seriously.
There were a lot of volunteers enthusiastic about his candidacy as opposed to being paid to pretend or feeling resigned to it like most of her volunteers. Wide enthusiasm gap.
Yes, volunteers are the lifeblood of campaigns, but they have lives and degrees of expertise that vary widely. Surely you know that you need full time field directors, field organizers, etc. to make the volunteer operation work. It sounds like you never stepped foot in a Clinton field office. If you had you would see there was plenty of enthusiasm to go around among the paid and the unpaid. Lack of enthusiasm for Clinton in both the primary and the general is nothing more than another lazy and inaccurate media narrative.
If the extent of the field offices that you experience is MA and NH, then you might think there was real excitement. I was in field offices in many of the east coast battleground state and there was zero energy. I even had a hard time finding an open office in the Tampa area in September and October, a Democratic area where turnout was low because there was zero GOTV.
n/t
People on the ground – including electeds – tried to warn the Clinton Campaign, but were ignored (at best). As mentioned earlier, the sheer scale of this became obvious to me in February when I was seeing Trump signs all over the place in Western Pennsylvania, in the context of an environment wherein local Democratic players were contacted by neither Clinton’s nor Sanders’ people.
While signs don’t vote, the sheer number and spread of the things -particularly on the lawns of registered Democrats – indicated pretty decent organizational prep – and this was before the primary.
When the campaign work in done in peoples’ homes, and every social nexus (taverns, social clubs, churches, etc) are used as an outreach venue; you don’t need campaign offices.
And such formal Clinton structures as existed were staffed by arrogant out-of-town player wannabees who seemingly went out of their way to condescend to locals… to Trump’s advantage.
Out-of-state phone banking is another story; suffice it to say that it accrues similar results.
Insofar as the media is concerned, there is a presumption of competence that is not always supported by the facts. As a rule, unless a given reporter is spoon fed, he would miss an earthquake when standing in the epicenter.
Hmm…
The ones I’ve been involved with always have in the script something like, “I’m calling our neighbors here in (insert name of community you are calling here)…” so, yeah, maybe a little white lie, but you never say I’m here in MA calling people in Ohio today. I’m just flabbergasted that you seem to be saying people didn’t know the basics. HOWEVER, you have to come back to the polls. The polls in WI, MI, and PA consistently showed Clinton leading right up until the day before and these are people trained to trust polls usually with good reason. For someone like me, regardless of the anecdotes I hear I will fall back on the idea that the plural of anecdote is not data, and believe that if the polls show us up why fix what ain’t broke.
I know what my neighbors sound like. I know many of my neighbors personally. People from Ohio do not sound like people from MA. People in GA recognize people from MA immediately and vice-versa.
One of the areas I did professionally for many years was systems to support telephone call centers. I fear you minimize both the ease with which people recognize “little white lies” and the hostility those generate.
Scripts are needed when the callers know nothing about the voter being pursued. But if you are a Trump supporter talking to a friend at church or a bar or in the workplace, you don’t need a script. You won’t call them “Joseph” when you know that they are “Junior” in real life. Many of us have nicknames and when my phone rings and the caller asks for me by my legal name, not TBD, I know it is a stranger who does not have my interests at heart trying to sell me something, whether it is life insurance or a candidate.
In a previous work project, I had the opportunity to visit a professional call training center in India. Operators are trained for days before they are allowed to take an incoming call from a person with a warranty problem or trying to make a reservation. After the initial training, they do a daily touch up by reading the local news, especially sports and weather, for the time zones they will be receiving calls from.
When you organize among your real social cohort, you will be using the same accent, not an MA accent calling into NC.
Nobody is suggesting that we dump phone banking or canvassing but anybody with experience in the real world of organizing know that it pales in comparison to peer=to-peer organizing.
I know in particular the Clinton campaign in NH adopted the neighborhood team approach that worked well for Warren. Of course Clinton did win NH so maybe that’s the variable and other states did not do that.
…and limited by sampling techniques. What occurred was a classic Truman-Dewey error, due in part to the Bradley Effect, and the fact that Trump’s campaign brought in voters that were largely off the grid from the polling models.
Re: Phone Banking.
Total studies are somewhat inclusive, but:
The single most important positive variable, in my experience, is an organic connection with the targeted voter. The script you cited generally doesn’t work, for the following reasons:
1.) “Calling our neighbors” as an intro begs the question of specific geography. Old style campaigns knew this; hence phone bankers always called people in their neignborhoods; at most people in adjacent neighborhoods. At the scale of a national campaign people never made out-of-state calls. (Your accent can adversely affects your credibility.)
2.) One of the biggest irony of politics in the social media era is that personal ties matter more, credibility-wise, than at any time since the pre-radio era.
3.) In an era of cell phones and caller ID, unsolicited calls from unknown numbers are considered to be nuisance calls by many voters; insofar as many lower-income voters are concerned, they are often thought to be from collection agencies.
4.) As a corollary to Point 3, many voters from low and low-moderate incomes (the majority of whom, for what it’s worth voted for Clinton) have prepaid phone plans, some of which means they pay for incoming calls.
This is not just a national phenomenon; the same dynamic operated adversely to Coakley in 2014. Furthermore, I’ve worked in local campaigns that were sunk by their phone banking.
Hence some suggestions, based upon experience:
Presuming a politically-literate field organization, phone bank lists should be generated, based upon an opt-in method derived from local canvassers. (Would you mind being contacted by us? Would you mind giving us your phone number? What would be the best time to call you?)
Callers should specifically identify themselves by name and immediate neighborhood: “Hello Mr. Smith, I’m Jim Johnson; I live over on Tremont Street. Is this a good time to speak to you…”
Back to polling, the problem goes to the greater issue of depending solely on models with flawed premises, and ignoring empirical data that refutes the model. Clinton’s campaign dealt with voters, based upon aggregates; Trump’s campaign identified individual supporters, and got them to the polls.
The election aside, there is the matter of the institutional class bigotry that is hard-wired into Democratic institutional culture. Clinton’s campaign concentrated upon reaching out to Republican-leaning suburban voters, to the exclusion of working class white voters and black voters of all classes. This pulled a lot of Trump voters to the polls.
Post-election, there is the issue of scapegoating white working class voters in isolation, despite the fact that:
Personally, I don’t care whether you live around the corner or in California, especially for a race that impacts the entire country, though it might seem odd if a CA person were trying to persuade me to vote a certain way for state representative. I’m certainly not going to discount your message based on your accent. What I recall from 1948 is that polls stopped in early October whereas polls this year were taken right up to the last weekend (though I’m not sure we’ll ever know exactly to what extent Comey’s last minute comments made).
The issue is one of credibility.
In the context of major declines in civic culture the tendency to trust only who you know, or who can establish personal commonality- never absent – is magnified.
A larger issue than low-information voters is that of low-trust voters, and this is a cross-economic, cross-cultural, and cross-generational factor in contemporary America.
Out-of-state phone bankers are at a cultural disadvantage from the time the recipient of the call picks up – assuming s/he does pick up, because the phone banker is often making a nuisance call, from the voter’s perspective. A recognizably out-of-state accent or other aspect that does not establish and reinforce some sense of commonality (and shared political viewpoints only work among activists) create what I call the presumption of assholetry response; as in “Who does this asshole think he is by interrupting my dinner?”
Or “my TV show”, or my anything.
When that same voter has been recruited by a physical neighbor who supports an opposing candidate, and when that support is reinforced socially within the voter’s geographic and social space, all that happens is to harden opposition to the phone bank’s candidate, via immediate hostility to the individual phone banker.
Innsofar as potential supporters are concerned: In the absence of physical contact by a trusted person, such phone banks often operate as political disincentives to vote at all; as happened with many black voters.
Back (again) to polling: Very few people who actually do this for a living were surprised. Pasted below is a layperson-friendly preemptive caution from October:
That doesn’t make any sense. It seems to me online respondents are very self-selecting with no filter for representativeness. I can’t count the number of times I’ve been asked to go to some website and vote in this or that online poll. Or is that not what this is? Are specific chosen people being emailed with the same precision that specific chosen people are called to create a random sample? I believe when Tip O’Neill said all politics is local he was mostly referring to House races which was his wheelhouse. Members of and candidates for Congress need to be attentive to how Washington actions impact locally and what local issues need federal attention. In our system that is as it should be. Even in presidential races candidates do well to make a local connection to the issues they are speaking to. I do not understand why nor believe it appropriate for that to be used as an excuse to retreat into parochialism. I don’t know how to explain it, but the idea that I should only trust those who are most like me is a very, shall we say, “foreign” one. Keep in mind, even Nate Silver, who IIRC correctly called every single state accurately in 2012, got this wrong, so while this election has taught us to reconsider some of what we thought about data and voter behavior, we can’t really be blamed for trusting information that has been reliable in the past.
The issue of internet versus telephone polling was not germane to the authors’ purpose, which was to caution people with little or no training in survey mathematics to avoid taking survey results as empirical fact without knowing the internal limitations of methodology and interpretation. The author further points out that even pollsters can miss what’s actually out there.
If Clinton’s person-to-person field data had confirmed the polling data, that would have been one thing: It didn’t.
The Clinton campaign had ample warning from the ground about what was happening in time to address the matter. For whatever reason, decisions were made to ignore empirical data (supplied by people who knew damn well how their neighbors intended to vote) in favor of their model.
This was a level of magical thinking that I tend to equate (in my kinder moments) with climate change denial.
The voter data was there, it was simply ignored.
The context, not that it matters, of Tip O’Neill’s quote, was that all elections especially national elections, are local.
…how to not let sampling bias creep into their universe in the first place.
All they can do is try to minimize it. It’s a matter of ongoing concern within the American Association for Public Opinion Research.
Furthermore, because variables can be both cultural (like low public trust) and structural (the rise of cell phone-only households and caller ID), the parameters for accurate results are always changing.
The map is not the road. Arguably the map isn’t even the map anymore when the topology is always changing.
In fairness, pollsters have been consistently and proactively honest about the inherent limitations of their craft.
Consider that Nate Silver gave Trump a slightly greater chance of winning the election than the Chicago Cubs had of winning the World Series…
Can’t be completely avoided, hence the concept of a margin of error, but in the classes I took for polisci undergrad and political management grad we were taught that the goal was to get a sample that is representative of the population on a whole host of socio-economic and cultural factors and if you don’t get the contact rate you want you add contacts in certain segments. It’s just that many of these outfits have been around for decades and have generally been reliable predictors so it definitely comes as a surprise that it suddenly wasn’t the case this time. They were also pretty consistent whereas if there were lots of disagreement among pollsters that would have been a more obvious red flag.
Consider this:
…but I’m actually surprised it took this long to be shut out of Southern state legislatures.
…at your individual vs. aggregate comment. I was under the impression Clinton inherited Obama’s much-vaunted micro-targeting operation. Also, I should clarify that while I used calls as an example, I know that evidence suggests canvassing is more effective in part due to drawbacks on calls that you mention. I’m very skeptical of an opt-in model because the only people I suspect would say yes are junkies like us who probably have our minds made up anyway, but just want to be polite. People hate getting called (and as a corollary I hate calling close to E-Day because voters scream at you about how many calls they’ve gotten), but part of me wants to say look, you live in a democracy so this is part of the deal.
… and for some time thereafter, if at all.
…I also know that people within the Democratic datatech world were making the same complaints as Democratic field folks insofar as arrogance and political cluelessness on the part of the Clinton campaign were concerned.
Re: Opt in models for phone banking.
I’ve found it successful, for the simple reason that people like to be asked. If they volunteer their phone numbers – and in my experience most identified supporters will – they are receptive, not only to GOTV calls, but also volunteer recruitment.
Ironically civilians, also in my experience, are better contacts as volunteers because they have, on balance, greater political work ethics…
If a campaign told me to say or imply I was from the target calling area, I would go off script for the reasons Tom states.
I would add, though, that anyone who claims to be offended by the source of the call wasn’t voting for your candidate anyway.
At least for me on the receiving end, it’s usually easier to tell that someone is reading a script than that their accent isn’t from around here. I know after a handful of calls to get comfortable, I revert to a more conversational tone and just make sure I make the points that I need to.
As Kushner said in Forbes, Trump rejected that approach. Two things may have happened with the Clinton effort.
Domke Syndrome – I am picking on poor Todd because he is often cited as a GOP expert after years of being away from campaigns and ground level. How many of her team were the same old, same old warriors that people don’t even hear anymore because the speeches are the same with the candidate de jour mentioned on an Insert Name Here basis? Anyone Trump used would be relatively new, as the GOP Death Star Establishment was shunning him as well. People NEW to the process generate interest.
Peer-to-Peer – IMO, one of the most effective ways to move votes in the age of Robo-calls and mass internet marketing ar what used to be called Dear Friend cards – a postcard message saying, I’m voting for Fred, to be signed and snail mailed by volunteers to their friends. This is not a talking head, or a far away presence, but somebody you actually know. Trump did a LOT of p-t-p stuff, and in a tight and nasty race, it may have been unusually influential.
Neither of these need staff or offices. Just people enthusiastic about a candidate. And they are both damn cheap, too. But the bipartisan political worker class tells every candidate about how they don’t work for exactly that reason. Trump may have won because political professionals were not a big part of his campaign.
and the antiquity of phone banks and canvassing cannot be ignored. Philippines president Rodrigo Duterte spent less than his opponents and he won, largely with a network of social media. My guess is that Trump took notes on this.
I visited them.
They just happened to be in people’s private homes or the occasional law or real estate office.
A slight qualification to your last sentence: Political professionals were in fact a big part of Trump’s campaign (and they will be equally important to his administration) , they just happened to be good at their job.
And unlike their Democratic counterparts, they didn’t self-promote or advertise their presence. They simply did their jobs.
Trump simply organized in plain sight in a perfect storm environment’ in that the Clinton people ignored what was happening right under their noses.
Despite the fact that Clinton won the most votes, Trump won more votes than just about anyone imagined he could. The very best pollsters thought Trump had a ceiling he couldn’t get beyond, which was meant to be well (many millions) short of his 60 million. He did this without anything close to the organized GOTV efforts of Clinton and the Dem party.
There is a lot of discussion below about proper deployment of phone banking and canvassing. But is this the right conversation? Where is the evidence that these are the things that win elections? And, as I suggested earlier, is it possible, if these things do win elections, that this is only because our voter base is so demotivated that you have to push them relentlessly to vote?
Here in MA, over the last decade of election cycles, our local committees have been almost forcibly prevented from deploying yard signs and doing visibilities because “it has been decided” that these activities do not increase voter participation or turnout. I have thought that this was a very unwise and counter-intuitive position. Imagine telling Patriots fans (or the Patriots organization!) that showing their colors: having and waving banners, wearing hats and shirts emblazoned with their team, etc., is stupid and doesn’t generate and spread fan enthusiasm. Imagine if the Patriots had to hire professional GOTF(fans) organizations in order to fill the seats. You’d say they were in deep trouble.
Well, I have read many reports from Dems in MI, WI and PA who say that the warning signs they literally(!) saw everywhere were oceans of Trump signs in communities all over the state. Trump apparently found yard signs (waving the flag, showing your community’s support) to be a very, very effective way for people to signal their identifying with the team, which in turn, possibly gave other voters to courage to do so too.
The enthusiasm gap was clearly much much bigger than anyone thought. And even more important, it seems that the ability to translate enthusiasm into self-motivated votes was and has for a long time, may have been vastly under-estimated.
There have been times when I and other volunteers have tried to tell staff that such and such won’t work or doesn’t seem to be working. I even did a bit of that when I was on the low rung of paid staff myself. Things like the constant contacts and circling back and verifying and other things that just feel like nagging voters. The response we get, and to be fair the people we talk to are themselves low level staff so mostly just messengers responsible for implementing the plan given to them, is always something that begins with, “We know that research shows…”