A year ago there seemed to be a serious threat to Trump in the GOP. His name was DeSantis.
In March 2023, DeSantis won 35% to Trump’s 40% of GOP registered voters. (Haley was at 5%)
But DeSantis only lost ground through the year, while Trump gained.
Still, Trump didn’t move much past 50% until after Jan. 2024 once the primary process began.
As of Feb 5-15, 2024, pre-South Carolina primary, Trump holds 73% nationally to Haley’s 15% and DeSantis is no longer a candidate.
Trump rose modestly in the spring, then more in the fall, with a big jump by February. Haley was slow to rise, then bumped up in Nov. and Feb., but to only 15%.
The race that could have been.
DeSantis not only started strong but was a threat to Trump from within the Trump wing of the party. In March 2023 DeSantis got 32% among Reps *favorable* to Trump, plus 45% among those unfavorable to Trump.
Had DeSantis been able to expand that incursion into Trump land this could have been a real race. He did not. Instead DeSantis’s support fell across both Trump-favorable and Trump-unfavorable Republicans. Whether because of Trump’s effective attacks on DeSantis or the failures of DeSantis, the race that could have been was not.
Nikki Haley instead is the last contender standing against Trump, and yet she fails to reach even 20% support. She never was a contender with those who like Trump, not reaching even 6% among the Trump-favorable Republicans. But she has captured the wing of the GOP that does not like Trump, winning 65% of those Republicans.
For her, the tragedy is that nothing in the campaign succeeded in increasing the share of those unfavorable to Trump. Instead the opposite occurred, Trump’s favorability hovered around 70% until July, then rose to 80% in Sept. and stands at 84% in February.
As Haley has won an increasing share, now 2/3rds, of Republicans unfavorable to Trump, that pool has declined by half, from 30% to 16%. Even taking all of this smaller pool cannot make an alternative to Trump competitive in the primaries.
A year ago, 30% of Republicans were unfavorable to Trump, and DeSantis was eating into Trump-favorable Republicans. His effort failed. Haley has never won more than negligible support from the Trump-favorables, and even as she has consolidated the support of Trump-unfavorable Republicans, that group has been shrinking, to now less than one in five Republicans.
In March 2023 the GOP had a significant 30% who did not like Trump and a majority who supported someone else or were undecided. That moment has passed.
Debates are followed by “instant reaction” polls of debate viewers, taken overnight and posted the day after. These have some value in capturing opinion of those so interested they actually watched the debate. But this is a poor measure of the impact of the debate, which to be meaningful neeeds to be lasting. Debate performance is as much, or more, about launching the next month for the campaign as it is about the snappy line of the moment.
Thus there is more than the usual need for a slow polling look at the debate. Now, nearly a month after, do we see any meaningful shift in candidate standing with voters? And was the response steady over time, or a blip that quickly faded? On the eve of the second debate, let’s get a clear view of the impact of the first debate.
A month before vs. a month after
I look at the polls taken in the month (26 days) following the debate, Aug. 24 through Sept. 18 (the most recent poll as of Sept 21), and compare polls taken in the 28 days prior to the debate (July 25-Aug. 22). Let’s call these the “pre-” and “post-debate” polls.
First, how do the candidates stand in national polls of Republicans and independents who lean Republican registered voters? Data are from the FiveThirtyEight.com collection of national polls.
Figure 1 shows the comparison for the eight debate participants. Only Nikki Haley has a meaningful increase in support, up 2.4 percentage points following the debate. No one else changes by even a single percentage point.
The trends before and after the debate tell the same story. Haley looks to have continued to rise after the debate, while Ramaswamy has slightly declined, as has DeSantis. All these changes are small, except for Haley.
The means and medians and number of polls are shown in the table below. In contrast to Haley’s increase, Christie and Trump rise by less than a tenth of a percentage point while the rest decline, all by less than a percentage point. Pence, Scott and DeSantis have the largest declines, though even DeSantis is down by only 0.6 percentage points.
And the candidate who wasn’t there didn’t get a boost or a drop: Trump’s support changed by 0.03 percentage points.
If the debate was thought to potentially scramble the standings, it did so only for Haley, though her support remains in single digits. Pence and Ramaswamy took much of the debate time, and a good bit of post-debate commentary. Neither has seen a lasting payoff.
Quick polls get attention, but slow polls are more illuminating.
In my previous post I argued that the indictments of former president Donald Trump did not in fact boost his standing with voters, despite the often repeated claim that they did.
Two folks I respect pushed back on the NY indictment, though agreeing the others had no effect. Here I respond, and agree in part. In any case, I think this shows a positive discussion is possible over such matters! My bottom line is Trump may have gained support in the GOP primary following the NY indictment. But I don’t find evidence of any gain in overall favorability with all registered voters. My new analysis (thanks to Philip and Sam) finds some evidence that the Florida indictment lowered his support and favorability, something I did not mention in my original post..
Since my original post, Philip Bump at The Washington Post published a nice story looking at my claim, and concluding there was in fact a boost from the first indictment in New York but agreeing there were no boosts from the subsequent indictments. See his piece here. (Link should not be paywalled.)
Sam Wang posted a similar point on Bluesky (sorry, don’t know how to copy a link to that)
Let’s start with the strong case, a rise in Trump GOP primary support following the NY indictment. Here is Trump’s percentage i.n GOP primary polls before and after each indictment. (I’ve updated the most recent polls since my original post, which affects only the post-GA data.)
We all agree the post-New York support is higher than pre-indictment. The only disagreement is that there was a trend of rising Trump primary support before the indictment and that rise continued for over a month after the indictment. My original post claimed the post-NY increase was primary due to the pre-existing trend, rather than an indictment effect. Bump and Wang disagree and see an indictment effect. Here is the chart with the full trend.
The red line shows the overall trend across all polls, the dots are individual polls. The high-frequency polls from YouGov (85 in all) make up the dense set of dots mostly above the red trend line. Those dense dots DO show a bump up immediately after the NY indictment. In my original post I discounted that, thinking the red line was increasing before and after and I gave more weight to that. But I discounted the dense set of polls that do rise. Wang raises the point but Bump dives into this and does more substantial analysis that I now think supports a NY indictment effect.
Here are my finding on this, in response to their points, and finding myself now agreeing there was an independent effect of the NY indictment even considering the trend before and after.
I fit several models, but here are the two that make me conclude NY had an effect. The dependent variable is Trump percent in primary polls. You do NOT want to use his margin over DeSantis because the latter has been declining steadily, so the margin confounds Trump support with DeSantis’ weakening. As you can see in the chart, Trump has been pretty flat since mid-May.
The models I fit are a polynomial in time, which allows for the curve of the support trend, which rises, then flattens. Obviously not a constant linear trend. I fit one quadratic and one cubic fit (the latter probably overfitting the trend but our focus in on the coefficients for the indictments.)
The first model is the quadratic trend.
The NY coefficient is a 4.12 percentage point increase in Trump support after the indictment, and is a statistically significant effect. So this agrees with Bump and Wang, and shows I dismissed this effect too quickly.
The model also suggests that the Florida indictment may have lowered Trump’s support by 2.8 points, a marginally statistically significant result. There is no evidence that DC or GA indictments changed his support. Both have negative but non-significant estimates.
An issue is whether the quadratic trend is sufficient to capture the trend over time, independent of the indictments. As a check I reestimate the model with a cubic in time. That model is shown next.
The NY effect here is a 4.8 point increase, which remains statistically significant. The Florida effect remains negative but falls short of statistical significance. So either way, the NY indictment seems to have boosted Trump with GOP primary voters over and above the trend leading up to the indictment.
On the other hand, Trump’s favorability ratings with all registered voters don’t seem to have gone up with NY, but may have declined slightly with the FL indictment.
Here the trend is nearly flat since January, so I estimate one model that is linear in time, and one that is quadratic as a robustness check. The linear time model is
The time trend is virtually flat, with no evidence of a NY indictment effect. However, the Florida indictment seems to have lowered favorability by 3.6 points.
Using a quadratic in time is similar:
The Florida estimate is a 3.8 point decrease in favorability, and marginally significant, with no evidence for a more complicated time trend. As a final check I ran a cubic time trend with the FL coefficient of -3.5, but p=.056 so not as convincing an effect.
Bottom line is the NY indictment didn’t show any evidence of boosting Trump’s favorability among all registered voters, but it does seem to have improved his standing in the GOP primary by 4.1 to 4.8 points depending on the model.
My thanks to Philip and Sam for pushing this point and adding to the analysis.
To cut to the chase, the indictments of Donald Trump have not boosted him in the polls, either for favorability or for his support in the GOP primary. This claim keeps being repeated as if the data support it. It does not.
Net favorability nationally
Trump’s net favorability (percent favorable minus percent unfavorable) inched up 1.5-2 points following the first indictment in New York on March 30, 2023, from a median of -14.0 to -12.5, and mean of -14.2 to -12.2. This is the only period of a (slight) improvement compared to the pre-indictment period (Jan. 1 to Mar. 29), Following the Florida indictment the median fell to -18, then -19 after the DC indictment and rose -17.5 after Georgia. (Table of full results below.)
The net favorability is for the national population, so perhaps Trump gained substantially among Republican voters, but not the full population. His support in the GOP primary vote gives us that test.
Trump GOP Primary support
Compared to pre-indictment, Trump did have higher support among Republicans in the primary vote following the New York indictment, a median of 49% prior to the NY indictment and 57% after NY but before Florida. There was no further change after Florida (still 57%), or after DC (still 57%.) After the Georgia indictment the median is 58%. If there was in indictment effect boost at all (see below for why you should doubt that) it was over by the Florida documents case indictment.
The remaining doubt about indictment effects is provided by the trend chart below. Trump’s support for the nomination had been rising steadily since Jan. 1 through the spring. It rose at the same rate following the New York indictment as it had been rising prior to the indictment. The trend levels off in early May, a month after the NY indictment and about a month before the Florida indictment. Since mid-may there has been very little trend in Trump’s primary vote. (Each point in the chart is a national poll and the red line is a local regression trend estimate.) Trump’s margin over DeSantis has continued to climb but that is due entirely to DeSantis’ collapse in the polls, not to any gains by Trump since May. Trump’s current 58% support is more than enough to win the nomination. But it hasn’t been increasing for four months.
There are reports of surges in donations following the indictments. But if so, that hasn’t been reflected in the polling for either favorability or for GOP primary support.
The table below shows the median, mean and number of national polls used in the charts above. Note there are only 4 favorability polls completed between the DC and GA indictments, though there were 16 primary polls in the same interval.
“Indictments help Trump” is folklore that needs to be corrected. There is no polling evidence the indictments help, or hurt, Trump. Any effect of criminal trials remains to be seen.
Using Republicans and independents who lean Republicans, Trump gets 31% and DeSantis gets 30%, a surprisingly close race. Trump leads by considerably more in other states and in national polls. What’s going on in my June Wisconsin @MULawPoll?
One might ask if independents who lean Republican are distorting this. But no. Among “pure” Republicans, Trump gets 35% and DeSantis gets 34%. Among independents who lean Republican, there are more other and undecided choices but the margin is not much changed: Trump gets 23% and DeSantis 25%.
So the close first choice race is not because of including leaners: it is a 1 point Trump margin without leaners too. (Scroll right for full table)
Group
Chris Christie
Ron DeSantis
Larry Elder
Nikki Haley
Asa Hutchinson
Mike Pence
Vivek Ramaswamy
Tim Scott
Donald Trump
Haven’t decided
All Rep+Leaners
1
30
0
3
0
6
3
5
31
21
Rep
1
34
0
1
0
5
2
4
35
18
Lean Rep
1
25
1
6
1
7
4
6
23
27
Head-to-head second choice
DeSantis picks up a big margin in the head-to-head second choice question, DeSantis 57% to 41% for Trump. Who moves from the first choice?
Those who pick another named Republican candidate (other than DeSantis or Trump) break 74-25 for DeSantis on the 2nd choice. And the undecided on first choice break 65-28 for DeSantis. See Table 2.
Trump is a lot of Republican’s first choice, but barely a quarter switch to him as their 2nd choice.
1st choice
Donald Trump
Ron DeSantis
Don’t know
DeSantis
1
98
1
Trump
98
2
0
Other candidates
25
74
1
Undecided
28
65
7
Is this because Republicans have turned against Trump?
Despite getting just 31% of first choice votes, Wisconsin Republicans still like Trump. Of all Reps and leaners, Trump favorability is 68% and unfavorable is 30% with a tiny 2% who say they haven’t heard enough about him.
Favorability is a shade higher with pure Reps, 72%, and a good bit lower, though still net positive at 60%, among leaners. Those ratings are shown in Table 3 (a). In October 2022, Trump was 78% favorable among Reps and 62% favorable among leaners, so a little decline with both, but not massive change.
DeSantis favorabiity is shown in Table 3 (b). He is 1 point behind Trump overall, 2 points above Trump on favorable with pure Republicans, and 7 points below Trump among leaners, but with much more haven’t heard enough and less unfavorable.
So there may be a small recent decline in Trump’s favorable ratings with Reps and with leaners, it isn’t a lot compared to where he was in October 2022.
Table 3: Favorability by Rep and Lean Rep Party ID
Group
Favorable
Unfavorable
Haven’t heard enough
All Rep+Leaners
68
30
2
Rep
72
25
3
Lean Rep
60
39
0
Group
Favorable
Unfavorable
Haven’t heard enough
All Rep+Leaners
67
14
20
Rep
74
10
16
Lean Rep
53
21
26
Well, did they just not vote for him in 2020?
Maybe these are somehow less Trump supportive Reps than they should be.
Asked to recall their vote in 2020, 95% of Reps say they voted for Trump, as do 79% of the leaners. In our late October 2020 poll, 90% of Reps and 79% of leaners said they were voting for Trump. This is not a peculiarly anti-Trump sample of Republicans.
Table 4: 2020 vote recall by party id, and Oct. 2020 pre-election vote
Group
Donald Trump
Joe Biden
Someone else
All Rep+Leaners
90
5
5
Rep
95
4
1
Lean Rep
79
7
13
Group
Donald Trump
Joe Biden
Jo Jorgensen
All Rep+Leaners
86
7
3
Rep
90
6
1
Lean Rep
79
8
6
Maybe they just aren’t as into you as they once were
There is a clear long term change in GOP views of Trump in Wisconsin since 2020. While he retains a core of vigorous supporters, Republicans and leaners do not have as favorable a view of Trump as they did in 2020. Table 5 shows this trend since January of 2020 until June of 2023.
Poll dates
Favorable
Unfavorable
Haven’t heard enough
Don’t know
1/8-12/20
83
16
0
1
2/19-23/20
87
10
1
1
3/24-29/20
84
12
0
3
5/3-7/20
88
9
1
1
6/14-18/20
81
13
3
3
8/4-9/20
83
14
1
2
8/30-9/3/20
80
15
3
1
9/30-10/4/20
85
13
2
1
10/21-25/20
88
11
1
0
8/3-8/21
75
18
3
3
10/26-31/21
72
23
2
4
2/22-27/22
70
21
3
5
4/19-24/22
68
24
3
4
6/14-20/22
75
19
3
3
8/10-15/22
71
21
2
5
9/6-11/22
72
24
0
4
10/3-9/22
70
22
2
6
10/24-11/1/22
72
17
2
6
6/8-13/23
68
30
2
1
For completeness, Table 6 shows the trends for pure Republicans and for leaners.
Table 6: Trump favorability, Jan. 2020-June 2023, among Replicans and leaners separately
Poll dates
Favorable
Unfavorable
Haven’t heard enough
Don’t know
1/8-12/20
87
12
0
0
2/19-23/20
91
7
1
1
3/24-29/20
88
10
1
2
5/3-7/20
92
5
1
1
6/14-18/20
84
12
1
4
8/4-9/20
85
13
1
1
8/30-9/3/20
84
11
3
1
9/30-10/4/20
90
9
1
0
10/21-25/20
89
10
1
0
8/3-8/21
83
12
3
2
10/26-31/21
77
18
2
2
2/22-27/22
80
11
2
5
4/19-24/22
73
19
2
4
6/14-20/22
81
17
1
0
8/10-15/22
77
15
2
5
9/6-11/22
79
16
0
4
10/3-9/22
78
15
2
5
10/24-11/1/22
78
12
3
5
6/8-13/23
72
25
3
0
Poll dates
Favorable
Unfavorable
Haven’t heard enough
Don’t know
1/8-12/20
76
22
1
1
2/19-23/20
78
18
1
3
3/24-29/20
78
17
0
5
5/3-7/20
81
16
1
1
6/14-18/20
76
17
6
1
8/4-9/20
78
16
2
3
8/30-9/3/20
73
25
2
0
9/30-10/4/20
74
19
5
2
10/21-25/20
84
14
1
0
8/3-8/21
60
30
4
5
10/26-31/21
60
31
2
6
2/22-27/22
53
37
5
4
4/19-24/22
58
35
4
2
6/14-20/22
62
23
7
8
8/10-15/22
61
33
1
6
9/6-11/22
57
39
0
4
10/3-9/22
54
35
2
7
10/24-11/1/22
62
26
1
8
6/8-13/23
60
39
0
1
Vote for Biden? Oh, goodness no.
Our GOP respondents aren’t as into Trump as in the past, but crossing over to vote for Biden remains a bridge too far. That goes for both Trump and DeSantis when matched against Biden, though DeSantis does a bit better with leaners, in Table 7.
Table 7: 2024 vote by party strength
Party stength
Donald Trump
Joe Biden
Haven’t decided
Don’t know
Refused
Rep
93
5
1
0
0
Lean Rep
78
12
6
1
3
Party stength
Ron DeSantis
Joe Biden
Haven’t decided
Don’t know
Refused
Rep
94
2
2
1
0
Lean Rep
87
12
2
0
0
Conclusions
Wisconsin Republicans are less attached to Trump than in 2020, are considering alternatives or are undecided, but not quick to embrace Trump if he isn’t already their 1st choice. But for the general election, they remain strongly Republican regardless of the candidates, and quite unwilling to vote for Biden.
Which GOP candidates pose a threat to Donald Trump among Republican voters?
In May Trump had 72% favorable and 26% unfavorable (2% didn’t have an opinion of him) among registered Republicans and independents who lean Republican in the Marquette Law School national poll. That high favorability rating is Trump’s great strength in the party. His favorability has been consistently high, though it declined a bit in the fall of 2022 then turned back up in May.
The other GOP candidates are less well known than Trump, even Pence and DeSantis lack the universal recognition of Trump. But low name recognition at this point can be an opportunity for those less well known to introduce themselves to many voters for the first time. Only Christie, Hutchinson and Sununu (who took himself out after the poll was completed) have net negative ratings. Note they are also the most openly critical of Trump. The others are not known, rather than disliked. (The chart is sorted by % able to rate the candidate.)
DeSantis is the clear 2nd place for favorability among all Republicans, and has low unfavorable ratings as well. Pence has relatively high favorables but also unfavorables that are almost as high. With the launch of his presidential campaign, and apparent willingness to criticize Trump, at least in his kick-off speech, his break with Trump is now clearer.
But how do these Republicans who like Trump feel about the alternative candidates?
The next two charts shows favorability to each candidate among those Republicans who are favorable to Trump and among those unfavorable to Trump.
DeSantis jumps out in the “favorable to Trump” chart with especially strong ratings compared to the other candidates, while views of Pence are notable for the very close division in the party over him.
For Christie, those favorable to Trump are 18% favorable to Christie, 43% unfavorable (and 39% don’t have an opinion.) But those Republicans unfavorable to Trump don’t like Christie either! 17% fav, 38% unfav. So Christie doesn’t have am obvious strength among anti-Trump Republicans. Or, basically, anywhere in the GOP.
Compare DeSantis who is 67% fav and 12% unfav among those favorable to Trump, and 37-26 among those unfavorable to Trump. DeSantis’s strength is WITHIN the pro-Trump GOP, not really so much in the anti-Trump quarter of the party. That means DeSantis is a danger to Trump from “inside the house.” He would benefit if Trump voters started looking for an alternative. The trick of course is how does DeSantis make this happen without turning the pro-Trump folks against himself.
DeSantis’ appeal to pro-Trump Republicans looked especially strong earlier. He got increasingly favorable ratings among those favorable to Trump until recently. In our May data his favorability dropped sharply among those favorable to Trump as Trump’s criticism of DeSantis mounted. Yet DeSantis has never been especially popular with the anti-Trump wing of the party, so he can’t turn to them for a boost, at least not yet. Does his favorability continue down, especially if they go after one another? Can DeSantis actually hurt Trump’s standing with GOP voters?
This is where Christie could potentially help DeSantis or other single-digit candidates. He can criticize Trump without worrying about hurting himself because he is already so unpopular with all kinds of GOP voters. He could do the dirty work for DeSantis & Co. But would Christie’s criticism be effective when so many in the party have turned against him?
And consider Mike Pence. The second best known candidate divides the party sharply, with only a slightly net-favorable rating. Interestingly those favorable to Trump are also pretty favorable to Pence, despite Trump’s view of his former Vice-President. Should Pence take ths offensive against Trump he might carry more weight than the disliked Christie. But as with DeSantis, how can he turn GOP voters against Trump without creating self-inflicted wounds?
Tables with values plotted in the charts are shown below.
Attention has focused on Ron DeSantis losing ground to Donald Trump in GOP primary polling since January. What has been largely overlooked is that DeSantis has also fallen in general election matchups against Biden. This is especially problematic for a claim that Trump is less electable.
Let’s work through our standard “four-charts” to see what has happened in the Biden v DeSantis horserace. (For an explanation of the value of these four different pictures of the race see my earlier post here.)
The trend chart, showing support for each candidate, shows a clear downturn for DeSantis since January. In late fall he enjoyed a surge and Biden declined a bit in their matchups. But since early January, DeSantis’ trend has sloped down, with one brief upturn. Over this same period, Biden has maintained stable support (as he has vs Trump too).
There is a lot of noise in the polls, as the mix of red and blue points in the chart shows. In January the candidates were tied but DeSantis’ decline has given Biden a consistent, though small, edge recently.
The clear change in direction of the trend is clearer in the margin trends in the second chart. Biden’s lead over DeSantis declined through the fall, but bounced in early January, changing direction quite clearly. The size of the margin is still small, under 3 percentage points, so no one should read this as an insurmountable advantage. But the shift of momentum is striking, and mirrors DeSantis’ more widely remarked upon decline vs Trump in the nomination contest.
Charts 3 and 4 are the longer-term view of how the distribution of poll results has shifted. Here the focus is not on the “point estimate” of the trend line, but rather the wide range of polling over two 2-month periods. Polls vary a lot, and it is hard to appreciate how that wide variation none-the-less reveals change in the distribution of candidate support. Shifts in the distribution of support over a longer period is more convincing evidence of meaningful change than a brief shift in the trend line, let alone a single poll that may be an outlier.
Chart 3 shows the distribution of DeSantis and Biden support in January-February compared to the distribution in March through early May. In Jan.-Feb. the DeSantis and Biden distributions largely overlap, though there is more spread to DeSantis support with a fair bit of poor performance to the left and some strong performance to the right, compared to the somewhat less variable Biden support. But in the two months since March, the DeSantis distribution has clearly shifted to the left and Biden a little to the right, with most of the overlap of distributions now to DeSantis’ disadvantage. There is also still more spread to DeSantis’ support than for the somewhat more consistent Biden.
This shift is also clear with the distributions for the margin between the two candidates in Chart 4. In Jan-Feb the distribution peaks very close to a tie, a margin of zero. Biden led in 19 polls and DeSantis in 13, with 4 ties. Since March the distribution has shifted to the right, with a peak at about a 2-point Biden advantage and considerably more polls showing Biden ahead, 28, than DeSantis, 11, with 3 ties.
All four of these charts show the decline in DeSantis support vs Biden in the general election. That doesn’t mean the outcome is set. Races shift over time, sometimes (at least in the past) rather sharply. In the modern period we have seen closer races and less variability as partisans have found it close to impossible to bring themselves to cross over for the other party. So small margins are to be expected and modest changes signal momentum shifts.
Ron DeSantis still has to defeat Donald Trump before he can face Joe Biden in November 2024. His recent weakening versus Biden undercuts his claim to be more electable than Trump. If he announces his candidacy in the coming weeks, as expected, he’ll have his chance to change the momentum once more. If he succeeds in that the charts will tell that story too.
I’ll be keeping track of the 2024 general election here and in subsequent posts under the “Pres2024” menu above. I’ll update regularly but not daily, let alone instantly for each new poll. Think of this as “slow polls” with some perspective, not “hot polls” for the latest shocking result.
I’ll be presenting three basic ways of looking at the horse-race trends. The first two are utterly familiar, but the third is not, and is my effort to give a better perspective on polling uncertainty than the usual. Give it a chance.
Biden vs Trump
The basic trend chart for each candidate shows both all the polls and the trend lines (estimated via local regression, as are all my trends). I scale these through Election Day to emphasize both where we are today and how much of the race is left to go.
A fundamental point is that I show these trends not to predict the outcome but to show how we get there. So much commentary either treats today’s polls as forecasts or dismisses them as meaningless. Neither is true. It is the ebb and flow of candidate support that defines the shape of the race but the outcome is uncertain even near the end. Don’t either overvalue or undervalue current polls. Watch how support is changing.
The candidate trends
The first “standard” figure shows the polls and trends for each candidate. This update shows Trump fell after the November elections, but has improved pretty steadily since January, closing the gap while Biden’s support has been flat. The Trump trend is now just slightly ahead of Biden. It is Trump’s gains over the last four months that are notable, not that Biden has fallen over that period, which he hasn’t. In fact, Biden’s support rose in the run-up to the 2022 elections and has remained at that (modestly) improved level compared to the first part of 2022 when Trump has a slight advantage.
The margin trend
The second standard trend is the margin between the candidates. I show Democrat minus Republican margin here. This is just the gap between the blue line and red line above, but focusing on the margin has a couple of advantages. First, “who’s ahead” matters, and this shows that directly. Movement in the margin may also be more apparent with just a single trend line than having to perceive the difference of two lines. Second, votes for 3rd party candidates or undecided responses can lower the support for both candidates but not affect the margin between them. I think that an advantage of the margin plot. (Some good folks disagree on this, but I say why choose between them? I show you both. Focus on what pleases you.)
The Biden minus Trump margin trend shows the tightening of the race since January. It also makes clear that Trump had a small advantage in 2022 until close to the election, after which Biden has generally had the edge, though the margin has tightened since January. Note here you can’t see which candidate changed, an advantage of the separate candidate trends above. But here you see how close the race has been and how the advantage has shifted regardless of who moved.
The distribution of support
One of the hardest things to get across is the uncertainty of polling results. Each new polls has the potential to upset the narrative, all the more so if it is an outlier. Stories and Tweets chase this new shiny object either saying it changes everything or it is a ridiculously flawed poll that means nothing. Alas, I’ve tried since Mark Blumenthal and I founded Pollster.com in 2006 to get people to appreciate the variability in polls and to stop freaking out over a single result. I have utterly failed in that mission. But not to let that stop me, here is another way to try for perspective on uncertainty.
In the trend plots above, I plot each individual poll to show how much variation there is around the trend line. This spread should draw the eye and invite the comparison of small trend changes and wide spread of individual polls around that trend. This was the fundamental contribution of Pollster.com’s graphics, compared to earlier simple lists of polls and an average.
But here is a third way to presenting how the polls vary and how the campaign changes. Let’s simply focus on the distribution of polls over a period of time and compare to another period.
The trend estimates give a line, the estimate at every point in time. Instead, let’s be a lot more modest and say “here is a group of 2 months worth of polls.” Instead of focusing on trend, let’s just pay attention to how those polls varied. And then let’s compare that with the next 2 months of poll. (I use 2 months to get enough polls in each group for the distribution to be reasonably clear.)
The third chart below shows these distributions for Biden and for Trump support for January & February 2023 and for March through the latest poll (May 7th). You can see how the Biden distribution was to the right of Trumps, even as some individual polls have Trump ahead in Jan-Feb. But in March-May 7, the two distributions have substantially converged, without a clear advantage for either candidate. The Biden distribution has shifted slightly to the left and Trump a little more to the right, producing the current overlap. (The little lines on the x axis are individuals polls for each candidate. Some overlap, obscuring the number of polls, which is shown in the subtitle of the char.)
You might also notice both distributions are a bit narrower in March-May, showing a little more agreement across polls.
The primary perspective is that over each of these 2 month groupings, each poll could be thought of as just a random draw from this distribution, so if you see polls varying by as much as shown in the chart, that just means “it’s Tuesday.” You can get more interested in how these distributions change over time though, because that IS showing you that the race is shifting. By grouping into 2 month periods, I want to de-emphasize daily noise that appears in the first two trend charts and encourage you to take a deep breath, be patient, but see more of the “real” evolution of the race that comes with a wider time interval and more appreciation for the variability in the polls.
The fourth and final chart shows the distribution of the margin between Biden and Trump, as the second chart shows the trend in margin. Here you again see the spread of poll margins in each 2 month period, and notice the shift to a closer race in March-May. I also add the number of polls with each candidate leading and the number of ties. The balance of number of leads also helps focus the mind that “your candidate” isn’t leading every poll (or trailing either.)
I don’t expect any of this to prevent the next freak-out over a surprising poll. I do hope those looking for a bit more perspective, even at the cost of momentary excitement, will find these charts helpful for appreciating the dynamics of the 2024 presidential campaign.
