Duterte lead over Robredo shrinks to 7 points, OCTA says

By Francis Allan L. Angelo

MANILA — Vice President Sara Duterte’s lead over Naga City Mayor Leni Robredo in a hypothetical 2028 presidential matchup has narrowed to 7 percentage points, the smallest gap OCTA Research has recorded between the two in 2026, according to its third-quarter Tugon ng Masa (TNM) survey released Nov. 3.

Nationwide, 40% of adult Filipinos said they would vote for Duterte if the May 2028 elections were held today, compared with 33% for Robredo.

Twenty-six percent were undecided or chose neither candidate, and 1% refused to answer.

The gap has narrowed in each 2026 survey wave, from 11 points in the first quarter to 9 in the second and 7 in the third.

The gap is also well below the leads of roughly 20 to 25 points that Duterte held in OCTA’s 2025 commissioned surveys, which were not publicly released.

OCTA described the TNM poll as an independent, non-partisan, and non-commissioned survey released as a public service.

Interviewers conducted the fieldwork face to face from Sept. 28 to Oct. 4 with 1,200 probability respondents aged 18 and above.

The survey has a margin of error of plus or minus 3 percentage points at the 95% confidence level.

OCTA said the results reflect an early-stage hypothetical vote-intention measure and should be read as a snapshot of current public sentiment, not as a prediction of electoral outcomes.

Among the 1,170 respondents who reported being registered voters, the results were identical: 40% for Duterte, 33% for Robredo, and 26% undecided.

Duterte’s decline, not a Robredo gain

Duterte’s share eased from 46% in the first quarter to 43% in the second and 40% in the third, a cumulative 6-point decline that exceeds the margin of error.

Robredo’s share held at 35%, 34%, and 33% over the same three waves.

The undecided share rose from 19% to 22% and then 26%, a 7-point increase since the first quarter.

OCTA said the narrowing therefore reflects lower figures for Duterte rather than gains for Robredo.

The pollster said voters moving away from Duterte appear to be withholding a choice rather than shifting to Robredo.

OCTA added that the large undecided bloc gives Robredo room to grow but does not yet show her growing.

Table 4. National presidential voting preference, first to third quarter 2026

Between the second and third quarters, Duterte’s share fell 3 points, Robredo’s fell 1 point, the undecided share rose 4 points, and the gap narrowed 2 points.

OCTA said these quarter-to-quarter differences are mostly within or near the margin of error.

The pollster said the consistent direction across three waves is the key feature of the release.

Duterte’s 7-point lead remains outside the margin of error.

OCTA noted that the plus-or-minus 3 percentage point margin applies to each candidate’s share and that the margin on the gap between the two is wider, so changes in the lead should be read with additional caution.

Neither candidate commands majority support.

The pollster cautioned that its comparison with the 2025 commissioned surveys should be read as broad context rather than a like-for-like trend, because those surveys were not publicly released and may differ in design from the TNM series.

Robredo’s numbers have held even as she says she will not seek national office in 2028.

Robredo said on April 21, 2026, that she would not run in the 2028 national elections and plans to seek another term as mayor of Naga City.

She ran for president in 2022 and finished behind Ferdinand Marcos Jr., then became the first woman elected mayor of Naga City three years later.

The undecided share, at 26%, is the largest in OCTA’s 2026 tracking of the matchup.

It is more than three times the size of Duterte’s lead.

The undecided share rose in Mindanao from 18% to 25% and in Class D from 21% to 28%.

Roughly one in three voters aged 18 to 34 chose neither candidate.

Robredo leads Luzon, Duterte holds Visayas and Mindanao

Table 1. Current presidential voting preference, Sept. 28 to Oct. 4, 2026 (in percent)

Robredo leads in the National Capital Region, 39% to 32%, with 29% choosing neither candidate.

Balance Luzon remains her strongest major area, where she leads 49% to 24%.

Duterte leads in the Visayas, 46% to 27%, and posts her strongest showing in Mindanao, where she receives 71% against Robredo’s 2%.

OCTA’s data tables flag Robredo’s NCR and Balance Luzon shares, and Duterte’s Visayas and Mindanao shares, as significantly higher than the corresponding national figures at the 95% confidence level.

Table 5. Presidential voting preference by major area, first to third quarter 2026

In NCR, the two candidates were tied at 34% each in the second quarter, after Robredo held a 20-point lead in the first quarter.

Robredo’s NCR share rose 5 points from the second quarter, beyond the margin of error, though OCTA said the 300-interview subsample means the change may reflect sampling variation.

Compared with the first quarter, Robredo’s NCR share is 13 points lower, from 52% to 39%.

Duterte’s NCR share has stayed within a 32% to 34% range across all three waves.

OCTA said the NCR figures may suggest that movement there has been mainly between Robredo and the undecided column, though the data alone cannot confirm this.

In Balance Luzon, Robredo’s share is 49%, down from 51%, and Duterte’s is 24%, down from 27%, differences within or at the edge of the margin of error.

The Balance Luzon gap is 25 points, compared with 24 in the second quarter, while the undecided share rose from 21% to 26%.

Within Balance Luzon, Robredo’s share is higher in North Luzon at 53% than in South Luzon at 45%, where 32% are undecided, including 41% in rural South Luzon.

In the Visayas, Duterte’s share fell 5 points from 51%, while Robredo’s was essentially unchanged from 26%, narrowing the gap from 25 points to 19.

OCTA said the 5-point Visayas difference is beyond the margin of error but modest for a 300-interview subsample and may reflect sampling variation.

Duterte’s Visayas share is 7 points lower than her first-quarter figure of 53%.

In Mindanao, Duterte’s share has fallen from 88% in the first quarter to 76% in the second and 71% in the third.

OCTA said the 17-point decline since the first quarter is larger than the margin of error and may indicate some easing in the intensity of her Mindanao support.

Robredo’s Mindanao share fell from 6% to 2%, while the undecided share rose from 18% to 25% and 2% declined to answer.

OCTA said the Mindanao figures suggest some respondents there are holding back from a choice rather than shifting to the alternative in the matchup.

Western Visayas leans Robredo

Robredo’s highest regional preferences are in Cagayan Valley (63%), Central Luzon (54%), Ilocos Region (50%), Western Visayas (49%), and CALABARZON (47%).

Duterte’s highest are in CARAGA (80%), BARMM (79%), Davao Region (74%), Central Visayas (73%), Zamboanga Peninsula (67%), SOCCSKSARGEN (66%), and Northern Mindanao (64%).

The Visayas are not uniform, with Western Visayas leaning toward Robredo, 49% to 18%, and Central Visayas leaning heavily toward Duterte, 73% to 6%.

Large undecided shares were recorded in Bicol Region (39%), Western Visayas (32%), Northern Mindanao (31%), and CALABARZON (30%).

OCTA cautioned that regional samples range from 10 to 100 respondents, so these estimates are less precise than the national figures.

Western Visayas, or Region VI, accounted for 80 respondents.

By ethnolinguistic background, Robredo leads among Tagalog/Filipino respondents, 44% to 28%, and Ilocano respondents, 56% to 19%.

Duterte leads among Cebuano/Bisaya respondents, 70% to 7%.

Ilonggo respondents are closely divided, with 38% for Duterte and 35% for Robredo.

Class E swings toward Robredo

Table 6. Presidential voting preference by socioeconomic class, first to third quarter 2026

Class E registered the largest swing of any socioeconomic class this quarter, with Robredo’s share rising 11 points from 31% to 42% and Duterte’s falling from 51% to 41%, leaving the two statistically tied.

Duterte’s 20-point lead in Class E in the second quarter closed within a single quarter.

Robredo’s Class E share has nearly doubled from 23% in the first quarter, when Duterte led the group 56% to 23%, a 33-point margin.

Both the quarterly gain and the cumulative 19-point gain since the first quarter are well beyond the margin of error, though the Class E subsample has only 109 respondents.

In Class ABC, Robredo leads 40% to 32%, with 25% undecided.

In the second quarter, Class ABC gave Robredo 27%, Duterte 36%, and 36% were undecided; in the first quarter, Robredo had 39% and Duterte 46%.

The 13-point rise in Robredo’s Class ABC share is beyond the margin of error, but OCTA said the 125-respondent subsample means it should be monitored before conclusions are drawn.

In Class D, which accounts for roughly four in five adult Filipinos, Duterte leads 41% to 31%, down from 43% and 36% in the second quarter.

The Class D undecided share rose from 21% to 28%, widening Duterte’s lead there from 7 points to 10.

OCTA said Class D’s larger sample puts these changes beyond the margin of error and that Duterte’s advantage there continues to anchor her national lead.

Age gaps shrink, gender and locale gaps widen

Table 2. Current presidential voting preference by age group (in percent)

Duterte leads every age group under 45, while Robredo holds narrow leads in every group 45 and above.

The undecided share is highest among the youngest voters, at 35% for those aged 18 to 24 and 34% for those aged 25 to 34.

In the second quarter, Duterte’s share among 25- to 34-year-olds was 61%, and Robredo’s among those 75 and above was also 61%; those figures are now 44% and 42%.

Duterte’s share among 25- to 34-year-olds is 17 points lower than in the second quarter, while the group’s undecided share rose from 25% to 34%.

OCTA said the 75-and-above subgroup, with 34 respondents, is very small and should not be over-read.

Table 3. Current presidential voting preference by locale, gender, and education (in percent)

Robredo leads narrowly in urban areas, 38% to 35%, while Duterte leads decisively in rural areas, 47% to 25%.

Duterte leads among men, 45% to 28%, while Robredo leads narrowly among women, 38% to 35%.

Duterte leads among respondents with no formal or elementary education, 45% to 32%, and among those with high school education, 42% to 32%.

Robredo leads among those with vocational education, 46% to 22%, though that subgroup has only 50 respondents.

Among college-educated and postgraduate respondents, the two are nearly level at 35% for Duterte and 33% for Robredo, with 29% undecided.

Table 7. Presidential voting preference by demographic group, second vs. third quarter 2026

In the second quarter, Duterte led urban areas 42% to 36% and rural areas 45% to 32%, suggesting a wider urban-rural difference now.

Among women, Duterte led 41% to 37% in the second quarter, while among men she led 46% to 31%, indicating a wider gender difference this quarter.

Among respondents with no formal or elementary education, Robredo led 49% to 42% in the second quarter, while among college-educated respondents Duterte led 42% to 22%; both shifts are beyond the margin of error.

OCTA said these education figures sit in some tension with the class figures, since Robredo’s share is higher in Class E but lower among the least educated, which may mean class and education capture distinct segments or reflect sampling variation.

Impeachment trial frames the results

Fieldwork coincided with an active phase of Duterte’s Senate impeachment trial, which opened in July.

The Senate, acting as an impeachment court, opened the trial on July 6 against the backdrop of Duterte’s bitter feud with President Ferdinand Marcos Jr.

In the days before and during fieldwork, the proceedings centered on the prosecution’s evidence regarding alleged unexplained wealth and corporate interests.

On Sept. 23, the impeachment court voted to lower the vote threshold required for conviction.

On Sept. 29, during fieldwork, Davao City Mayor Sebastian Duterte testified before the court, and the court ruled that the vice president could not be compelled to admit ownership of bank accounts whose records had been submitted as evidence.

On Oct. 2, the court’s spokesperson said a verdict could be issued as early as Dec. 16.

The vice president had earlier declared her intention to seek the presidency in 2028.

Developments after fieldwork, including the prosecution’s Oct. 13 withdrawal of three of the four articles of impeachment, which left the charge relating to unexplained wealth, are not reflected in the results.

OCTA said a judgment of conviction may extend to disqualification from holding public office, which bears directly on the hypothetical matchup.

Because Robredo is not a prospective candidate by her own account, OCTA said her inclusion is best read as a measure of her standing with voters rather than a test of an actual candidacy.

The survey did not ask respondents whether the impeachment proceedings or Robredo’s statements affected their choices, and OCTA said no causal conclusion should be drawn.

What the numbers suggest

OCTA offered a working hypothesis, not a causal finding, that the lower figures for Duterte and the higher undecided share in Mindanao may be associated with sustained trial coverage.

The pollster noted that the survey did not measure exposure to trial coverage and that the pattern may reflect other factors.

In the second quarter, OCTA suggested that Robredo’s statements ruling out a 2028 run may have moved some NCR respondents toward ambivalence.

With her national share within the same 33% to 35% range and her NCR share higher this quarter, OCTA said her support may not be strongly affected by her stated intentions, though this requires confirmation.

The pollster said both hypotheses should be validated or set aside as further waves become available, particularly after the impeachment court renders its verdict.

OCTA said undecided voters may not behave as a single bloc, but their size may make them a central factor in how the matchup develops.

The survey tests a single hypothetical pairing roughly 19 months before the May 2028 elections.

OCTA said candidate filings, alliances, the trial’s outcome, and shifting public issues could substantially change these preferences.

The pollster said the findings describe a race that may be closer and less settled than earlier in the year, not one that has been decided.

How the survey was done

The survey covered 240 sample barangays distributed among the 17 regions in proportion to population size, with five probability respondents selected in each.

Sample distribution by area and locale

Sample size and barangays by region

In NCR, the 60 barangays were distributed among 17 cities and municipalities roughly in proportion to population, with every sixth household sampled from a random starting street corner.

Outside NCR, sample cities and municipalities were selected with probability proportional to household population, with a sampling interval of five in urban barangays and one in rural barangays.

Respondents in each household were chosen using a probability selection table that alternated male and female members to ensure a 50-50 gender split.

The questionnaire was prepared in Filipino with an English translation and translated into Bicol, Cebuano, Ilocano, and Ilonggo.

Field interviewers were trained in Quezon City, Cebu City, Iloilo City, Zamboanga City, and Davao City.

Field supervisors observed 12% of the interviews, and 21% of unobserved interviews were back-checked during spot checks conducted after 30%, 60%, and 90% completion.

Unavailable respondents received two valid same-day callbacks before being replaced by a substitute with the same gender, age bracket, working status, and socioeconomic class from another household.

Group supervisors reviewed at least 50% of completed questionnaires, and 74% of interviews were encoded twice to check for accuracy.

Complete results and data tables are available from OCTA Research fellows Ranjit Rye and Guido David at octaresearchphilippines@gmail.com.