Share of all respondents · unweighted
Share of all respondents · weighted
Sitiveni Rabuka · all respondents
Completed interviews across Fiji
How Fiji intends to vote
National voting intention, adjusted for Fiji’s geographic and ethnic population profile. Select a party to see its electorate in detail.
National party support
Share of decided voters · denominator excludes undecided, non-voters and refusals.
All-respondent support
Includes undecided, non-voters and declined responses; denominator is all 1,673 respondents.
A fragmented electoral landscape.
A competitive multi-party contest.
Voting intention is distributed across several political parties, with no party commanding the support of a majority of decided voters. The results point to an increasingly competitive electoral landscape, while a significant proportion of respondents remain undecided.
Illustrative Parliament
Illustrative allocation of 55 seats using D’Hondt and a 5% party threshold.
| PAP | 16 seats |
| Unity Fiji | 8 seats |
| SODELPA | 7 seats |
| NFP | 7 seats |
| People First | 6 seats |
| FLP | 4 seats |
| NextGen | 4 seats |
| One Nation | 3 seats |
Who should be Prime Minister?
Who would respondents most prefer as Prime Minister after the next general election? Results include all respondents, including those without a preferred candidate.
Named candidates
Share of all survey respondents · population-adjusted
Not everyone names a leader
Some respondents are undecided or have no preferred candidate.
Leadership within parties
Party profiles show preferred PM among that party’s actual supporters.
Explore party leadership profiles ↗Preferred PM by voter group
Select a demographic dimension. Each chart is a weighted share of all respondents in that subgroup. Small bases are suppressed.
Central
n=697 respondentsWestern
n=598 respondentsNorthern
n=276 respondentsEastern
n=102 respondentsiTaukei
n=1035 respondentsIndo-Fijian
n=550 respondentsOther
n=88 respondentsFemale
n=848 respondentsMale
n=810 respondents18-24
n=400 respondents25-34
n=442 respondents35-44
n=367 respondents45-54
n=243 respondents55-64
n=150 respondents65 or older
n=49 respondentsSmall base. Interpret cautiously.
Urban
n=1210 respondentsRural
n=462 respondentsThe issues shaping the vote
What issues will influence voter decisions? Respondents identified priorities ranging from employment and leadership to public safety, education, healthcare and household costs.
Fiji's
priorities.
Quantitative analysis of responses to the open-ended question about the issues most likely to determine voting decisions.
Issues mentioned by respondents
Percentage of all respondents who raised each theme
The survey asked respondents for the single most important issue influencing their vote. Some answers mentioned several subjects. Each identifiable theme is counted once per respondent, so percentages may total more than 100%. Figures use the full weighted sample of 1,673 respondents, including those who did not identify an issue. Themes were grouped using a standardised topical coding frame.
What matters across different voter groups?
Compare the share of respondents mentioning an issue across divisions, ethnic groups, genders, ages and place of residence.
Central
n=697Western
n=598Northern
n=276Eastern
n=102iTaukei
n=1,035Indo-Fijian
n=550Other
n=88Female
n=848Male
n=81018-24
n=40025-34
n=44235-44
n=36745-54
n=24355-64
n=15065 or older
n=49Interpret cautiously: small subgroup.
Urban
n=1,210Rural
n=462Each percentage uses the weighted total of respondents in the selected group. Groups with fewer than 30 interviews are not displayed.
How different groups vote
Explore how party preferences differ by division, ethnicity, gender, age and residence. Select any party to view its detailed profile.
Central
n=572 decidedWestern
n=495 decidedNorthern
n=197 decidedEastern
n=96 decidediTaukei
n=866 decidedIndo-Fijian
n=427 decidedOther
n=67 decidedFemale
n=666 decidedMale
n=683 decided18-24
n=341 decided25-34
n=371 decided35-44
n=297 decided45-54
n=192 decided55-64
n=107 decided65 or older
n=33 decidedSmall base. Interpret cautiously.
Urban
n=998 decidedRural
n=361 decidedWho supports whom?
Open a party to see where it performs strongly, who its supporters are, their preferred PM and how firm their voting decisions are.
Who participated in the survey?
Explore the demographic profile of the people interviewed for the CDD National Opinion Tracker. This page describes the survey sample itself, not political party support.
A national picture,
built from 1,673 voices.
The survey reached eligible adult respondents across Fiji’s four divisions, documenting different ages, ethnicities, genders and places of residence. The figures below show how respondents are distributed across these groups.
Division
Where participants ordinarily reside, compared with estimated Fiji adult-population shares.
Ethnicity
Self-identified ethnicity, compared with estimated Fiji adult-population shares.
Gender
Share of all completed interviews. Counts and percentages are unweighted.
Age group
Share of all completed interviews. Counts and percentages are unweighted.
Place of residence
Share of all completed interviews. Counts and percentages are unweighted.
Interview language
Language in which the interview was conducted.
How the survey was conducted
CDD's National Opinion Tracker is an independent, non-partisan research programme designed to measure public opinion, electoral preferences and national priorities in Fiji.
Survey population and coverage
The target population is Fijian citizens aged 18 or older, ordinarily resident in Fiji and eligible to vote in the next general election. Wave 1 includes 1,673 completed interviews covering the Central, Western, Northern and Eastern divisions. The final survey records confirm consent and eligibility screening for all included respondents. See respondent demographics.
Stratification and respondent selection
The study uses a division-by-ethnicity stratified survey design: four administrative divisions and three broad ethnicity categories (iTaukei, Indo-Fijian and Other). Gender was monitored within each division. The fieldwork protocol specifies objective respondent selection within the sampling frame and supplementary recruitment to address geographic or demographic gaps. The completed dataset records all interviews as telephone interviews.
Questionnaire
The standardised questionnaire asks voting intention, whether a party choice is final or could change, preferred Prime Minister, and the most important issue affecting voting choice. Political-party options were not to be read out unless requested. Preferred PM and national-priority questions were open-ended, with no suggested names or issues. Gender, age, ethnicity, division, residence and interview language were also recorded.
Consent and impartiality
Participation was voluntary and required consent. Respondents were told the purpose of the survey, its independence, their right to decline questions, and the confidential handling of answers. The fieldwork guidance prohibits interviewers from prompting political choices or choosing respondents on the basis of their political preferences.
Preparation of the results
Voting intention is reported both for all respondents and for the subset naming a party or independent candidate. Preferred-PM names are standardised where they clearly refer to the same person. Open-ended issue answers are classified by recurring topics. Non-responses are retained in the relevant all-respondent denominator; missing answers are not filled in with assumptions.
Percentages and uncertainty
The decided-voter base is 1,360; it excludes 278 undecided responses, 25 who said they would not vote, eight who preferred not to answer, and two who preferred not to say. Other national results generally use all 1,673 interviews. Small demographic groups are not charted when their unweighted base is below 30. Sampling and non-sampling uncertainty should be considered when interpreting close results.
How the weighting works
Weighted results reduce differences between the achieved sample and Fiji's demographic distribution, using geographic and ethnic population targets.
strata
Estimate the adult population
We use Fiji Bureau of Statistics population-by-division and ethnicity figures from the 2017 Population and Housing Census. To estimate the adult population in each division, we add the population aged 20 and over to two-fifths of the 15–19 age bracket, representing ages 18 and 19.
Set the 12 target shares
The estimated adult share of each division is applied to that division's ethnic population totals. Each division-ethnicity group is then expressed as a share of the estimated national adult population, approximately 580,646 persons.
Adjust each respondent
The population share of a stratum is divided by its share of the completed sample to obtain the respondent's adjustment factor. Respondents in underrepresented groups receive a factor above 1; those in overrepresented groups receive a factor below 1.
weight = (stratum share of estimated adult population) ÷ (stratum share of completed survey sample)100 × (sum of weights for the party's decided voters) ÷ (sum of weights for all decided voters)The adjustment is applied to the full sample before calculating subgroup or decided-voter percentages. For an all-respondent issue percentage, the sum of weights for respondents mentioning that topic is divided by the sum of weights for all respondents.
View the 12 weighting strata and their adjustment factors +
| Division | Ethnicity | Adult population target | Interviewed | Sample share | Weight |
|---|---|---|---|---|---|
| Central | iTaukei | 28.476% | 441 | 26.36% | 1.080 |
| Central | Indo-Fijian | 11.698% | 213 | 12.73% | 0.919 |
| Central | Other | 2.652% | 43 | 2.57% | 1.032 |
| Western | iTaukei | 21.418% | 338 | 20.20% | 1.060 |
| Western | Indo-Fijian | 16.520% | 240 | 14.35% | 1.152 |
| Western | Other | 1.076% | 20 | 1.20% | 0.900 |
| Northern | iTaukei | 9.096% | 162 | 9.68% | 0.939 |
| Northern | Indo-Fijian | 4.628% | 95 | 5.68% | 0.815 |
| Northern | Other | 0.593% | 19 | 1.14% | 0.522 |
| Eastern | iTaukei | 3.585% | 94 | 5.62% | 0.638 |
| Eastern | Indo-Fijian | 0.057% | 2 | 0.12% | 0.479 |
| Eastern | Other | 0.201% | 6 | 0.36% | 0.560 |
The 12 population targets total 100%. Figures may not add precisely when rounded. Interview counts are unweighted. The factor is applied individually to respondents in the associated stratum.
The benchmark is an estimate of the resident population aged 18+ derived from the 2017 census, not a direct count of registered electors in 2026. Single-year ages 18 and 19 were approximated as 40% of the 15–19 bracket. The adult share within each division was applied across its ethnic groups because age-by-ethnicity figures for each division were not supplied.
These published estimates are adjusted by division and ethnicity. They are not separately calibrated by gender, age or urban/rural residence. Weighting can reduce observed demographic imbalances, but cannot fully address every source of sample-selection, coverage or response bias.
How we report voting intention
Decided-voter shares exclude those without a selected party or independent candidate. All-respondent shares retain undecided and non-voting choices in the denominator. The parliamentary seat illustration applies a 5% threshold and D'Hondt to weighted decided-voter preferences, but is not an election forecast.
Interpreting national priorities
Respondents were asked which single issue would most affect their vote. Some answers mentioned several issues. Results on the issues page measure the weighted proportion mentioning each theme at least once. Themes can overlap; their percentages therefore need not sum to 100%.
Precision
For comparison, a simple random sample of 1,673 would have a theoretical maximum 95% sampling error of about ±2.4 percentage points. This is not a validated design-adjusted margin of error for the survey and does not capture non-response, coverage or other non-sampling error.
Research standards
The CDD methodology manual is informed by Fiji's Electoral Act 2014, the Electoral Commission's opinion-polling guidelines, and ESOMAR/WAPOR research and publication principles. Survey results reflect attitudes recorded at the time of interview, not predictions of future voting behaviour.