A public opinion survey conducted by the Sustainable Development Policy Institute (SDPI) and the Herald magazine shows that the Pakistan Tehreek-e-Insaf (PTI) enjoys a slim lead at the national level over the Pakistan Muslim League-Nawaz (PMLN). This lead of 4 percentage points is outside the survey’s national margin of error – + 1.3 percentage points – this, however, does not take into account the fact that 13 per cent of the respondents remain undecided. PMLN, in turn, is leading the Pakistan Peoples Party (PPP) by 5 percentage points.

In the election’s main battleground, Punjab, the survey shows a competitive race: PMLN has a province-wide lead of 7 percentage points over PTI which is outside the survey’s provincial margin of error of + 1.9 percentage points. What, however, is of concern for PMLN, and gives hope to PTI, is that outside central Punjab, the former’s 5 percentage points lead lies within the survey’s margin of error of + 2.6 percentage points in this region. This, combined with the fact that this region accounts for 55 per cent of the province’s National Assembly seats, makes the contest between the two parties a real cliffhanger.

What is missing in the current debate on elections in Punjab is that the final result will depend on how the undecided voters finally choose to cast their ballot and also how many voters turn out to vote on the polling day. The Herald-SDPI survey finds that 14 per cent of the respondents in Punjab remain undecided. It is this group of voters that will clearly determine the final result of the 2018 election. This finding is consistent with the polls conducted by Gallup and Pulse Consultants during May 2018 which also show that undecided voters hold the election in Punjab in the balance.

To win a majority in the province, therefore, PTI has to swing a substantial portion of the undecided voters to its side. PMLN, too, will need to ensure that a large part of these voters ultimately vote for it. To achieve these goals, the two parties will have to ensure a healthy turnout of their supporters at the polling booths.

The survey shows strong voting intentions among the supporters of both the parties. Approximately 70 per cent of their supporters have reported a ‘strong’ or ‘very strong’ intent to vote. A bigger challenge for both of them is that a much smaller portion of the undecided voters in Punjab (52 per cent) reports a ‘strong’ or ‘very strong’ intent to vote. Being able to get these voters out on the polling day will be the real organisational test for the two parties.

It is also important to underscore here that support for different parties as shown in the survey does not allow a forecast of the number of seats that each party can win. This is because of two key reasons. Firstly, how support for different parties as reported by the survey translates into vote shares at the national level depends on which way the 13 per cent undecided voters swing and which party can motivate its supporters to turn out in larger numbers on the polling day to vote. Secondly, the number of seats a party wins depends on how concentrated or spread out its support base is across constituencies. Because the survey sample is not representative at the constituency level, it is not possible to estimate whether and by how much the support bases for different parties are concentrated or spread out.

Outside Punjab, PMLN is facing another challenge: While 40 per cent of the survey respondents in the province support the party, backing for it falls dramatically in other provinces. It has 10 per cent support in Khyber Pakhtunkhwa, 11 per cent in Balochistan and only 4 per cent in Sindh. To have a chance of forming the next federal government, it has to not only hold on to its lead in central Punjab but also win by more than a slim margin in the province’s districts outside this region.

The big question for the party is what effect Nawaz Sharif’s incarceration – and that of his daughter Maryam Nawaz Sharif – will have on the undecided voters in Punjab. Unfortunately, the timing of the survey precludes the possibility of assessing this effect.

For PTI, the path to forming the federal government runs through Punjab and Khyber Pakhtunkhwa. The party seems to have bucked the historical trend of anti-incumbency in the latter province and has sustained support there.

The survey shows that 42 per cent of the respondents in Khyber Pakhtunkhwa support PTI which has a 30 percentage point lead over other main contenders — PPP, Muttahida Majlis-e-Amal (MMA) and PMLN. This is also in line with other polls.

Like PMLN, however, PTI has to convert its improved performance in polls in Punjab into a substantial win in terms of National Assembly seats. The key to this lies with the all-important undecided voters in the province as well as the party’s electoral performance in the districts outside central Punjab.

The survey shows that 54 per cent of the respondents in Sindh support PPP, giving it a substantial lead of 40 points over its challengers — PTI and Grand Democratic Alliance (GDA). The party’s continued dominance in Sindh is likely to make it an important player in the next parliament. If Punjab returns a divided verdict, PPP may end up having an important say in who forms the next federal government.

The survey also suggests that a major change may take place in Sindh’s urban areas where it is difficult to predict the victory of any party. The 2018 elections, thus, have the potential to open up politics in urban parts of the province if other parties are able to translate the reduced support for the Muttahida Qaumi Movement (MQM), as shown in the survey, into gains in terms of National Assembly seats.

Like urban Sindh, Balochistan appears to be a divided polity in terms of party support. No single party has a clear lead here – with PPP, PTI and PMLN enjoying between 11 per cent and 15 per cent support and MMA not being far behind at 8 per cent. Regional parties, such as the Balochistan National Party (BNP) and Pakhtunkhwa Milli Awami Party (PkMAP), enjoy the support of 6 per cent and 4 per cent of the survey respondents respectively. This dispersed support suggests that the electoral fortunes of the province may be decided by the alliances and seat adjustments that different parties and candidates have made.

The Herald-SDPI survey was carried out in 55 districts across Pakistan between June 25 and July 12 this year, with 6,004 randomly chosen respondents — 2,848 from Punjab, 1,117 from Khyber Pakhtunkhwa, 1,055 from Sindh and 984 from Balochistan. The sample is provincially representative in Khyber Pakhtunkhwa and Balochistan. In Punjab, the survey has a sample representative of both the central region (which includes the divisions of Lahore, Gujranwala and Sahiwal and the district of Faisalabad) and the rest of the province. In Sindh, the sample is representative of urban areas (which include Karachi and Hyderabad districts) as well as the rest of the province.

To collate the survey findings at the national level, provincial samples have been given weightage in accordance with each province’s actual share in the country’s latest population as per the census carried out in 2017. Within each province, district samples have been given weightage according to each district’s population share in the census. Provincial samples have been weighted to achieve an equal gender representation and also to reflect the educational attainment of the population as per the latest Pakistan Social and Living Standards Measurement survey.

Ali Cheema is Associate Professor at Lahore University of Management Sciences and Senior Fellow at the Institute of Development and Economic Alternatives. Asad Liaqat is a PhD Candidate in Public Policy at Harvard University.

Data coordination: Namrah Zafar Moti, Sarah Dara and Aliyah Sahqani.

Sampling and data analysis: Ahsan Tariq, Fatiq Nadeem and Ahsan Zia Farooqui.

Data collection: Institute of Research, Advocacy and Development (IRADA), Punjab Lok Sujag, Nari Foundation, Yusra Jabeen, Sharjeel Arshad, Tariq Ahmed, Gulab Ahmed, Muhammad Arif, Qadir Dino, Zafar Musayni, Saddam Jamali, Mumtaz Sajidi, Fataullah Kasi, Gohar Rafique, Aimal Khan, Muhammad Arif, Aziz Khan, Abid Sherani, Masood Achakzai and Abdullah Jan.

This survey has been financially supported and supervised by the Sustainable Development Policy Institute (SDPI), designed by the Institute of Development and Economic Alternatives (IDEAS) and carried out by the Herald magazine.