Two Serbian polls split over opposition threshold

Two polls published two weeks apart hand the same electorate two different endings, and they part ways exactly where the opposition is weakest. The survey by New Serbian Political Thought (NSPM), released on the evening of July 29, shows four more lists clearing the electoral threshold alongside the two front runners. The survey by CRTA, conducted with the Democracy Action Lab at Stanford University, measures opposition parties one by one and gives none of them a full percentage point.

The gap is not a matter of wording. It decides whether opposition parties end up with parliamentary groups or with nothing at all, and it opens before a single voter says a word, inside the questionnaire the pollster carries.

What the NSPM numbers show

Among decided voters, NSPM records the following:

• student list, 39.6 percent
• Serbian Progressive Party list, 37.5 percent
• coalition of the People’s Movement of Serbia, New Face of Serbia, the Ecological Uprising and the Movement for the Decentralisation of Serbia, 5.6 percent
• Socialist Party of Serbia, 4.9 percent
• coalition of the Party of Freedom and Justice, Serbia Centre and the Movement of Free Citizens, 4.2 percent
• the group around Branimir Nestorovic, 3.8 percent
• New DSS, 2.3 percent

Roughly a quarter of respondents declined to answer the question on party choice, a reserve large enough to move any figure on that list.

The same parties disappear at CRTA

The CRTA and Democracy Action Lab survey gives the student list 44.9 percent, the Serbian Progressive Party 35.7 and the Socialist Party of Serbia 3.8. Opposition parties there were measured individually and land between 0.3 and 0.6 percent, so not one of them comes close to the threshold on its own.

Two things belong next to that headline number. First, the 44.9 percent figure is not from last night; it was published in mid July and has been retold since as the final word on the ratings. Second, the fieldwork behind it ran from June 10 to June 24, 2026, face to face, on a representative sample of 2,324 adult citizens of Serbia, and the results cover decided voters, who make up 70 percent of those surveyed. Almost six weeks of a political season separate that fieldwork from the reading it is now being given.

Why does one poll produce a threshold and the other does not?

Because they do not ask the same question. NSPM measures coalitions that have already been grouped, offering voters a list as it would appear on the ballot, with three or four parties under one name. CRTA measures parties as they exist today.

Measured separately, the People’s Movement of Serbia, New Face of Serbia, the Ecological Uprising and the Movement for the Decentralisation of Serbia all sit inside the margin of error. Added into a single line on the form, they take 5.6 percent. No voter changed their mind between the two surveys; what changed was the menu they were shown.

NSPM editor Djordje Vukadinovic faults CRTA for leaving several parliamentary parties out of its analysis, among them the Ecological Uprising and New Face of Serbia. He summed up his own findings in one line:

Our survey is both an encouragement and a warning

The encouragement is the combined opposition bloc; the warning is the margin of just 2.1 percentage points between the student list and the Progressives. On the position of President Aleksandar Vucic, Vukadinovic said:

For the first time, Vucic is entering an election story that, at the very least, does not have that already known and guaranteed ending

A decision no poll can make

When we wrote about an opposition heading into the elections in three separate columns, the open question was how many of those columns would actually reach the ballot. These two surveys have now put a price on that question. Grouped, the same opposition holds two lists above the threshold and a say in who forms a majority. Scattered, it holds zero.

Thresholds are not cleared in opinion research but at the moment an electoral list is filed with the Republic Electoral Commission. Until that deadline every one of these figures stays conditional, and the distance between 39.6 and 44.9 percent says less about the student list than about who asked, and when.