Five weeks, 4,000 Facebook posts: social media campaigning in the 2023 election

Predictions this year’s election would see a deluge of disinformation didn’t pan out, but analysis of parties’ Facebook posts found not everyone stuck to the unvarnished truth.

Close up image of Facebook logo

Over five weeks leading up to the 2023 election, more than 4,000 Facebook posts by political parties and their leaders were analysed by Dr Mona Krewel, director of the Internet, Social Media and Politics Research Lab at Te Herenga Waka—Victoria University of Wellington, and her team.

Social media is firmly entrenched in modern election campaigns, but results show most parties put less effort into social media this year than they did in the 2020 election, Dr Krewel says. Results also show the disinformation deluge that was predicted by some pundits didn’t arrive.

Sure to rise? Trends in social media campaigning

Ahead of the 2023 election, various commentators named social media as the key battleground of modern campaigning and the place to be for parties and candidates, Dr Krewel says. However, forecasts the number of Facebook posts would increase this year didn’t play out.

In the five-week period leading up to election day on 14 October 2023, the Labour Party, the National Party, Te Pati Māori, and the Green Party put up fewer Facebook posts compared with same period in the 2020 election. New Zealand First posted about the same amount.

“Only the ACT Party and leader David Seymour posted more. ACT quadrupled the number of its Facebook posts and clearly placed a high emphasis on its social media campaign in 2023,” she says.

Why did the expected increase in social media campaigning fail to materialise?

“The most likely explanation is that face-to-face campaign events and traditional door-to-door canvassing made a comeback in 2023. In the 2020 election, during which parties needed to be prepared for possible COVID-19 lockdowns, they focused more on social media campaigning. They knew it would provide a fallback option to reach voters if a lockdown occurred.

“In the 2023 election, this was no longer necessary. In this sense, our results indicate that every campaign is different. Statements about social media campaigning inevitably increasing from election to election should therefore be treated with caution, as the specific context of the campaign affects campaigners’ decisions.”

Graph showing number of Facebook posts

Personalised campaigning

The two major political parties both ran highly personalised social media campaigns in this election, Dr Krewel says.

“More than half of posts (54 percent) by Labour and National focused on the party leader, indicating that personalisation of social media campaigns has become ingrained across the political spectrum from left to right. It also shows this type of personalised campaigning, seen internationally, has reached party-centred political systems such as New Zealand’s.”

Some researchers have attributed this trend to the “Americanisation” of political campaigning.

“Since the mid-1990s, there’s been a lot of debate in political communication research about the alleged Americanisation of political campaigns—the view that campaigns outside of the United States increasingly look like campaigns inside the US.

“However, many scholars have pointed out the changes we see should really be labelled as a ‘modernisation’ of campaigns rather than an Americanisation, because they would have occurred with or without the US as a role model for campaigning.

“Parties in the US and elsewhere face the same trends in voting behaviour, with voters becoming more volatile, and the same changes in media technology be it the introduction of TV to campaigns in 1960s, or the internet in the early 2000s, or social media now.”

Parties adapt their campaigns and modernise them, which affects their campaign styles, she says.

“However, no matter how you label these changes, they’re associated with an increase in the personalisation of campaign communication—that’s the case even in more party-centred political systems. In the past, this was mostly a feature of more candidate-centred political systems such as the US.”

Graph showing % of personalised posts

Getting out the vote

One of the core tasks of political parties in election campaigns is to mobilise voters to turn out, Dr Krewel says.

“In an ideal world, voters get drawn into the political process as the campaign progresses, their political interest over the campaign increases, and they learn something about the political positions of the parties.”

The use of early voting in New Zealand means parties can no longer leave their mobilisation efforts to the last minute, she says.

“We’d expect to see parties increase their mobilisation efforts earlier in their campaigns and particularly at the point when early voting starts. In addition, with an estimated 80,000 overseas voters, one would expect mobilisation efforts to increase when overseas voting begins.”

So did parties mobilise strategically on social media?

“Parties increased their mobilisation efforts over the campaign. In mid-September, about a third of Facebook posts included a call to vote. By the end of the campaign, it was more than 50 percent. At this point, more than half of posts called out to voters and tried to convince them to vote.

“There was a slight push towards more calls to vote when overseas voting started (27 September), but the parties did not keep up this momentum and ‘get out the vote’ posts stagnated for a while after this time,” she says.

The focus on "get out the vote” posts also increased in the period immediately after early voting opened (2 October). However, parties didn’t maintain this focus for long and efforts slowed down around the last weekend before the election. "Get out the vote" posts increased again in the final stretch of the campaign period, Dr Krewel says.

“In summary, one could argue parties are doing their job and trying to get voters to turn out. This is particularly important given turnout is declining worldwide and New Zealand is no exception. Turnout in 2023 was 78.2 percent, compared with 82.2 percent in 2020, though this is still good by international standards.”

However, Dr Krewel believes parties could have done better: “Their mobilising efforts only lasted a short time after the two pivotal dates when overseas voting and early voting started.”

Graph showing get out the vote posts

Style over substance?

Compared with the 2020 election, parties’ social media campaigns this year had less focus on policy issues.

“Most overseas research on social media campaigns has found they are often superficial and don’t focus much on policy issues. By international standards, campaigns in New Zealand can’t be called superficial, but their focus on policy issues has decreased.

"For most of the campaign, about half of parties’ Facebook posts focused on policy. In contrast, for periods in the 2020 campaign as much as 75 percent contained policy content. Shortly before election day in 2020, political issues were mentioned in about 90 percent of posts.”

The economy was the key topic when Facebook posts focused on policy.

“In this respect, the picture was no different from 2020 and I would predict that we will see the same picture in 2026. The economy is almost always the most important topic or at least one of the most important topics for voters in an election campaign and it is rational for parties to focus on topics voters care about.”

The topic in second place? At some points in the campaign, it was domestic policy issues such as co-governance or law and order, and at others it was social issues such as free dental care.

“This result is telling, as it reveals the issues agenda in this campaign favoured National,” Dr Krewel says.

“Centre-right and conservative parties are usually considered by voters to be more competent in taking care of the economy. Centre-left and Labour or social democratic parties are seen as the most competent to deal with social issues.

"However, the fact social issues were not always the second most frequent topic of Facebook posts means Labour did not manage to make its perceived competence issue salient on social media.

“Overall, posts on social issues were outnumbered by posts on domestic policy—another area where centre-right parties are perceived as more competent and so this might therefore have helped National.”

Graph showing posts on policy issues

Negative vibes

The leaders of Labour and National both accused each other of running the most negative campaign of all time and complained about personal attacks, Dr Krewel says.

“However, the percentage of Facebook posts that included some form of negative campaigning stayed below 50 percent. The 2023 social media campaign can therefore not be considered an extremely negative one.”

Labour’s social media campaign was less negative than National’s and also less negative than the campaigns of other parliamentary parties and their leaders, she says.

“This does not come as a surprise as incumbents are always more positive than challenger parties. An incumbent defends its record and seeks confirmation from voters, while a challenger attacks and criticises the incumbent for its record and offers voters an alternative.

“From this perspective, both National and Labour did their job. The leaders’ mutual accusations of negative campaigning must therefore mostly be seen as a strategic move in the hope that voters would reject the alleged negativity of their opponent.”

However, Labour’s campaign got a little more negative closer to election day, Dr Krewel says.

“There is a plausible explanation for this. Research shows parties behind in the polls go on the attack more, as negativity usually generates more media attention than positivity. In contrast, a party leading in the polls does not risk negative campaigning, as there is always a possibility it will backfire.”

National’s campaign followed this trend with negative Facebook posts decreasing over time, as the polls turned more in its favour.

“We saw a similar picture in 2020 with a different cast: The more National and then leader Judith Collins were behind in the polls, the more negative its campaign became. Labour and then leader Jacinda Ardern simply had no incentive to go negative and instead focused on positive self-representation from what looked like a comfortable majority.

“The other point to note is that attacks usually led to counterattacks. So the more Labour and Chris Hipkins came under fire from National and Luxon this year, the more rebuttal posts Labour sent out. The media also engaged in this and handed out advice to Hipkins to counter Luxon.”

Across all parties tracked in the study, parliamentary parties were slightly less negative than others, Dr Krewel says.

Graph showing negative posts

Trickle of disinformation

The level of fake news in this campaign remained similar to that seen in the 2020 election.

“At the start of the 2023 campaign, 2.5 percent of posts were fake news, the same level as in 2020. The share of fake news then fell, before slightly increasing at the end of the campaign. It peaked at 3.8 percent at the start of the last week of campaigning. These results don’t provide grounds for spreading fears about unprecedented levels of disinformation.”

Throughout the five-week period, fake news posts fluctuated between 0.7 and 3.8 percent of all parties’ Facebook posts, Dr Krewel says.

“That will hardly sway an election, especially when these posts mostly come from a few fringe parties. When you look at the number of Facebook followers these fringe parties have, as well as their election results, not a lot of people are exposed to their posts nor do many New Zealanders seem to believe their claims.”

The picture changed slightly when posts containing half-truths were counted.

“Half-truths were categorised as posts that for the most part are correct but often contain small inaccurate details. The percentage of posts containing half-truths was higher than the percentage of fake news posts. However, the amount by no means can be considered alarming, peaking at about 6.5 percent by the end of the campaign.”

Graph showing posts with half truths and fake news

The rhetorical strategies used in these posts ranged from emotional appeals to ad hominen attacks.

"Just over a third (37 percent) of disinformation posts relied on appeals to people’s emotions, reflecting the fact the parties posting them could not present any hard evidence for the accusations or conspiracies they were spinning,” she says.

"Disinformation posts can also jump to conclusions quickly (26 percent) as they tend to make mountains out of molehills. Arguments can be very simple (23 percent) and facts are cherry-picked (21 percent) without discussing any counterevidence.

“Sometimes they cite pseudo experts (18 percent) or anecdotes (15 percent), voice unrealistic scientific demands such as 100 percent certainty (7 percent), claim they have a (silent) majority behind them (5 percent), or make ad hominem attacks (5 percent), personally attacking their declared enemies instead of debating their issue.”

While the amount of disinformation wasn’t high, investigating the strategies disinformation posts employ can shed light on why some people fall for them and get dragged down the rabbit hole, Dr Krewel says.

“The level of disinformation posts might not always stay as low as it was in the 2020 and 2023 campaigns. Policymakers need to be prepared should a future election campaign get flooded by massive amounts of disinformation posts.

“We need to keep an eye on social media campaigns and monitor the development of disinformation, but at the same time we need to refrain from panic mongering. There is no empirical evidence social media campaigning in Aotearoa is turning us into a post-truth society.”

Graph showing types of rhetorical strategies used