NEW YORK: Television ads can move the needle for aspiring presidential candidates, but the impact is largely short term in nature, according to figures from The Economist.

"We found that paid TV airtime did matter, accounting for a modest 13% of the week-to-week changes in polling," the news title revealed in a new report.

To reach this total, the Republican field was split into pairs. Each day both candidates scored over 10% in the polls, The Economist counted how many TV spots – positive and negative – had been aired about the two candidates in the last week in Iowa and New Hampshire.

By measuring the ratio of favourable and unfavourable TV ads – a medium which remains a "the staple on campaign shopping lists" – against polling averages, the study could show if these messages had made a tangible impression.

"If there were any pay-off to media spending, then candidates who appeared in lots of positive ads and few negative ones should have gained ground when compared with their rivals," it argued.

Placing this idea into context, viewers in Iowa saw 866 more positive ads about Marco Rubio than Donald Trump from January 24th to January 30th. The business mogul was also the subject of 220 more negative spots than Rubio.

"After adjusting for their standing in national polls, the front-runner's advantage over the Florida senator duly shrank by 5.1 percentage points," reported The Economist.

"Overall, holding nationwide polls constant, we found that candidates could expect to gain a one-point edge over their rivals in the next week's early-state polling for roughly every 200 net positive ads about them, or every 500 net negative ones about their opponents."

Another key finding, however, was that TV commercials had a short-term effect: the impact of positive ads during the Republican race so far has been 4.4 times greater in the week they aired than over the next seven days.

Such conclusions came with methodological caveats, too. In the first instance, The Economist sought to control for the impact of news events relating to each presidential hopeful by comparing polls in Iowa and New Hampshire with national figures.

But it also warned that factors like rallies, local media coverage and natural "random variation" inevitably influence the scores registered in surveys – as well as remarking that polls themselves are often a "poor proxy" for actual outcomes.

Data sourced from The Economist; additional content by Warc staff​