# Does Spending Money Help Maltese Politicians Rake in More Votes?

We found out

Do local politicians win more votes depending on how much they spend on campaigning and advertisements? After politicians published their declared expenses of June 3rd, we decided to find out. To do this, we plotted a graph showing their declared spend against their first count votes. If a candidate stood for election in two districts, the number used is the sum of these districts, since the amount spent they declared is for both districts.

An immediate problem is that the two candidates who earned most votes, Prime Minister Joseph Muscat and Opposition Leader Simon Busuttil, both declared they spent zero on campaigning â€“ which makes sense given they were the leaders of both of their parties and making the majority of appearances for duration of the campaign. But both Muscat and Busuttil earned so many votes that they were adversely affecting any relationship between spending and votes - so they were both excluded from the graph.

Scatterplots are useful because they allow you - in the past through mathematics and nowadays through the magic of computer software - to determine lines of best fit through the points by a process known as linear regression.

If the number of votes went up as politicians spent more money, the graph would look like the figure on the left. But if there was no relationship, it would resemble the one in the right.

### So how does our scatterplot look like when you fit the line?

Unless youâ€™re Joseph Muscat or Simon Busuttil, spending more money on your campaign generally helped. Since the lower left-hand corner is a bit hard to make out, given that most elected candidates spent less than â‚¬10,000 and got less than 4,000 votes, hereâ€™s that area zoomed in:

The graph shows that on average, above 1,666 votes, each candidate would win an additional vote with every â‚¬7.5 spent. Perhaps the best way to think of this scatterplot would be like this:

Additionally, we can also divide the total money spent by each candidate by the number of votes earned, allowing us to determine much each candidate spent per vote. The average spent was â‚¬4.5 per vote, although individual candidates varied wildly across that, with Stefan Zrinzo Azzopardi spending â‚¬23.3 per vote on one extreme, and Stephen Spiteri spending â‚¬0.3 per vote on the other.

### How does this compare internationally?

The most comprehensive look at how much campaigning impacts votes is probably carried out in the United States â€“ the closest elections to our MPâ€™s would-be members of congress. In 2012, the 435 elected congressmen on average spent $9.84 per vote, although the median for the past three congress races has been lower, at $7.3 per voter. The cost also varies greatly depending on the state.

The effectiveness of advertisement is another issue entirely: in the United States television advertisements tend to make only slight 2-3% gains for candidates, yet that could be enough in a close race. Ultimately, campaigning has evolved into an arms race where one candidate canâ€™t afford not to outdo the other, driving up costs with each election cycle.

### For the mathematically inclined…

…who know that linear regressions give you equations and want them, the lineâ€™s equation is y = 0.134x + 1666.8. This means that above 1,666 votes, for each additional euro spent, candidates would get 0.13 of a vote. The RÂ² of the line is 0.22.

The sharp-eyed among you might have also noticed that the average from dividing the votes by the costs was different to that we got in the linear regression. The reason for this is that a linear regression calculates an intercept, in this case, the number of votes a candidate would get if they spent nothing on campaigning. In this case, 1666.8 votes.

Calculating the numbers for each individual candidate on the other hand spreads all that cost over each voter, including for instance their mother.

## Comments