Parliamentary elections in Latvia have a proportional representation electoral system (similar to most of the EU countries) where seats in parliament are allocated in a way that they represent the proportion of the casted votes. In Latvia that means a voting system where the voters cast a vote for a party list, which contains all of the candidates (list of candidates differs between electoral districts). Each voter has an option to vote for a party list without making any changes but they also can make changes by placing a plus sign beside the name or names of their preferred candidates or by striking out the candidates who they do not support.

Therefore, every political candidate has an incentive to gain individual support from their electorate in order to increase their individual chances to be elected. One of the characteristics of parliament elections in Latvia is the noteworthy number of regional candidates who have most of their supporters localized in a specific geographic area (e.g. local municipality) of the whole electoral district. This research paper focuses on the research methods on how to identify these candidates and how to analyse their key characteristics. In this research the approved parliamentary election results of the 8th, 9th, 10th, 11th Saeima elections (2002 - 2014) was used. The results of all political parties participating in the parliamentary elections were acquired and analyzed but for those political parties who overcame the 5% threshold (minimal treshold to get elected members in the parliament), the statistics of the pluses and strike outs of each candidate in all 5 electoral districts of Latvia were also examined.

In order to do quantitative analysis that focuses on parliamentary election results of Latvia there are certain limitations of the data that needs to be addressed. The most detailed level of analysis is available on polling places that have a considerable variation in their size. In the most recent 12th Saeima elections in 2014 the average polling station had 860 voters (varying from 4 to 5692 voters between polling stations). Therefore any quantitative analysis must take in to account large degree of uncertainty in the results of individual polling stations. It must be also noted that only a part of the electorate are making changes (by adding pluses or striking out certain candidates) in the party list. In 12th Saeima elections, only 62 % of voters made any changes in their casted ballots.

Parliamentary election results of Latvia for election candidates can be quantitatively analysed in different ways but most significant results can be obtained by analysing the number and frequency of pluses and strikeouts for each candidate since this parameter directly influences a candidates chance to be elected. It must be mentioned that these indicators are hard to meaningfully compare between different political parties and electoral districts.

By analysing the frequency of pluses and strikeouts the number of individual voters each election candidate has “attracted” to his political party can be approximated. Analysis carried out for candidates in 11th Saeima elections in 2011 shows that for 410 candidates in those political parties who gained at least one elected member of the parliament the average number of approximate attracted voters is 740 with a significant variation between candidates. The best 13 candidates had attracted more than 5000 voters each and these candidates are in almost all cases nationally known politicians with considerable support in almost all municipalities in an electoral district. The worst 70 candidates have attracted less than 10 voters each which means that these voters have almost no municipalities (or even polling stations) where their support is statistically noticeable.

Regional election candidates usually have specific characteristics that define their support and its geographical variation inside the electoral district. Number of attracted voters for these candidates are less than for those candidates who are well known nationally and they usually are significantly higher in polling stations located in some (usually adjacent) municipalities. Therefore these candidates can be quantitatively defined by analysing the spatial autocorrelation of their individual election results (by using Moran I spatial autocorrelation). Those parliamentary election candidates whose results have the highest spatial autocorrelation are usually local level politicians or local specialists or entrepreneurs who decided to be on the ballot in parliamentary elections. The analysis of these results also shows that these candidates are more frequent in rural municipalities than in cities perhaps indicating a stronger role of local identity these rural municipalities.