Forecasting Election Results in the Czech Republic
| Název práce v češtině: | Předpovídání výsledků voleb v České republice |
|---|---|
| Název v anglickém jazyce: | Forecasting Election Results in the Czech Republic |
| Klíčová slova: | předpovídání, volby, dynamický lineární model |
| Klíčová slova anglicky: | forecasting, elections, dynamic linear model |
| Akademický rok vypsání: | 2017/2018 |
| Typ práce: | diplomová práce |
| Jazyk práce: | angličtina |
| Ústav: | Institut ekonomických studií (23-IES) |
| Vedoucí / školitel: | prof. PhDr. Tomáš Havránek, Ph.D. |
| Řešitel: | skrytý - zadáno vedoucím/školitelem |
| Datum přihlášení: | 07.11.2017 |
| Datum zadání: | 07.11.2017 |
| Datum a čas obhajoby: | 16.09.2019 09:00 |
| Místo konání obhajoby: | Opletalova - Opletalova 26, O105, Opletalova - místn. č. 105 |
| Datum odevzdání elektronické podoby: | 31.07.2019 |
| Datum proběhlé obhajoby: | 16.09.2019 |
| Oponenti: | PhDr. František Čech, Ph.D. |
| Kontrola URKUND: | |
| Seznam odborné literatury |
| Lewis‐Beck, Michael S. "Election forecasting: principles and practice." The British Journal of Politics & International Relations7.2 (2005): 145-164.
Lewis-Beck, Michael S., and Ruth Dassonneville. "Forecasting elections in Europe: Synthetic models." Research & Politics 2.1 (2015): 2053168014565128. Malkiel, Burton, G.. 2003. "The Efficient Market Hypothesis and Its Critics ." Journal of Economic Perspectives, 17(1): 59-82. Rothschild, David. "Forecasting elections: Comparing prediction markets, polls, and their biases." Public Opinion Quarterly 73.5 (2009): 895-916. Rothschild, David, and Neil Malhotra. "Are public opinion polls self-fulfilling prophecies?." Research & Politics 1.2 (2014): 2053168014547667. Rothschild, David M., and Justin Wolfers. "Forecasting elections: Voter intentions versus expectations." (2011). Available at SSRN: https://ssrn.com/abstract=1884644 or http://dx.doi.org/10.2139/ssrn.1884644 Silver, Nate. "How the FiveThirtyEight Senate Forecast Model Works." Internet, http://fivethirtyeight. com/features/how-the-fivethirtyeight-senate-forecast-model-works/, accessed on February 8 (2018): 2015. Silver, Nate. "How FiveThirtyEight Calculates Pollster Ratings." Internet, https://fivethirtyeight.com/features/how-fivethirtyeight-calculates-pollster-ratings/ accessed on February 8 (2018): 2015. Williams, Leighton Vaughan, and J. James Reade. "Forecasting elections." Journal of Forecasting 35.4 (2016): 308-328. Wolfers, Justin, and Eric Zitzewitz. 2004. "Prediction Markets." Journal of Economic Perspectives, 18(2): 107-126. |
| Předběžná náplň práce |
| The first step for building the forecasting model is to collect the data for polls and pre-election surveys from previous elections in the Czech Republic and rating them based on their historical performance. This rating will then be used to calculate the aggregate forecast. Each poll has its own statistical error and therefore aggregating them into an average improves the precision of the predicted outcome as it decreases the overall error of the forecast. The information from the betting markets will be added to the model. The methodology will follow models used by David Rothschild (predictwise.com) and Nate Silver (fivethirtyeight.com).
The forecast is created in five steps: 1. Collection of polling data. Determining weights for each poll based on the historic accuracy, sample size, recency, etc. Then calculating the weighted average. 2. Adjusting polls when necessary, depending on the type of the elections, methodology of the poll conducted. Adjusting for trends and possible biases. 3. Combining polls with demographic and economic data. Determining regional and demographic differences in voters’ preferences and evaluating the current economic situation. The forecast assumes that better economic situation favours the incumbent. Allocating the undecided voters. The demographics have decreasing weight in the forecast as the actual election approaches and more and more voters are decided as well as more polls are available. 4. Calculating forecast form prices on the betting market, which should include all available information in given time. Combining this forecast with the averaged polls only forecast and averaged polls plus demographic data. 5. Accounting for uncertainty and simulating the election to get the probabilities. The uncertainty decreases towards the elections. There are three types of errors included in the forecast. Firstly, the national error, which accounts for systematic bias in the polls across country, it is based on the time until the elections, number of undecided voters. Secondly, there is demographic error, which considers common demographic factors such as religion, race, education and accounts for bias in the polls for these groups. And thirdly, the region-specific error which accounts for bias in polls in given region as the elections are conducted by regions or voting districts. |
| Předběžná náplň práce v anglickém jazyce |
| The first step for building the forecasting model is to collect the data for polls and pre-election surveys from previous elections in the Czech Republic and rating them based on their historical performance. This rating will then be used to calculate the aggregate forecast. Each poll has its own statistical error and therefore aggregating them into an average improves the precision of the predicted outcome as it decreases the overall error of the forecast. The information from the betting markets will be added to the model. The methodology will follow models used by David Rothschild (predictwise.com) and Nate Silver (fivethirtyeight.com).
The forecast is created in five steps: 1. Collection of polling data. Determining weights for each poll based on the historic accuracy, sample size, recency, etc. Then calculating the weighted average. 2. Adjusting polls when necessary, depending on the type of the elections, methodology of the poll conducted. Adjusting for trends and possible biases. 3. Combining polls with demographic and economic data. Determining regional and demographic differences in voters’ preferences and evaluating the current economic situation. The forecast assumes that better economic situation favours the incumbent. Allocating the undecided voters. The demographics have decreasing weight in the forecast as the actual election approaches and more and more voters are decided as well as more polls are available. 4. Calculating forecast form prices on the betting market, which should include all available information in given time. Combining this forecast with the averaged polls only forecast and averaged polls plus demographic data. 5. Accounting for uncertainty and simulating the election to get the probabilities. The uncertainty decreases towards the elections. There are three types of errors included in the forecast. Firstly, the national error, which accounts for systematic bias in the polls across country, it is based on the time until the elections, number of undecided voters. Secondly, there is demographic error, which considers common demographic factors such as religion, race, education and accounts for bias in the polls for these groups. And thirdly, the region-specific error which accounts for bias in polls in given region as the elections are conducted by regions or voting districts. |