Voter Turnout in Ukraine

By Alisha Armas

          In this post, I examine the effect of Ukrainian as the native language of voters on differences between voter turnout in the November and December rounds of the 2004 Ukrainian presidential election. Language and politics in Ukraine are deeply divided, especially between its western and eastern/southern regions. In recent elections, this divide has become visible in the support of political parties. In 2004, the Our Ukraine candidate, Viktor Yushchenko, won the northern and western regions, while support for the Party of Regions candidate, Viktor Yanukovych, came from the East and South (see Figure 1). Furthermore, as Figure 2 shows, native Ukrainian speakers are concentrated in the West and North, and fades as one moves southward and eastward. More than reflecting a regional divide, language influences the media that Ukrainians consume, which can shape their political motivations.

          Western and Eastern Ukrainian media provided different coverage between the two rounds of the 2004 elections that I examine. After Yanukovych claimed victory in the fraudulent November runoff against Yushchenko, hundreds of thousands of Yushchenko’s supporters protested in Kyiv’s central square. This “Orange Revolution” ended on December 3 when the Supreme Court of Ukraine ordered a revote to occur on December 26, which Yushchenko won. Throughout these events, the Ukrainian media provided fair coverage of the protests and the candidates prior to the December revote while non-Ukrainian speaking regions tend to get their news from Russian sources, which had a negative impact on the Eastern regions. [1] Because of these factors, I hypothesize that electoral districts with higher percentages of native Ukrainian speakers will have greater increases in voter turnout between the November runoff and December revote. It is important to look at the effect of language on changes in voter turnout to help understand if the Orange Revolution was a one-time surge of political participation that was confined to the Kyiv area.

           The literature identifies several factors that influences voter turnout. For example, Wolfinger and Rosenstone find that education in particular, and age and income to a lesser extent, affect whether or not individuals vote. [2] Most literature on the Orange Revolution focuses on the outcome of the election, and may only deal tangentially with voter turnout. For example, Clem and Craumer examine voter turnout to explain Yanukovych’s victory in the November election, arguing that Donetsk and Luhansk oblasts had large increases in turnout that may have been the result of ballot box stuffing or false reporting. [3]

Table 1. Summary Statistics

Variable	Mean	Std. Dev.	Min	Max	Description
Change in Voter Turnout	-3.36	5.17	-21.61	19.44	December Revote Voter Turnout – November Voter Turnout
Native Language: Ukrainian (%)	66.37	31.33	6.35	99.58	% of population whose native language is Ukrainian
Disposable Income per Capita	4394.85	892.56	380.20	7623.90	Ukrainian Hryvnia
Average Age	38.88	2.05	33.50	45.10	Average age of the population
Higher Education (%)	31.18	5.65	21.30	47.70	% of the population with higher education
Fraud	0.14	0.34	0.00	1.00	1 if the region was subject to fraud in the November runoff (Donetsk and Luhansk regions)
Observations	207

           I look directly at the differences between voter turnout in the November runoff and the December revote. This will provide a more detailed look than analysis at the regional level while still using politically significant units. I calculate the change in voter turnout from turnout data reported by the Ukrainian Central Election Commission for each of Ukraine’s 225 electoral districts. [4] I use voter turnout data from the November runoff rather than first round data, because both the November and December elections were between Yanukovych and Yushchenko, while there were over 20 candidates in the first round. However, accusations of fraud plagued the November election. Reports indicate that fraud was particularly prevalent in the Donetsk and Luhansk regions. [5] Because international observers consider the December fair, fraud in the November elections likely drives much of the decreases in voter turnout in Donetsk and Luhansk. As Figure 3 shows, many of the greatest decreases in turnout are in Donetsk and Luhansk electoral districts. Therefore, I include a dummy variable that is 1 for electoral districts in Donetsk and Luhansk, and zero for all other districts. For native Ukrainian speakers, I use data from the 2001 All Ukrainian Census, which is Ukraine’s most recent census. [6] While I prefer more recent data, the State Statistics Service of Ukraine reports low levels of migration at the the oblast level between 2001 and 2004, indicating little change in Ukrainian demographics. Therefore, the use of the 2001 data should not cause drastic problems. Additionally, to correct for negative skewness, I change Ukrainian language to the percent of the population whose native language is not Ukrainian (100 – native Ukrainian speakers (%)) and log this new variable. Therefore, for my hypothesis to be true, electoral districts with higher percentages of non-native Ukrainian speakers should have greater decreases in voter turnout between the November and December elections.

           I additionally control for income, age, and education. Income data at the subregional level is not available. Therefore, I apply 2004 regional data on per capita disposable income from the State Statistics Service of Ukraine to all districts that fall within a given region. Because most electoral districts consist of both urban and rural areas, the differences in income between electoral districts should be minimal. For age, I use the average age as reported in the 2001 census. [6] Finally, for education, I use data from the 2001 census at the regional level, since data at smaller units are not available. I choose to measure education as the percent of people who possess higher education because differences between regions are more pronounced for higher education than secondary education. While it may seem prudent to control to political affiliation, language is high correlated with support for either Yanukovych’s Party of Regions or Yushchenko’s Our Ukraine. Since I believe that language, partially through the media, influences both voter turnout and political ideology, I do not include a variable for political affiliation or ideology.

 

Table 2. Determinants of Changes in Voter Turnout in 2004 Ukrainian Presidential Elections

DV: Changes in Voter Turnout [7]	(1)	(2)	(3)	(4)	(5)
Native Language Not Ukrainian(log) [7]	-0.076***	-0.073***	-0.076***	-0.078***	-0.015
	(-4.00)	(-3.73)	(-4.02)	(-3.65)	(-0.68)
Income(log) [7]		-0.059	0.024	0.004	0.101
		(-0.49)	(0.20)	(0.02)	(0.76)
Average Age			-0.054***	-0.055***	-0.036**
			(-4.18)	(-4.18)	(-2.85)
Higher Education(log) [7]				0.054	-0.266
				(0.25)	(-1.26)
Fraud					-0.486***
					(-5.84)
Constant	3.132***	3.617***	5.051***	5.048***	4.480***
	(52.84)	(3.64)	(4.98)	(4.96)	(4.73)
R-Squared	0.07	0.07	0.15	0.15	0.27
Observations	207	207	207	207	207

Note: OLS estimates with t-stats in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001

Source: Ukrainian Central Election Commission, 2001 All Ukrainian Census, and State Statistics Service of Ukraine

           For a 1% increase in the percent of non-native Ukrainian speakers, the change in voter turnout will, on average, decrease by 0.02%, all else constant, which supports my hypothesis. However, this is not proof of causality since changes in voter turnout may cause reported native language. After all, Ukrainians may vote out of patriotism, and so, also out of patriotism, report their native language as Ukrainian, the only official language of Ukraine. However, high voter turnout is not confined to Ukrainian-speaking districts. In the next presidential election in 2010, districts with few native Ukrainian speakers also had high turnout rates (see Figure 4). In fact, districts where native Ukrainian speakers composed less than 50% of the population had an average turnout rate of 70.34%, while districts where 50% or more of the population were native Ukrainian speakers had an average turnout of only 68.65%.

           Furthermore, this relationship lacks statistical significance, meaning that language may have no effect on the change in voter turnout. This lack of statistical significance only occurs once the fraud variable, which is based on region, is introduced. It is possible that region has a powerful effect on voter turnout. However, in 2004, language, region, and fraud were all closely related: regions that experienced fraud in the November elections were also in the East and had few native-Ukrainian speakers. Yet, even in the absence of the fraud variable, the effect of language on voter turnout is small. It seems unlikely that language was a powerful motivator in the differences in voter turnout. While this does not support my hypothesis, the small effect of language may mean that increased voter turnout and the political participation that occurred during the Orange Revolution may occur among non-native Ukrainian speakers as well.

Bibliography

Blais, Andre. "What Affects Voter Turnout?"  Annual Review of Political Science 9(2006): 111-25. doi:10.1146/annurev.polisci.9.070204.105121.

Clem, Ralph S., and Peter R. Craumer. "Shades of Orange: The Electoral Geography of Ukraine's 2004 Presidential Elections."  Eurasian Geography and Economics 46, no. 5 (2005): 364-85. doi:10.2747/1538-7216.46.5.364.

"Course of Voting in Oblasts of Ukraine." Central Election Commission of Ukraine - "The Elections of the President of Ukraine" 2004. Accessed February 26, 2015. http://www.cvk.gov.ua/pls/vp2004/wp0011e.

"Databank." 2001 All Ukrainian Census. 2015. Accessed February 26, 2015. http://database.ukrcensus.gov.ua/MULT/Database/Census/databasetree_en.asp. 

Karatnycky, Adrian. "Ukraine's Orange Revolution." Foreign Affairs, March/April 2005. Accessed March 13, 2015. www.foreignaffairs.com/articles/60620/adrian-karatnycky/Ukraines-orange-revolution.

"Рорulation Income in Region of Ukraine." State Statistics Service of Ukraine. January 21, 2015. Accessed February 26, 2015. http://ukrstat.gov.ua/.

Tora, Alex. "Population of those whose mother language is Ukrainian in Ukraine (2001)." WikiMedia Commons. December 12, 2008. Accessed February 26, 2015. http://commons.wikimedia.org/wiki/File:Ukraine_census_2001_Ukrainian.svg.

"Ukraine Elections Map Nov2004." Wikipedia. December 4, 2004. Accessed February 26, 2015. http://en.wikipedia.org/wiki/File:Ukraine_ElectionsMap_Nov2004.png.
Wolfinger, Raymond E. and Steven J. Rosenstone. Who Votes?. New Haven: Yale University Press, 1980.

---

[1] Adrian Karatnycky, "Ukraine's Orange Revolution," Foreign Affairs, March/April 2005, accessed March 13, 2015. www.foreignaffairs.com/articles/60620/adrian-karatnycky/Ukraines-orange-revolution.

[2] Raymond E. Wolfinger and Steven J. Rosenstone, Who Votes? (New Haven: Yale University Press, 1980).
[3] Ralph S. Clem and Peter R. Craumer, "Shades of Orange: The Electoral Geography of Ukraine's 2004 Presidential Elections."  Eurasian Geography and Economics 46, no. 5 (2005). doi:10.2747/1538-7216.46.5.364.
[4] For a list of electoral districts by region, see Appendix A.

[5] Karatnycky, "Ukraine's Orange Revolution."
[6]  This data is available at the raion and major city level, which must be combined to obtain information for electoral districts. For more details on calculating electoral district level from census data, see Appendix B.

[7] Because change in voter turnout, income, and higher education are significantly positively skewed, I log these variables.

APPENDIX A. Electoral Districts by Region

Region	District Numbers
Crimea	1-10
Vinnytsia	11-18
Volyn	19-23
Dnipropetrovsk	24-40
Donetsk	41-63
Zhytomyr	64-69
Zakarpatska	70-84
Ivano-Frankivsk	85-90
Kyiv	91-99
Kirovohrad	100-104
Luhansk	105-116
Lviv	117-128
Mikolaiv	129-134
Odesa	135-146
Poltava	147-154
Rivne	155-159
Sumy	160-165
Ternopil	166-170
Kharkiv	171-184
Kherson	185-189
Khmelnyts	190-196
Cherkasy	197-203
Chernivtsi	204-207
Chernihiv	208-213
Kyiv (City)	214-223
Sevastopol	223-225

 

APPENDIX B. Calculating the Native Language: Ukrainian and Average Age Variables

           The 2001 All Ukrainian Census reports information on native language and average age at the raion and major city level, while electoral districts are composed of one or more of these areas. In the case of some of the larger cities, electoral districts may be composed of a portion of the city. In the cases where only a portion of the city composes an electoral district, census data for the entire city are used for the electoral district. In cases where electoral districts are composed of several raions and/or cities, the language or age is weighted by the population, and I add the weighted values together and divide by the total population. For Kyiv (city), the electoral districts are the same as the districts for the census, so no changes were made.

Example: Electoral District 180 (Kharkiv Oblast)

           The Ukrainian Central Election Commission states that Electoral District 180 consists of the cities of Izium and Kupiansk, and Borivskyi, Iziumskyi, Kupianskyyi and Shevchenkivskyi raions. The 2001 census reports the following information for native Ukrainian speakers and population for each of these areas as follows:
 

City or Raion	Native Language: Ukrainian (%)	Population
IZIUM	74.22	56,114
KUPIANSK	69.28	62,620
BORIVSKYI RAION	91.52	21,128
IZIUMSKYI RAION	91.67	22,002
KUPIANSKYI RAION	88.94	29,581
SHEVCHENKIVSKYI RAION	91.23	23,213
TOTAL:		214,658

Weighting the language by population:

City or Raion	Weighted Language (Ukrainian Language * Population)
IZIUM	4164781.08
KUPIANSK	4338313.6
BORIVSKYI RAION	1933634.56
IZIUMSKYI RAION	2016923.34
KUPIANSKYI RAION	2630934.14
SHEVCHENKIVSKYI RAION	2117721.99
TOTAL:	17202308

Finally, the total weighted language is divided by the total population and the resulting value is used as the percent of the population for the electoral district who are native Ukrainian speakers. In this case, 17202308 ÷ 214658 = 80.14%.