Gender Differences in Reactions to TV Political Advertising: An Empirical Study of the 2015 Presidential Election in Poland

Many studies on TV political advertising suggest the gender differences in the reactions to advertising in the affective aspect, both at the level of liking and evaluation of the candidates. The article follows this trend, presenting an empirical study carried out in Poland during the 2015 presidential electoral campaign. The results show gender differences in the evaluation of candidates. We found out that women are more susceptible to changing the perception of candidates’ images as a result of the influence of TV electoral ads. Interestingly, watching advertisements resulted in a statistically significant decrease in the evaluation of most candidates. Women’s susceptibility to political advertising is also proved by changes in the emotional attitude to the candidates after watching the spots. In addition, the research showed that TV electoral ads cannot change the perception of candidates whose images are consolidated in public discourse.


INTRODUCTION
The 2015 presidential election in Poland resulted in many surprising outcomes and confirmed that a proper communication strategy can enable even quite an obscure candidate to defeat the popular incumbent president in the fight for the office of the head of state. The specificity of that campaign was even visible at the stage of candidates' registration. As many as 11 candidates managed to collect 100,000 signatures required by law: Grzegorz Braun (independent), Andrzej Duda (Law and Justice, PiS), Adam Jarubas (Polish People's Party, PSL), Bronisław Komorowski (incumbent, endorsed by Civic Platform, PO), Janusz Korwin-Mikke (KORWiN), Marian Kowalski (National Movement, RN), Paweł Kukiz (independent), Magdalena Ogórek (endorsed by Democratic Left Alliance, SLD), Janusz Palikot (Your Movement, TR), Paweł Tanajno (Direct Democracy, DB) and Jacek Wilk (Congress of the New Right, KNP). Most of them were substitute candidates, whose main goal was to convince the voters that they are "politically independent" candidates representing the interests of specific political parties and their supporters, not "political puppets assuming the role of party leader's under study" [Kolczyński 2015: 26]. With this end in view, each of them had to create their own image and consolidate it in the voters' awareness.
The 2015 presidential campaign was interesting, not because of political marketing tools used by candidates, but due to the dynamics of political preferences. Even three months before the election, according to electoral polls, Bronisław Komorowski was widely considered as the winner. He had such a great advantage over the others that his victory in the first round seemed quite real. In the following months, the incumbent president was gradually losing support in favor of Andrzej Duda. 1 Actually, the fact that Duda was chosen as the PiS's candidate by the party leader Jarosław Kaczyński was surprising in itself. According to public opinion polls, six months before the election, Duda was unrecognizable for nearly half of the Polish voters. 2 However, by beginning the campaign early and using effective marketing strategy, he did not only manage to get into voters' awareness but also, eventually, to win in the second round of the election. What was surprising was that Paweł Kukiz ranked third -a rock man, criticizing the activities of all the Polish political parties, demanding a thorough transformation of the political system through the introduction of single-member districts and greater empowerment of citizens by means of direct democracy institutions. During the campaign, Kukiz consistently maintained the image of a rebel from out of the world of politics, 3 running for the election not for some personal reasons but out of real care for the future of the state. Maintaining the authentic image helped him obtain the support of over 20% of voters and in the final 1 In the past, Andrzej Duda had served as a member of the State Tribunal, a Cracow town councillor and a deputy in the Sejm. At the moment of appointment as a candidate he was a Member of the European Parliament. The fact that he was appointed by PiS resulted from the high proportion of negative electorate of the party leader, Jarosław Kaczyński.
2 According to the Public Opinion Research Center (CBOS) poll of December 2014, 43% of voters did not know Andrzej Duda. Out of the remaining respondents, 18% declared trust in that candidate, and 14% -the lack of trust in him [CBOS 2014].
3 Kukiz managed to maintain the image of a candidate from out of the sphere of politics, although practically since the 1990s, he had been supporting various political options. In the 1997 parliamentary election, Kukiz endorsed the rightist Solidarity Electoral Action (AWS). In the 2005 presidential election, he supported Donald Tusk. Five years later, he encouraged to vote for Marek Jurek. Since 2014, he has been serving as a councillor in the Dolnośląskie Voivodeship. phase of the campaign attract the electorate who wanted to express their disapproval of the government and President Bronisław Komorowski [Pankowski 2015]. The other candidates obtained marginal support, not exceeding 4% of the valid votes.
The perception of the candidates as "secondary" was reflected in the budgets of campaign election committees. The leaders in this respect were definitely the two main rivals fighting for the office of the president, Bronisław Komorowski (over PLN 18 million) and Andrzej Duda (approx. PLN 13.7 million). The expenses of the other candidates were incomparably lower. Even the candidates of parties with strong positions in the Polish politics (SLD and PSL) allocated only slightly more than a million zlotys. The greatest outsiders were Marian Kowalski and Paweł Tanajno, who had the amounts of several hundred or even below PLN 20,000 (in the case of Tanajno).
Despite the growing role of the Internet in political communication, in the Polish conditions the majority of campaign activities still take place on TV, and the most often used instrument of political marketing is campaign spots. This can be inferred from the amounts allocated for this purpose in the 2015 presidential campaign in Poland. Out of nearly PLN 40 million spent on the campaign by all the candidates, almost 38% was the expenditure for the production and broadcasting of TV political ads. The highest budgets for this purpose were allocated by the candidates of the two main political forces: Civic Platform (PO) and Law and Justice (PiS). Bronisław Komorowski running for re-election spent more than PLN 7.3 million on the producing and broadcasting of TV ads, and Andrzej Duda spent over PLN 5.6 million. Due to limited campaign budgets, the other candidates allocated much lower amounts for TV political advertising.

THEORETICAL fRAMEWORK
Studies carried out both in Western democracies [Miller, Wattenberg, Malanchuk 1986;Pierce 1993] and in Poland [Zakrzewski 1991] show that voters much more often describe candidates in terms of their personal traits rather than in terms of their views or party affiliations. This may be explained by the fact that an average voter is not interested in politics very much and does not regularly follow political events [Kinder 1986], but also by the way electoral campaigns are presented in the media -usually as a kind of horse races. Bruce Newman [1999] calls these phenomena in a more straightforward way, figuratively stating that politics is entering the age of manufactured images. This tendency is related to the increasing importance of television during electoral campaigns. Actually, it is part of broader social changes: the transformation of a contemporary human into homo videns who has rejected written word in favor of images [Sartori 2007: 13]. Thus, the TV utterly determines the electoral process, turning the traditional representative democracy into a "video-politics" [Revelli 2019: 172]. In electoral campaigns, therefore, the decisive factor is not the real qualifications of the candidates but their created political images. Furthermore, television produces a false sense of intimacy between politicians and voters, allowing the viewer to see the candidates' qualities more clearly. Thereby television facilitates and promotes voting on the basis of the candidates' personal traits [Hayes 2009: 234]. Political entities have also noticed this phenomenon, which is the reason why electoral campaigns today are more and more personalized [Garzia 2014: 12], not only in presidential elections but also in elections for collective representation bodies. 4 Studies on the influence of advertising focus on three dimensions: cognitive (ads' information efficacy and such mechanisms as attention, exposure, awareness, recognition, comprehension, recall and engagement); affective (attitude change, level of liking/disliking and involvement); and conative (intention to buy and purchase behavior) [Wimmer, Dominick 2010: 382-388]. With reference to political advertising, the cognitive dimension is connected with the influence of advertising messages on the voters' level of knowledge on the traits and programme of each candidate. The affective dimension involves voters' perception of the image and their emotions concerning the presented candidate, and the last dimension refers to the influence of political advertisements on voters' specific behaviors, i.e. voting for a specific candidate, or even a permanent change in electoral attitudes [Kaid 2004: 166-171;2008: 564]. In the presented article we will only focus on the affective dimension, in particular on the effects of campaign spots on the evaluation of political images and the emotional attitude towards candidates.
One direction of analyses in the affective dimension is the study of differences in responses to political advertising between men and women. Some of them, i.a., international comparative studies carried out by Lynda Lee Kaid and Christina Holtz-Bacha in the United States, Great Britain, France, Germany, Romania and Poland show that gender determines the difference in response to advertising messages. Comparative analyses have confirmed that women are more susceptible to the influence of exposure to TV political spots. They are more likely to more positively evaluate the candidates after watching the ads than men [Kaid, Holtz-Bacha 2000]. Similar conclusions are drawn in studies carried out in the United States [Kaid, Tedesco 1999;Tedesco, Kaid 2003;Bystrom et al. 2004;Kaid, Holtz-Bacha 2006: 454] and Western European countries [Kaid, Gagnère 2006: 91]. Actually, the same tendency is observed with regard to commercials. Studies show that in the case of this kind of message, women are more easily influenced than men, and the factor that increases the susceptibility to the influence of advertising is the woman accepting the role attributed to her. Women who attach more importance to the traditional role of women are more susceptible to the influences than those who attach less importance to it [Eagly 1978;Kotler 1999: 548]. Other issues that may explain women's higher susceptibility to advertising messages are pointed out by Mary Banwart and Lynda Lee Kaid. These researchers have identified five factors that may potentially explain this phenomenon [Banwart, Kaid 2003: 92]. First, women generally display a lower interest in the sphere of politics, which facilitates the creation of its image through the mass media. Second, women put more stress on the personality factors of candidates, often referred to in political ads. Third, women are interested in political issues other than those men are keen on. fourth, women and men respond differently to media messages. fifth, the differences may result from the feminine and masculine communication styles. Differences in the reception of negative ads by women and men are also interesting. It has been found that women tend more than men to perceive the sponsor of the ad more negatively than the political opponent attacked in the ad [Kern, Just 1997].
However, some studies suggest that there are no significant differences in the sphere of influence of TV political advertising on voters of both sexes [Holtz-Bacha, Kaid 1995;Banwart 2007Banwart : 1160Banwart -1161. Similar conclusions have been drawn in studies carried out in Poland, yet those studies referred to the influence of spots on the perception of political parties by men and women [Olszanecka 2012;. This may mean that the influence of TV electoral advertisements depends on the conditions in which the study is performed and the campaign context -first of all, on the parties or candidates taking part in the election and the way of carrying out the electoral campaign.

METHODS
The presented study was carried out according to the quasi-experimental research design without a control group, 5 one week before the first round of the 2015 presidential election in Poland. This kind of before-and-after research procedure has been commonly used to measure effect of TV political advertising for almost 30 years [Kaid, Leland, Whitney 1992;Mazzoleni, Roper 1995;Tedesco, Kaid 2003;Cwalina, Falkowski, Kaid 2005;Kaid, Fernandes, Painter 2011;Olszanecka-Marmola 2019].
The study involved three stages: In the first stage, the group of participants anonymously filled in a study questionnaire (pre-test), including, i.a., questions concerning their interest in the sphere of politics, 6 political preferences, self-identifications on the left-right scale 7 and their emotional attitudes to the candidates competing for the office of the head of state. The instruments used in the study were feeling thermometers and semantic differential scales. Then, the respondents were shown three randomly chosen TV electoral ads of each running candidate (in each group of respondents, the order of the candidates was also chosen randomly). The ads were selected from the spots broadcast in free electoral blocks presented on public television (TVP1, TVP2). 8 Due to the random sampling, the spots used in the study were both positive ads and negative ads, as well as issue ads and image ads. The contents presented in the ads were not directly targeted at women; they were universal messages that could easily reach various segments of the electorate.
Watching the ads of a given candidate was followed by the third stage of the study, in which the respondents filled in a post-test referring to the person's image qualities and their emotional attitudes to the candidate.
Both applied measurement instruments are successfully used in psychological, political and media analyses. feelings thermometers are the standard method used to measure emotional attitudes to people, phenomena or objects [Pieńkowski, Podlaszewska 1991: 167]. Diana Mutz [2007: 82] also points out that they are very useful in studying changes of voters' emotional attitude to the candidate in a specific time frame, especially as regards the impact of particular political events. They can 5 Unlike a classic experimental study, this research design is characterized by the lack of random sampling of the experimental group and no control group [Babbie 2009: 371]. 6 Interest in the sphere of politics was measured using a 1-5 scale, where 1 meant that the respondent was not interested in politics at all and 5 meant a very high interest in that sphere. 7 Self-identification on the left-right scale was tested using a 1-7 scale: 1 reflected an extremely left orientation, 2 -left, 3 -center-left, 4 -center, 5 -center-right, 6 -right, and 7 -extremely right. Apart from that, the respondents could choose the option "I don't know / It's hard to say". 8 In the Polish Election Code there is a distinction between free "election broadcasts" and paid "electoral advertisements". Each registered candidate has the right to broadcast free election broadcasts in public radio and TV channels at the expense of the broadcasters from the 15 th day before the election until the end of the election campaign. In the present study, we used those free election broadcasts. Basically, there were two reasons for that. first, most candidates did not choose to use paid electoral advertisements during the campaign. Second, in the Polish conditions, the candidates who decide to buy broadcasting time tend to present the same ads in that time as in the free electoral blocks. also be a useful tool in the process of designing experimental or quasi-experimental studies, being a measure of the influence of the stimulus on the research group. In the present study, an emotional attitude to candidates was evaluated on a scale from 0 to 100 degrees. The results ranging from 0 to 50 degrees meant that the respondent had negative feelings toward the candidate, 50 degrees meant a neutral attitude, and values above 50 degrees indicated the respondent had positive feelings toward the candidate.
Using semantic differentials it is possible to find out what meanings certain concepts or objects have for individuals. This instrument is based on the thesis that a human absorbs descriptive and connotative (emotional) information from the environment [Osgood, Suci, Tannenbaum 1957]. By applying differentials we are able to diagnose the connotative meanings measured with pairs of opposite adjectives. For the purpose of this study, we produced 13 seven-degree scales: dishonest vs honest, conflicting vs conciliatory, unreliable vs reliable, effective vs ineffective, focused on the good of the party vs focused on the good of Poland, inexperienced vs experienced, incredible vs credible, lazy vs hard-working, incompetent vs competent, weak leader vs strong leader, arrogant vs tactful, looking bad in the media vs looking good in the media, physically attractive vs physically unattractive. These scales produced high reliability in this study: Cronbach's alpha for respective pre-/post-test image scales usually exceeded 0.90 for each candidate. 9 Hence, they could be used to construct the index of the general perception of the image of each candidate, being the mean of the values marked on 13 scales of the semantic differential.
The study sample consisted of students of international security, internal security, European diplomacy, journalism and communication, European studies, foreign language studies, physiotherapy, medicine, political studies and sociology from six universities: University of Warsaw, University of Wrocław, University of Silesia in Katowice, University of Economics in Katowice, Medical University of Silesia and Silesian University of Technology. The final analysis involved 507 people -311 women (61.3%) and 196 men (38.7%). Their voting preferences were highly compliant with the distribution of support in this age cohort in the discussed election [TVP INFO 2015]. The highest number of the respondents declared they would support Paweł Kukiz in the first round. Women intended to vote for him more often than men. Bronisław Komorowski, Andrzej Duda and Janusz Korwin-Mikke had slightly lower support from the respondents. Men more often than women intended to vote for the latter two. Interestingly, more than one-fourth of the women participating in the poll did not know which candidate they would vote for in the first round. A higher proportion of men than women declared electoral absence. Just like in other studies [Verba, Burns, Schlozman 1997 Two research questions were formulated during the study: 1. Does the voters' gender determine the degree of influence of TV electoral advertisements on the perception of images of candidates running in the 2015 presidential election?; 2. Do TV electoral advertisements influence in the same way the electoral images of different candidates running in the 2015 presidential election?
The following hypotheses were verified as part of the study: H1: Women are more susceptible to changing the evaluation of a political image as a result of the influence of TV electoral ads.
This hypothesis is based on the fact that the majority of empirical studies concerning the influence of TV electoral advertising lead to this conclusion [see: Kaid, Holtz-Bacha 2006;Kaid, Gagnère 2006], which is explained in the theoretical part of the article.
H2: Regardless of the gender of their viewers, TV electoral ads will not cause a reconfiguration of image qualities of commonly known candidates.
This hypothesis is connected with the fact (diagnosed in social psychology) that consolidated knowledge on certain objects, phenomena or people is a change-resistant cognitive structure. Thus, the more memory-rooted is the politician's image, which generates a certain attitude to the person, the more difficult it is to change it by a piece of external information [Falkowski, Michalak 2009: 252], in this case an electoral campaign spot. In the analyzed election, the following candidates had images that were strongly consolidated in the electorate's awareness: the incumbent president Bronisław Komorowski, Janusz Korwin-Mikke, and Janusz Palikot. This did not occur in the case of Andrzej Duda or Paweł Kukiz. The candidate of PiS had not been recognizable for nearly half of the Polish electorate even six months before the election [CBOS 2014], while Kukiz was perceived through the image of a musician, not a politician.

RESULTS
In order to verify the hypotheses, we decided to calculate the mean values of the image qualities of all the candidates (included in the 13 scales of the semantic differential) for each respondent. Then, we calculated the mean value of those indices for the whole sample. This was done both for the results obtained in the pre-test and in the post-test. Then, we carried out a test to check the significance of differences between the obtained means, whose results are presented in Table 3. The obtained results confirm hypothesis 1, saying that: Women are more susceptible to changing the evaluation of a political image as a result of the influence of TV electoral ads. This is proved by the fact that exposure to TV spots in the group of women resulted in the image reconfiguration of 7 out of 11 candidates (in the case of men, this only referred to 5 candidates). In the case of women, there were differences concerning: Grzegorz Braun, Andrzej Duda, Adam Jarubas, Janusz Korwin-Mikke, Marian Kowalski, Paweł Tanajno and Jacek Wilk. What is more, the differences in the pre-test and post-test means (in the case of women) displayed a higher level of statistical significance. In the case of men, statistically significant differences between the pre-test and the post-test were diagnosed in the case of political images of the following candidates: Andrzej Duda, Adam Jarubas, Marian Kowalski, Paweł Tanajno and Jacek Wilk. In-depth analysis involving other variables used in the study showed that the greater susceptibility of women to a change in perception of the candidates' images under the influence of TV political advertising was related to the lower level of interest in politics displayed by this group of respondents. Women's evaluations of candidates' images in the pre-test and post-test were strongly positively correlated with the evaluations made by people declaring no or little interest in the sphere of politics (Rho = 0.912; p < 0.001). Such relationships were not found for political preferences or the level of political alienation.
The direction of influence of the presented electoral ads is also interesting. We found out that in the vast majority of cases the exposure to TV spots worsened the evaluation of the candidates' images. The only exception in this regard was Janusz Korwin-Mikke, whose image qualities were evaluated higher by women. This could have been the result of the strategy adopted by the candidate, also present in his electoral ads. In that presidential election campaign, Korwin-Mikke slightly softened his controversial image. In the spots he refrained from aggressive attacks on his political opponents, instead trying to persuade the viewers to participate in the election. 10 In the case of Korwin-Mikke, considerable professionalization in the production of the campaign spots was also evident. Long ads in the form of "talking heads" referring to complex economic issues, which he had used in the previous presidential elections, were replaced by short, more attractive messages, which -as evidenced by researchmanaged to improve a little the negative image of this candidate in the group of women.
It is also interesting that out of all candidates, only the image of Paweł Kukiz was evaluated positively in both groups (both in the pre-test and post-test). This confirms the tendency diagnosed in Poland to evaluate politicians negatively regardless of the parties or orientations they represent [CBOS 2017].
The decrease in the evaluation of image qualities of some candidates (such as Paweł Tanajno, Jacek Wilk, Grzegorz Braun, Marian Kowalski, Adam Jarubas) can be partly explained with their poor recognizability before broadcasting the spots. 11 Having watched the advertisements, the participants managed to obtain some information 10 Janusz Korwin-Mikke has been functioning in the Polish politics since the early 1960s. After the democratic transition in 1989, he ran for presidential elections four times. But he never received more than 2.5% support. Each time, his strategy was based on aggressive attacks on other politicians and controversial utterances, insulting considerable parts of the Polish electorate, i.a., the less educated, women, or people with disabilities. 11 The questionnaire also included the option "I don't know how to assess this", which was interpreted as a neutral rate of 4, in the middle of the semantic differential scale. The mean proportions of responses "I don't know how to assess this" for each candidate were: Paweł Tanajno (72.7%), Jacek Wilk (67.0%), Grzegorz Braun (54.6%), Marian Kowalski (53.6%), Adam Jarubas (52.9%), Andrzej Duda (20.2%), Magdalena Ogórek (18.0%), Paweł Kukiz (12.6%), Janusz Palikot (12.4%), Janusz Korwin-Mikke (10.9%), Bronisław Komorowski (5.9%). about those (mostly niche) candidates, which resulted in a negative reception of their image. In the context of the research, the case of Grzegorz Braun seems interesting, as his image evaluations in the post-test only changed significantly among women. Presumably, it was connected with the very poor recognizability of characteristics of that candidate in the case of women (much lower than in the case of men). 12 Many women probably first learnt about Braun from the presented spots, and they did not like the extremely conservative image of that candidate created in political ads. This could have partially resulted from the views diagnosed in the study sample: women more often than men declared left-wing orientations.
The fact that women are more susceptible to the influence of TV electoral advertisements is also evidenced by the changes in emotional attitude to the candidates before and after watching electoral spots, expressed on the scale of the feeling thermometer (Table 4). Also in this case, the TV spots more often (and to a greater degree) caused a change in feelings toward the politicians in women rather than in men. Just like in the case of semantic differential scales, women's scores in feeling thermometers displayed a higher level of statistical significance. Another issue interesting from the political studies point of view is the differences in perception of the candidates' images between men and women. Men evaluated more positively, both in the pre-test and in post-test, the image qualities of: Grzegorz Braun, Janusz Korwin-Mikke, Marian Kowalski, Jacek Wilk, and Magdalena Ogórek, whereas women evaluated better the incumbent president Bronisław Komorowski and Janusz Palikot. Similar statistically significant relationships were found in terms of emotional attitude. They probably result from the ideological differences between the two groups, mentioned before. The only result that did not fit this pattern was the higher evaluation of Magdalena Ogórek's image and the emotional attitude to her among men. This probably results from the fact that despite the SLD endorsement, she was perceived as an independent candidate with a non-specific ideological profile.
To show the relationships between the feelings thermometers and semantic differential results, we decided to calculate the correlation coefficients for these values, as well (Table 5). In all the cases we found a strong relationship between these indices, which was stronger for candidates who were more well-known. Interesting conclusions can be drawn from the analysis of correlation between these variables with reference to the respondents' gender. Actually, the values of correlation coefficients are always higher in the case of women than in the case of men. This shows that the evaluation of the candidates' image qualities in the case of women is less related to the emotional attitude to them. This phenomenon can be explained on two planes. first, it may mean that when evaluating politicians, women to a lesser degree rely on emotions and are able to evaluate the candidate's traits rationally, disregarding any liking or disliking they may have. Second, it may result from the diagnosed lower level of interest in politics among women, which may have led to the discrepancies between the emotional response to candidates and the evaluation of their images. The second hypothesis, assuming that TV electoral ads, regardless of the gender of their viewers, will not cause a reconfiguration of image qualities of common-ly known candidates, was basically confirmed, too. In the case of Bronisław Komorowski and Janusz Palikot, differences in the mean values in semantic differentials and feeling thermometers did not display any statistical significance. Only the evaluation of Janusz Korwin-Mikke was somewhat different, as he was better assessed by women and had a higher level of liking after the exposure to TV spots rather than in the pre-test (although in his case, this was rather a lower level of disliking rather than a higher level of liking). In this case, the spots in a way disrupted the image of politician who hates women, 13 deeply rooted in the public awareness, thus being an effective tool of reconfiguration of image qualities.

CONCLUSIONS
The object of the presented study was the influence of TV electoral advertising on a change in the images of the candidates running for the 2015 presidential election in Poland. Our empirical research suits the trend of studies of the influence of TV electoral ads on the evaluation of candidates' image and the emotional attitude to them. Using methods commonly applied in this kind of analysis, we diagnosed women's higher susceptibility to the influence of such communication tools, as well as lower effectiveness of TV spots in the process of reconfiguration of image perception of well-known politicians whose image is consolidated in the public awareness. In-depth analysis shows that the higher susceptibility of women to advertising messages revealed in the study was connected with a lower level of interest in politics than that displayed by men.
The presented analysis had some limitations that may have affected the study results. first, there may be some doubts regarding the study sample, made up of students. Such samples may be characterized by a high level of homogeneity, resulting not only from the age or education level but also from the similarity of psychological qualities. However, we tried to minimize this factor, including students of different fields: social, humanistic and medical sciences. Furthermore, other studies of this kind, including most studies mentioned in the theoretical part of the paper, are also successfully carried out in student groups. This also gives the opportunity to compare the obtained results with the analyses performed in the USA and Western Europe. Definitely, in order to verify the identified relationships, it would be worth carrying out a similar study on a more varied group, involving respondents from other age groups and with different socio-demographic characteristics. An additional limitation of the study is that the sample was not randomized, which surely limits the possibility of generalizing the obtained results to the whole population. However, other studies presented in the theoretical section are carried out in a similar mode. Unlike in those studies, we decided to use a much more numerous sample, which in a way strengthens the identified relationships. But in this context, it is worth doing an experiment with a control group, so as to be sure that the change between the pre-test and post-test resulted from the exposure to the advertisements, not from other events that occurred in the interim between stage 1 and stage 3 of the study.
Second, because we used real advertisements of election committees of the candidates running the presidential election, the participants may have obviously had seen them before. This factor may have potentially weakened the influence of TV electoral ads. To diagnose the potential scale of this phenomenon, in the post-test we included questions concerning familiarity with the presented spots. Most of the participants declared they did not know the ads or had had little exposure to them. The only exception was Bronisław Komorowski, whose ads had been watched by the respondents more often. Yet, we may assume the image of the current president was so consolidated in the voters' awareness that this did not have a significant impact on the results. Besides, this limitation refers to all studies carried out with the use of real TV electoral spots. An ideal but virtually impossible situation would be to perform such studies before the broadcasting of spots during the electoral campaign.
Third, the results may have been affected by the already mentioned specificity of the analyzed election, in which even the leading political groups decided to put up "secondary" candidates (or even, in the case of SLD and Magdalena Ogórek, not a party member). Not knowing the image qualities of candidates (actually, not knowing the very candidates) running for the election definitely caused the worsening of their evaluations after watching the electoral spots. In this case, the exposure to TV ads let people really get familiar with the candidates and evaluate them on the basis of the first, apparently not very positive, impression. In further studies it might be good to include a way to control the special campaign conditions, so as to be sure about the correct interpretation of the obtained results.
finally, an interesting direction of research is to test what contents included in campaign spots make them more persuasive, also with reference to the viewers' gender. In this case, due to the problems with clear classification of advertising messages pointed out by researchers, both in the dimensions of image ads vs issue ads 14 and negative ads vs positive ads, 15 it seems to be extremely difficult to carry out studies based on actually broadcast TV spots.
14 Frank Biocca [1991: 18] approaches very critically the analyses classifying spots as issue ads or image ads. In his opinion, only few spots can be doubtlessly classified as one or the other. The division of ads into single frames or verbal utterances shows that most of them include both the elements of issue and image ads. Therefore, we need to approach with caution the analyses that unidimensionally classify spots as issue or image ads, especially that the content of issue ads also naturally affects the candidate's image. 15 Categorizing advertisements as positive or negative is often a kind of simplification. As Glenn W. Richardson [2001] observes, researchers often approach the negative aspect of the message too narrowly, only focusing on the verbal component of the ad and ignoring the audio-visual elements.