Attitudes toward Specific Advertising Media ( AM ) : Informative or Manipulative ?

Attitudes toward advertising in general (AG) is a concept which often raises ethical, social and cultural concerns, and yet in the context of specific media, consumers are often more positive, and find advertising entertaining, thought-provoking, amusing and informative. This paper focuses on attitudes toward advertising in three specific media (AM) (print, television and online advertising) and aims to provide input into the design decisions of an advertising campaign. The survey is based on a sample of 425 Malaysian and international tourists, and one key finding is the positive view that respondents have about advertising in print, online and on television. In terms of attitudes to print advertising, social role and image makes the strongest unique contribution to AP; for television advertising product information makes the strongest unique contribution to explaining attitudes to ATV; and in terms of online advertising, hedonic pleasure makes the strongest unique contribution to explaining attitudes to AO.


Introduction
Advertising can stimulate amusement, sadness, laughter and pathos.Advertising also reflects how people see themselves: who they are and who they identify with, and can penetrate and reflect status, economically and socially.Images shown in advertising provide prototypical expectations about consumption patterns, characteristics of consumers, young or old, male or female, and blue collar or professional (Hirschman & Thompson, 1997).Advertising is able to promote positive values, behaviors and attitudes such as sociability, affection, generosity, patriotism, ecumenism, personal enrichment, and security (Holbrook, 1987).Advertising spend is also positively linked with economic growth (Kopf et al., 2011).
In contrast, advertising in general is viewed as a discipline and practice which has negative social and cultural repercussions (Pollay & Mittal, 1993;Sandage & Leckenby, 1980) and is criticized for unintended consequences which relate to its emphasis on a range of negative behaviors including materialism, cynicism, irrationality, selfishness, anxiety, social competitiveness, sexual preoccupation, powerlessness and a loss of self-respect (Pollay, 1986;Pollay, 1987).Advertising faces a barrage of criticisms (Kopf et al., 2011) much of which can be placed into one of two broad categories: first, advertising is blamed for being inherently wasteful by inflating the prices paid for goods and services; and second, advertisers are accused of creatively promoting perceived obsolescence and imbuing products with self-worth, freedom, adventure, and success (Dauvergne, 2010).The question of ethical values in advertising practices is also one of the controversial issues driving negativity towards advertising, especially in relation to advertisements on the television.Despite the strict guidelines in many countries designed to protect children from negative advertising influences and to ensure that advertisements directed at them are ethically and morally acceptable, parents and other consumer protection groups feel that television advertising causes harm to certain audiences, especially young and vulnerable viewers (Bandyopadhyay, 2001).publishing to broad masses of people' (Ünsal, 1982 p.12 cited by Kes, 2011) and this definition includes e-advertisements, which seek to generate sales via the Internet (Kes, 2011).Many consumers also enjoy advertisements, and indeed find advertising entertaining, thought-provoking, amusing and informative, but which elements of separate advertising media best predict attitudes to advertising in general?Do these contradicting opinions continue to reflect people's general attitudes towards advertising, even in the context of specific media?Are consumers more positive or more negative about advertising especially since the rapid developments in online advertising?Previous literature on attitudes toward advertising in general (A G ) tends to focus on generic advertising and rarely focuses on attitudes in terms of particular media, such as print, television, or online media and the Internet.In this article the authors develop and test a predictive model relating to attitudes to advertising in general (A G ), by identifying and comparing the factors affecting consumers' attitudes towards advertising in more specific media, and measure the predictability of responses of advertising in the overall explanation of print advertising (A P ), advertising on television (A TV ) and online advertising (A O ).

Attitudes to Advertising in General (A G )
Advertising in general (A G ) is a learned predisposition to respond in a consistently favorable (or unfavorable) manner to advertising (Lutz, 1985).Research has also (e.g.Sandage & Leckenby, 1980;Muehling, 1987) measured and tested A G in relation to other variables, in addition to measuring the dimensions that comprise the construct.Bauer and Greyser (1968) also examine two effects of advertising in an attempt to predict beliefs about advertising using eight items in two categories: economically-and socially-oriented statements.
Based on the results using this measurement scale, Americans are in agreement that the functions of advertising help them to get better products and act as a good instrument for the economic system.However, Americans show a significant dislike toward individual advertisements because they find them offensive and annoying (Bauer & Greyser, 1968).Later studies include consumers' attitudes towards certain attributes of advertisements, and the findings aim to predict advertising effectiveness, brand attitudes and purchase intention.For instance, (Biehal et al., 1992) examine the relationship between customers' attitudes toward the advertisement (A Ad ) and its effect on brand choice; (Olney et al., 1991) investigate advertisement content which arouses emotion; (Muehling & McCann, 1993) examine attitude toward specific advertisements (A Ad ); and the effect of an advertisement's claim and content on attitude toward the advertisement, is the subject of a study by (Burton & Lichtenstein, 1988).Studies on advertisements that are specifically targeted at certain audiences such as children (Mallalieu et al., 2005) or specific ethnic groups (De Run, 2007) and special target groups (Orth & Koenig, 2007) raise ethical concerns in some cases.
Attitude toward advertising in general (A G ) is an important element affecting attitude toward advertisements (A Ad ) which, in turn influences purchase behavior (Durvasula et al., 1999).The applicability of the A G framework outside the United States is the focus of work by Durvasula et al. (1999) and research by Pollay and Mittal (1993), which extends the model.A more recent A G model charts the relationship between determinants and antecedents of advertising attitudes in the context of European Union accession countries, namely, Bulgaria and Romania (Petrovici & Marinov, 2007).The focus of this research is on A G but also specific advertising media.

Attitudes towards Advertising in Specific Media (A M )
Researchers examining attitudes towards A G (Sandage & Leckenby, 1980;Muehling, 1987;Andrews, Lysonski, & Durvasula, 1991;Ramaprasad, 2001;Petrovici & Marinov, 2007) claim to measure attitude toward advertising as a whole -both as an instrument and as an institution -regardless of the medium used and without providing respondents with specific examples of advertisements.Some studies look in particular at the relationship between attitudes towards advertising in some specific advertising media and the subsequent effects on viewers, although such studies pre-date online or internet advertising.James and Kover (1992) measure respondents' attitudes toward advertising in relation to their level of involvement with specific advertisements on the television and in printed media, and the research is one of the few studies to focus on specific media.Advertisements in print media, such as newspapers or magazines have advantages: reading a printed advertisement largely relies on a reader's personal choice, whereas experiencing television advertising is more difficult to avoid as part of a television experience (James & Kover, 1992).Conventional print advertising is also superior to television with respect to memory measures -specifically advertising recall (Sundar et al., 1998).In print advertising, respondents with more favorable attitudes towards advertising recall a higher number of advertisements after exposure (Schwaiger et al., 2010).
Television advertisements have the capacity for more emotional content and television advertisements also offer greater opportunity to entertain compared with print advertisements which have greater capacity for providing information (Resnik & Stern, 1977).Nonetheless, the television medium has more potential to be manipulative and exploitative because television can more easily reflect reality and the unimaginable -beyond reality.This practice is well-established in television commercials directed at children where fantasy is used widely on the basis that children have a great capacity for imagination (Bandyopadhyay et al., 2001) Online advertising is a well-established as a medium (Hawkins 1994) which is characterized by ease of entry, relatively low set-up costs, global accessibility, time independence, and interactivity (Berthon et al., 1996).Consequently, this medium creates an opportunity for firms to deliver fast and timely messages at a low cost.Electronic advertising is essential for the majority of businesses and the number of users connected to the Internet is increasing exponentially every year (Kes, 2011).The nature of electronic advertising with high-speed interactive capability is one of the factors contributing to the number of users.Advertising content is one of the biggest segments on the Internet resulting in a total of 21.2 billion dollars for the full year of 2007 in the United States alone (IAB, 2012).Online advertising, electronic advertising or web advertising has seldom been used in past studies on the topic of advertising in general (A G ). Previous studies on attitudes towards online adverting have mainly focused on the acceptance level of consumers as the users of the information technology, and such research is becoming increasingly out-dated as the Internet is more widely used.Despite the growing number of studies on technology acceptance, studies on the perceptions of consumers, policy-makers or advertisers on perceptions of online advertising remain scarce.
A study exploring how Web managers perceive the effectiveness of a web site compared with eight other traditional media (Leong et al., 1998) used ten key media attributes to measure the gaps between the media in a two-dimensional perceptual map.In 1998 web managers perceived a web site as comparable to direct mail media as a source for delivering rich information, but were unable to see the potential for using emotion in online advertising.Owing to the rapid development of the Internet, including such websites as YouTube, these findings are somewhat obsolete, and all forms of advertising are now available online via the Web in an integrated platform.

Factors Associated with Attitudes towards Advertising in Specific Media
This research study adopts indicators used by Pollay and Mittal (1993), Petrovic and Marinov (2007) and Tan and Chia (2007) to measure the correlation between personal experience and use of advertising and attitude toward advertising in specific media.Therefore, the research framework has multiple indicators of personal experience and use of advertising: (i) product information, (ii) social role/image, (iii) hedonic/pleasure and (iv) falsity (the potential to distort, mislead, misinform and exaggerate) Firstly, product information is an important justification of advertising in previous studies (Pollay & Mittal, 1993;Petrovici & Marinov, 2007).The product information dimension measures the individual's beliefs about advertising as a valuable source of information for and about products or services.In this study the researchers measure product information using the following items: 'advertisements help me to keep up to date about products/services available in the marketplace'; 'advertisements are a useful source of information'; ' advertising is a valuable source of information about local sales/products'; 'advertising contributes to my knowledge about quality products'; and 'advertisements tell me which brands have the features I am looking for' (Petrovici & Marinov, 2007); and better value can be achieved from products advertised in the media compared with products that arenot advertised' (Sohail & Saeed, 2004), which measures whether or not advertising in any specific media adds value to the advertised product.
Print advertising is a good source of product information, and many previous studies support this view (Tan & Chia, 2007;La Ferle et al., 2000).Leong et al. (1998) carried out a cluster analysis and produced a dendrogram showing that attributes such as creating brand, product or corporate awareness; communicating product, brand image and corporate image, are the best way to characterize the press (or print advertising media).Even adolescents aged 14 to 19 prefer to read newspapers to find out about news and current events (La Ferle et al., 2000).H1: Consumers' attitudes towards print advertising (A P ) predict beliefs about advertising in printed media as a source of product information.
A perceptual map by Leong et al., (1998) compares the effectiveness of online advertisements with other traditional media and online advertisements are placed closer to the rational dimension vis-à-vis the emotional dimension.This finding is similar to previous research showing that online advertising has the ability to deliver rich information and detail, which is also in line with other traditional media such as print media (Schlosser & Shavitt, 1999;Ducoffe, 1996;Brackett & Carr, 2001;Lei, 2000).Wolin et al. (2002) also argue that the product information role has a dominant role in web advertising.H2: Consumers' attitudes towards online advertising (A O ) predict beliefs about online advertising as a source of product information.
Secondly, the social role and image dimension describes an individual's beliefs about advertising that reflect and shape personal image.A study of advertising attitudes among consumers in Romania shows that a majority of the respondents think advertising is fulfilling a social role (Petrovici & Marinov, 2007).Social role and image (Petrovici & Marinov, 2007) includes the following items: 'advertisements tell me what people like me are buying or using'; 'advertising has an important role in the education of children'; and 'advertising keeps me up to date with trends in various fields'.Another four items are added to this measure, previously used by (Pollay & Mittal, 1993), that is: 'advertising helps me to know which products will or will not reflect the sort of person I am'; 'sex is used widely in advertising'; 'from advertising I can learn about fashions to buy to impress others; and 'advertising distorts the values of youth'.These additional items are important for framing respondents' encounters with social images relating to sex and youth in advertising and advertisements.
H3: Consumers' attitudes towards television advertising (A TV ) predict beliefs about advertising on the television as a medium of building social role and image.
Thirdly, hedonic pleasure is a dimension that encompasses an individual's beliefs about advertising in terms entertainment, amusement and pleasure.The following statements measure this construct: 'sometimes advertisements are even more enjoyable than other media content'; 'advertising is often amusing and entertaining'; 'I like to talk to my friends about advertisements'; 'sometimes I take pleasure in thinking about what I saw, heard or read in advertisements'; and 'sometimes advertisements bring to my mind pleasant memories' (Petrovici & Marinov, 2007).
Unlike the print media, the television medium has an entertainment function and has the ability to evoke feelings through the use of motion, color, sound and special effects (Leong et al., 1998).Television advertising has the ability to arouse emotion and to entertain viewers, which is quite distinct from the other advertising media and influences people's overall attitudes towards advertising (Shavitt et al., 1998).For example, one of the functions of television which is highly inter-correlated with people's attitudes towards television advertising is its entertainment function: people find television advertisements fun to watch (Alwitt & Prabhaker, 1992).Research by Tan and Chia (2007) also supports the notion that the hedonic/pleasure function of advertising is positive and significant in predicting attitudes towards television advertising.H4: Consumers' attitudes towards television advertising (A TV ) are predicted by their positive beliefs about advertising on television as a source of fun entertainment and pleasure-seeking material.
Fourthly, falsity is a dimension which explains an individual's beliefs about advertising as a source of manipulation, exaggeration or misleading information.This scale is developed based on the items based on research by (Larkin, 1977) and (Tan & Chia, 2007).The items are 'in general advertising in misleading'; 'advertising should be more realistic'; 'most advertising insults the intelligence of the average consumer'; 'in general, advertisements present a true picture of the product advertised'; 'advertising often persuades people to buy things they really don't need or should not buy'; and 'there are too many exaggerations in advertising'.Some people find television advertising intrusive, especially those who anticipate enjoying television programmes (Alwitt & Prabhaker, 1994).This is intensified when advertisements are seen as manipulative and misleading.For instance, Alwitt and Prabhaker (1992) report that the majority of their respondents agreed that the advertised products do not perform as well as claimed in the television advertisements and that television advertisements do not present a true picture of the product being advertised.Consequently, these findings, and the prevalence of advertising credibility in other advertising attitude models, results in Brackett and Carr (2001) including A TV as an additional perceptual antecedent to advertising value.
Advertising credibility represents consumers' perceptions of the truthfulness and believability of advertising in general (MacKenzie & Lutz, 1989).Indeed, this additional perceptual antecedent together with the existing three (i.e.informative, entertainment and irritation) have a significant relationship with advertising value, which in turn is significant in predicting attitude toward advertising (Brackett & Carr, 2001).This exemplifies respondents' negativity towards television advertising due to its perceived capacity to irritate and manipulate (James & Kover, 1992).H5: Consumers attitude toward television advertising (A TV ) is predicted by negative beliefs about advertising on television as a source of misleading and manipulative information.H6: Consumers attitude toward advertising in general (A G ) is predicted by positive attitudes towards advertising in print, on television and in online advertising (A M ).

Scaling of Items
This research aims to measure attitude towards advertising in specific media (A M ) in addition to measuring A G .Tan and Chia (2007) also measure this construct with a set of Likert-scaled items, for example 'Overall, I consider television advertising to be a bad thing'; 'Overall, I like television advertising' and 'I consider television advertising to be very essential'.This study uses semantic differential scale items in order to be consistent with the items used by (Tan & Chia, 2007) therefore, three semantic differential pairs of adjectives are utilized in this study.The adjectives are good/bad, negative/positive and essential/not essential.

Research Instrument
The questionnaire design uses same-respondent replies for both predictor and criterion variables: that is, the respondents are asked to provide answers to the same questions and items for each separate media -print (A P ), television (A TV ) and online advertising (A O ).Common method variance (CMV) is a potential problem with this approach, that is, variance that is attributable to the measurement method, rather than to the constructs the measures represent (Podsakoff et al., 2003).In order to reduce the problem or minimize the likelihood of CMV, Podsakoff et al. (2003) suggest four remedies, one of which the current authors use as a solution for this potential problem.In this research, the remedy is implemented at the ex ante research design stage: the authors use different scale endpoints and formats in addition to some reversal of negative/positive (for both predictors and criterion variables) to minimize the mesmerizing effect of the simple repeat.
The first author conducted a pilot study in the United Kingdom over a period of two weeks to test the reliability of the questionnaire items and to further refine them.Following the final amendments to the questionnaire, data collection took place in Malaysia at the international airport to secure respondents from wide range of nationalities.The researchers used mall-intercept techniques (de Brewer & Haydam, 1997;Sudman, 2000) and removed incomplete responses, which resulted in a total of 425 questionnaires for analysis.The amount of data collected corresponds to the 1:5 item-cases ratio (Tabachnik & Fidell, 2007) and is large enough to have a negligible impact on the expected non-normality of individual metric variables (Hair et al., 1992)

Reliability Assessment of the Advertising Beliefs Variables
Researchers tested all dimensions of advertising beliefs (i.e.product information, social role/image, hedonic/pleasure and falsity) using Cronbach's Alpha coefficients, corrected item-total correlations, squared multiple correlations and item means across three different advertising media.All values of Cronbach's alpha coefficient exceed the minimum recommended threshold (α ≥ 0.70), suggesting that the scales are reliable and stable (Churchill 1979).More specifically, the internal consistency of product information is the highest, with the Cronbach's alpha coefficients ranging from 0.78 to 0.82.For social role/image, the Cronbach's alpha coefficients range from 0.76 to 0.78.For the hedonic/pleasure dimension, the Cronbach's alpha coefficients range from 0.75 to 0.78.Lastly, falsity records Cronbach's alpha coefficients from 0.72 to 0.75.Furthermore, analysis of the corrected item-total correlation coefficients also shows that each item makes a considerable contribution to the total scores of the four respective dimensions.The corrected item-total correlation coefficient for product information ranged from 0.47 to 0.67, social role/image from 0.45 to 0.66, hedonic/pleasure from 0.44 to 0.58 and, for falsity the correlation coefficient is from 0.33 to 0.65.Squared multiple correlations are as follows: for product information, the values are from 0.23 to 0.56; for social role/image, from 0.24 to 0.44; for hedonic/pleasure, from 0.23 to 0.34; and for falsity from 0.13 to 0.46.Therefore, these results indicate that all items can be accepted for inclusion in their respective summated scales and the scales can be treated as indicators for advertising belief variables for further analyses.

Research Findings and Hypotheses Testing
Attitudes toward specific advertising media (i.e.A P , A TV , A O ) are measured by three items each, on three semantic differential adjectival pairs (i.e.bad/good, negative/positive and not essential and essential).Table 1 reports the mean scores, standard deviation (SD), skewness and kurtosis values of advertising attitudinal scales.The respondents report different degrees of intensity in terms of their attitudes toward these advertising media.For instance, overall attitude toward internet advertising (A O ) is least favorable with a mean score of 4.71 compared with A P and A TV (reporting mean scores of 5.05 and 5.03 respectively).All items measuring A O record the lowest mean scores compared with the other two advertising media (ranging from 4.63 to 4.80 with standard deviation ranging from 1.41 to 1.49).The highest mean score is A TV (bad/good) with a mean score of 5.15 and standard deviation of 1.38.In contrast, A O has the lowest mean score for one item (i.e.negative/positive: 4.63) and a standard deviation of 1.45.A series of statistical analyses i.e. correlation analysis, reliability test and validity test are carried out to test the hypotheses using Statistical Package for the Social Sciences (SPSS 16.0).The researchers then group beliefs about advertising into four dimensions related to experience with advertising (i.e.product information, social role/image, hedonic/pleasure and falsity).The testing uses a total of 22 items on a 5-point Likert scale.
The researchers conducted Pearson Correlation test to gain greater understanding the relationships between the impendent variables and A P , A TV and A O .The falsity scale is omitted from the regression analysis due to its weak correlation with A G (r = -0.09.All assumptions of normality, linearity, multicollinearity and homoscedasticity appear not to have been violated, based on results from preliminary analyses.The multicollinearity effects show tolerance values ranging from 0.41 to 0.81; variance inflation factor (VIF) values ranged from 1.22 to 2.44 (smaller than 10).

Attitude towards Advertising on Printed Media (A P )
The regression analysis results reveal significant associations between personal uses of print advertising and A P (R 2 = 0.17, F (3, 421) = 28.69,p = 0.00) (see Table 2).The three personal experience and use of advertising measures were analyzed, however, only two variables were statistically significant in explaining A P , with Social Role/Image reporting the highest beta value (β= 0.209, p < 0.001) followed by Hedonic/Pleasure (β= 0.145, p < 0.001).
Although the results show a positive association between Product Information and A P (i.e.high levels of product information associated with high levels of attitude toward print advertising), the regression analysis provides no conclusive evidence to support H1. Regression analysis also measures the effect of personal experience and uses of advertising on attitude towards television advertising (A TV ).In this model, all scales measuring beliefs about advertising are based on respondents' experiences of television advertising.Similar to the previous analysis, falsity was excluded due to its weak relationship with the dependent measureA G (r = -0.15).Therefore, H5 is not supported.
Prior to the analysis the researchers checked the tolerance and variance inflation factor (VIF) values to identify the effect of multicollinearity.The values are within the acceptable range (tolerance values ranged from 0.39 to 0.79 and VIF ranged from 1.26 to 2.59), thus confirming no violation of multicollinearity.Table 3 shows that the model explains 21 percent of total variance of A TV .Two out of the three personal experience and use of advertising factors of television advertising; are significant (product information, β = 0.229; social role/image, β = 0.210).* significant at the p < 0.001 In the initial stage, a correlation coefficient reveals the relationship between social role/image and attitude toward television advertising and the result shows a significant association (r = 0.42, n = 425, p < 0.01), with high levels of social role/image associated with high levels of attitude toward television advertising.In this regression analysis, the positive coefficient confirms the predictability of Social Role/Image on A TV .Therefore, H3 is accepted.However, the finding does not provide substantial evidence to support H4.The relationship between hedonic/pleasure and attitude toward television advertising is positive, however, the strength of the relationship is small (r = 0.34, n = 425, p < 0.01) and does not substantially corroborate the hypothesis, although predictive validity measure is sufficiently provided.

Attitude towards Advertising on the Internet (A O )
Finally, the researcher use regression modeling to assess the predictability of personal uses of advertising on consumers' attitudes towards online advertising and Table 4 shows the results of the analysis.Prior to the analysis, the authors checked the tolerance and variance inflation factor (VIF) values to detect any effect of multicollinearity.Similar to the previous analysis, falsity is excluded due to its weak relationship with the dependent measure A G (r = -0.07).* significant at the p < 0.001 In Table 4, all personal uses of advertising are significant in predicting A O .In the regression model, the independent variables also indicate significant association with the mediator (R 2 = 0.25, F (3, 421) = 47.69,p = 0.00).Hedonic/Pleasure reporting the highest beta value (β = 0.240, p < 0.001) followed by Social Role/Image (β = 0.196, p < 0.001) and Product Information (β = 0.130).Correlation analysis also reveals that the association between product information and attitude toward online advertising was significant and the strength is distinctly moderate (r = 0.42, n = 425, p < 0.01).This signifies that high levels of product information are associated with high levels of attitude toward television advertising.At the same time, the correlation provides predictive validity of product information on attitude toward internet advertising.With this evidence, H2 is supported.

Post-hoc Analysis
The findings from this research suggest that attitude toward specific advertising media (A M ) is related to advertising in general (A G ).This likelihood of this relationship is also supported by Baron and Kenny (1986).Baron and Kenny (1986) suggest that a set of three prerequisites be met to establish the mediational model.Prior to the analysis, the fundamental assumption of multicollinearity was checked and no evidence to show violation of the assumption was detected.
Table 5 shows the results of the regression analyses.First, all personal use factors (i.e. the independent variables) are entered into a regression model with the dependent variable: A G .The results reveal a significant relationship between personal experience and use of advertising and A G (R 2 = 0.23, F (3, 421) = 42.57,p = 0.00).All three personal experience and use of advertising items are significant (product information, beta = 0.17; social role/image, β = 0.19; hedonic/pleasure, β = 0.18).In this second model (Model 2), the independent variables also show significant association with the mediator (R 2 = 0.29, F (3, 421) = 55.94,p = 0.00).The final step (Model 3) confirms the effect of A M as a mediator (R 2 = 0.44, F (4, 420) = 80.87, p = 0.00).The beta value for A M was the strongest (β = 0.53, p = 0.00) followed by hedonic/pleasure (β = 0.11, p = 0.05).These results show that A M significantly predicts A G , therefore H6 is accepted.** significant at the p < 0.001; * significant at the p < 0.05 In Model 3 product information and social role/image have no effect (see Baron and Kenny, 1986).With the reduced power revealed in the third regression model, the results confirm that A M provides a mediating pathway in the relationship between product information → A G and social role/image → A G .However, since the beta coefficient for hedonic/pleasure in Model 1 holds in Model 3, A M was established as a partial mediator in the relationship between hedonic/pleasure and A G . Figure 1 illustrates the effect of A M in the relationship between personal experience and uses of advertising and A G and shows the correlations for all the key variables.

Discussion and Conclusion
First, one of the key issues in A G is the fundamental view consumers have about advertising in terms of its value and potential to provide information and enjoyment, and the potential to convey social roles and images to a large audience -but also the potential for advertising to mislead, to be unrealistic, insult the audience's intelligence, distort information, exaggerate the positive attributes, and to persuade people to buy things they do not need (falsity).A key finding from this research is the positive view that respondents have about advertising in print, online and on television.The scores for falsity are low and the correlations for falsity in terms of all three media show that in respect of these items -items related to the view that advertising is misleading, exaggerated and dishonest, have r values so low that this variable can be excluded from the regression analysis.Respondents are therefore very positive about A G and A M , and support the role of advertising in providing information, and offering fun and pleasure.
Second, the results show that the independent variable that contributes uniquely to attitudes toward each specific media is different for A P , A TV and A O .In terms of print advertising, social role and image (make the strongest unique contribution to explaining attitudes to A P .Results for television advertising show that product information (makes the strongest unique contribution to explaining attitudes to A TV ; and in terms of online advertising hedonic pleasure (makes the strongest unique contribution to explaining attitudes to A O. Third, on a macro level A G is significantly predicted by respondents' attitudes towards advertising in specific media (A M ).A M emerged as the strongest predictor of A G -a combination of attitudes to all the three types of advertising predict attitudes to advertising in general.The variable with the strongest effect on A G is Ap: the results of regression analysis show that A G is best predicted from the print advertising standpoint.The R 2 value is the highest among the three models (R 2 = 0.489), suggesting that all of the independent variables are able to explain 48.9 per cent of the variability in A G .Post-hoc analysis confirms that A P fully mediates the relationship between hedonic/pleasure and A G , and partially mediates the relationship between social role/image and A G .
Fourth, A G can also be predicted from the television advertising perspective and the correlation is highest (r= .58).This is due to the explanatory power of the model, in which the R 2 value is 0.486, suggesting that the independent variables are able to explain 48.6 percent of the variation in A G .Although hedonic/pleasure and product information have a direct relationship with A G , the post-hoc analysis shows that in this model A O reduces the predictability of both variables, and acts as a partial mediator.Attitude toward online advertising (A O ) is not a prime contributor to A G , however the effect is positive and significant.The social role/image → A O → A G pathway is not confirmed and is therefore eliminated.Further research is necessary to explore and test this hypothesis.
Fifth, advertising functions are at their best in the context of a creative combination of motion, color, sound and special effects (Bland et al., 2005) together with the potential to use humor.Thus both television and online advertising are capable of delivering the complex messages that comprise product information but also offer pleasure and enjoyment which are more difficult to capture in the other formats, particularly print.
Sixth, the rapid expansion in online advertising and related promotion via the Internet seems to have almost replaced traditional forms of advertising.However, this scenario is not fully supported in the results of the present study.The results show overall, that the respondents view online advertising more negatively than print and television perhaps because online advertising can also be intrusive and unwelcome.Authors of previous studies frequently agree that online advertising is more likely to be realistic and avoids exaggeration.Nonetheless, findings in this study indicate that attitude toward online advertising (A O ) is influenced by all of the factors: hedonic/pleasure, social role/image as well as product information but the unique contribution to explaining attitudes to A O is hedonic pleasure and shows that online advertising is perhaps enjoyed more than either A P or A TV .Online media are increasingly becoming a hybrid of television, radio, newspapers, magazines, billboards, direct mail and so forth (Miller, 1996).Hence, the Internet's ability to deliver all the functions: product information, social images and hedonic pleasure.The results suggest that to optimize online advertising designers could make the best use of multimedia capability and need to explore ways of attracting the viewers' attention by integrating product information with interactive techniques to make advertisements more appealing and memorable.
Finally, the results indicate therefore that print advertising and television advertising are still perceived as strong in terms of providing product information, and conveying social role and image effectively.One of the more recent developments in online advertising over the past five years, however, is the integration of all the three media: print, online advertising and TV or movies through websites such as YouTube (Mangold & Faulds, 2009) in conjunction with company websites.All three kinds of advertisements are now feasible online as the technology becomes increasingly more accessible and affordable.All it needs is the continuing and ever-creative imaginations of the designers.

Limitations of the Study
The authors acknowledge a number of limitations with this research study.Firstly, the scope of the research was limited to examining attitudes towards advertising from three different media frames of reference, i.e. print, television and online advertising.However, with rapid advancement in mass media technology, particularly on the Internet, means that the concept of online advertising is changing on an ongoing basis.The focus of the study is online advertising but respondents may vary considerably in what they perceive online advertising to consist of.This is also true in terms of print advertising and television advertising which is not a static entity.However, this is also the case for previous research in the field of A G .
Secondly, the sample is a non-probability sample, and in this respect the researchers acknowledge the limitations in terms of generalisability.The first author collected the data using a mall-intercept technique which also has implications in terms of generalization of the findings for other populations.Although statistical tests demonstrate homogeneity of the sample across different demographic backgrounds, the authors are careful when interpreting the results because advertising attitude patterns of other consumers from diverse interest backgrounds might be different -example the sample respondents are predominantly from eastern Asia, and a sample from a western population might produce different results.

b
Items measured on a semantic differential scale with 1 = negative and 7 = positive c Items measured on a semantic differential scale with 1 = not essential and 7 = essential a All correlations are significant at the 0.01 level (2-tailed) unless otherwise stated; b correlation is not significant

Figure 1 .
Figure 1.Relationships between personal beliefs and use of advertising, A M and A G

Table 2 .
Regression testing the relationship between personal use of advertising and A P Attitude towards advertising on printed media (A P ) * significant at the p < 0.0018.2Attitude towards Advertising on Printed Media (A TV )

Table 3 .
Regression testing the relationship between personal uses of advertising and A TV

Table 4 .
Regression testing the relationship between personal uses of advertising and A O

Table 5 .
Regression testing the mediating effect of A M on the relationship between personal experience and use of advertising and A G