The Impact of Online Advertising on Tunisian Consumers ’ Purchase Intention

The impact of online advertising on consumer behavior has been little studied and research on the topic reached mixed results. Our study focuses on the determinants of online advertising and the role of the mediating variable “attitude towards online advertising” in the relationship between perceived advertising value and consumer purchase intent. The results show that online advertising value depends positively on its informativeness, credibility and entertainment value. They also indicate that credibility and online advertising value affect consumers’ attitude towards this type of advertising. Finally, we show that attitude plays a mediating role between perceived advertising value and purchase intent.


Introduction
Today, Facebook has become one of the most solicited means of entertainment, a marketing tool and an advertising channel to which consumers have direct and interactive access.This led advertisers to integrate this channel into their marketing strategy.Thus, ads that are channeled through traditional advertising media are now being run on Facebook.Assessment of online advertising value is a key element in shaping consumer attitudes towards it.Perceived value and the resulting attitude, in turn, have an impact on the purchasing behavior.
Our study puts an emphasis on the determinants of perceived value and on the causal relationships between value, attitude and purchase intent.Our interest in online advertising is justified by the fact that this field is not fully explored.In addition to the mixed results of previous research, research on the factors determining online advertising value and its impact on attitudes and purchase intent were little studied in the marketing literature.Therefore, our aim in this study is to find out the determinants of value, to assess the measurement scales of variables and investigate the mediating effect of attitude in the relationship value-purchase intent.__________________________________________________________________________________________________________________ Our paper is organized as follows: we first develop our conceptual framework and the hypotheses of our study.Then, we present the methodology adopted.Next, we report and analyze the obtained results.Finally, we present the implications and limitations of our study and we conclude with future research.

Conceptual framework and research hypotheses
Our study focuses on the direct and indirect impact of perceived value of online advertising on purchase intent.The study of the relationship between perceived value of online advertising, on the one hand, and attitude towards it and purchase intent, on the other, was the focus of several authors, including Wang et al. (2002Wang et al. ( , 2009)); Ducoffe (1995Ducoffe ( ,1996)); Brackett and Carr, (2001); Kim and Hun (2014); Mehta (2000).Inspired by this line of this research, we extended their proposals in view of validating a model integrating the determinants of perceived value of online advertising and attitude towards online advertising and purchase intent.Online advertising value is influenced by the variables entertainment, credibility and informativeness, and in turn it acts on purchase intent, directly and indirectly, through attitude, which then acts as a mediating variable.

Perceived value and its effects
Value is defined as the relative preference of an interactive experience between a subject and an object (Holbrook, 1999;Aurier, Evard and N'Goala, 2001;Woodruff, 1997).The measurement scales of this concept, developed in the marketing literature, report to different approaches and vary from one field of research to another.In this study, we opted for the scale proposed by Ducoffe (1995).However, perceived value is itself a result of informativeness, credibility and entertainment provided by online advertising, (Wang et al., 2002;Ducoffe, 1995and 1996, Brackett and Carr, 2001).

Informativeness
According to Wang et al. (2002), information is valued by its relevance.Advertising that provides information relevant to the needs may be actively processed and result in a response.For consumers, advertising is, among other things, information about the availability and quality of products.It has an informational role and has the ability to present a true picture of the products.Hence, we formulate the first hypothesis: H1: Information content positively influences online advertising value.

Credibility
of advertising influences consumers' attitudes and behavior.Wang et al. (2002) found that credibility relates directly to value and attitude toward advertising.This result confirms the findings of Ducoffe (1995Ducoffe ( , 1996) ) and Brackett and Carr (2001), which stipulate that credibility impacts advertising value and attitude toward advertising.Therefore, we propose the following two hypotheses: H2: Credibility positively influences online advertising value.
H3: Credibility has a positive effect on attitude toward online advertising.

Entertainment
Ability to entertain can improve the advertising experience.Ducoffe (1996) found a positive correlation between entertainment and advertising value.Similarly, Wang et al (2009) find that this factor brings pleasure and enjoyment.Then, our fourth hypothesis follows: H4: Entertainment acts positively on online advertising value.

Attitude towards online advertising
Attitude towards online advertising is defined as the aggregation of assessments of perceived attributes.It results from the value that an individual attaches to.Furthermore, the concept of attitude is defined as a predisposition to respond, favorably or unfavorably, after an assessment of a phenomenon (Burton et al, 1998).Ducoffe (1996) shows that online advertising value positively relates to attitude towards it.Hypothesis H5 reflects this relationship._________________________________________________________________________________________________________________

Purchase Intent
Hun and Kim (2014) point out that few studies have investigated the direct effect of advertising value on purchase intent.They found that advertising value positively affects purchase intent.Hence, our sixth hypothesis: H6: Online advertising value has a direct effect on purchase intent.

The mediating role of attitude towards online advertising
Consumers' attitude towards advertising influences their likelihood to purchase.Indeed, Mehta (2000) and Wang and Sun (2010) found that a positive attitude positively influences purchase intent and can result in frequent purchases.As we have assumed that attitude is induced by perceived value and that it acts on purchase intent, it seems that the variable "attitude" could play a mediating role in the relationship between value and purchase intent.Therefore, we formulate the following two hypotheses: H7: Attitude towards online advertising has a positive impact on purchase intent.H8: Attitude towards online advertising plays a mediating role in the relationship between advertising value and purchase intent.
In figure 1, we present the research model.We assume that the informativeness, entertainment and credibility determine perceived value of advertising, which in turn has a dual effect on purchase intent: a direct and an indirect effect through consumer attitude towards online advertising.A questionnaire was administered online, with the fans of Tunisie Telecom's Facebook page.The measurement scales used and the items measuring our variables are taken from the literature and as such they have already been validated.Therefore, we are not in a scale development process, but rather in a scale confirmation process.Moreover, because any principal component analysis (PCA) should be followed by a confirmatory factor analysis (CFA), the step of running a PCA becomes doubly superfluous and therefore in our study we proceeded directly to a confirmatory factor analysis (CFA) using structural equation models (SEM) processed by AMOS software.In this phase, the retained estimation method is maximum likelihood.However, as the sample size (210) is relatively small, we run a bootstrap procedure at 95% confidence intervals, and this for all the estimated parameters of our model.

The Independent variables
To measure informativeness of online advertising, we adopted the scale developed by Ducoffe (1995).In this scale, informativeness consists of four items.Credibility of online advertising scale is taken from Wang (2009).This scale consists of three items.To measure Entertainment of online advertising, we have adapted the scales developed by Ling et al. (2010) and Wang (2009), retaining only five items.

The Dependent variables
The dependent variables in our model are Value of advertising, Attitude and Purchase Intent.To measure advertising value on a Facebook page, we used the scale of Ducoffe (1995).This scale consists of three items.As for attitude towards advertising on a Facebook page, we combined the scales developed by Zang et al. (2009) and by Ling et al. (2010).The retained scale thus consists of four items.Finally, we used the three-item scale of Kim et al (2010) to measure purchase intent.

Reliability
The reliability of a measurement scale corresponds to its ability to produce observations coherent with each other.It is assessed through Jöreskog's rho coefficient of internal consistency, which should be above 0.7 so that reliability is judged to be good.

Convergent validity
Convergent validity of a measurement scale refers to its ability to provide results similar to those obtained using the same type of measures.It is considered good when the individual t-test of each factor loading exceeds 1.96.It is assessed by rho (ρVC) test whose value should exceed 0.5.

Discriminant validity
Discriminant validity measures whether two theoretically distinct variables are distinct in reality.When the value of rho of a variable exceeds the square of the correlations between this variable and all other latent variables in the model, the construct is considered to represent good discriminant validity.Table 3 shows the indices of convergent validity and reliability (ρVC and ρF).They also have a good discriminatory power.

Test of our research hypotheses
Estimation of the structural model was then conducted, since the scales of the variables of our model have been validated.The different equations modelling our hypotheses were tested simultaneously using AMOS software.
As for the estimation of the overall measurement model, we used the maximum likelihood method, coupled with a bootstrap procedure.We set the results of the different estimations at the 95% confidence intervals, for all the hypothesized relationships in our model.We first present the results of hypotheses H1 to H7.

Testing hypotheses H1 to H6
As shown in Table 5 below, the model provides good fit indices.The more online advertising provides information, the more credible and entertaining it is, the better it is perceived by Internet users.Hypotheses H1, H2 and H4 of our model are thus validated.
We note also that the effects of varying value and credibility on attitude are both positive and significant with respective CR values of 10.617 and 2.801 (Appendix 3A).Hypotheses H3 and H5 are thus validated.The more people attach value to online advertising and the more this latter is credible in their eyes, the more they develop positive and supportive attitude towards it.
We should note that although the obtained indices (observable variables) violate the multinormality assumption required to use the maximum likelihood method to estimate the parameters of the research model, our conclusions about the significance of these indices remain valid, as confirmed by the confidence intervals provided by the bootstrap procedure (Appendix 3B).The only confidence interval having a negative lower bound is that of the regression coefficient of the relationship PURCH INT <---VALUE.Hypothesis H6 which assumes a direct effect of perceived value on purchase Intention is rejected.

The mediating role of "Attitude"
Let us reproduce the estimated regression equations between the three variables Value, Attitude and Purchase Intent (Appendix 2 To conclude to the existence of a mediating role, we should check that the exogenous variable "value of advertising" has a significant effect on the endogenous variable "Attitude" in equation (1).The second condition relates to checking the existence of a significant effect of the variable assumed to play a mediating role (Attitude) on the endogenous variable Purchase Intent in equation (2).Reviewing CR values presented in parentheses, these two conditions are met.Moreover, as we found a CR of 0.842 (Appendix 3A), the effect of the variable "value" is not significant in equation ( 2) and thus mediation is complete.This means that perceived value has no direct effect on purchase intent.Therefore, the influence of perceived value on purchase Intent is observable via Attitude, following perception of online advertising value.It remains to confirm the presence of this effect by performing the Sobel test.The statistic obtained by this test is 5.0207> 2.
We reject the null hypothesis of a zero indirect effect and retain the existence of a comprehensive mediating role of Attitude in the relationship between perceived value and purchase Intent, and this with a firstrank risk of 5%.It follows that hypotheses H7 and H8 are retained.

Discussion of Results
Our study contributes to understanding the impact of advertising value on attitude of Internet users and to determining its effects on purchase intent.We found that informativeness positively affects a customer's perceived value of online advertising.This result is consistent with the results of previous studies (Ducoffe, 1995 and1996;Wang, 2009

Implications for management
The results of our study have implications for advertisers, and other agents who are interested in online advertising, specifically advertising in Facebook pages.Marketers and advertisers need to create credible and entertaining advertisements that provide value to customers and generate a positive attitude.Consequently, managers and advertisers should focus on creating value and informationally encapsulated messages for online advertising.Indeed, the design of an informative, credible and entertaining advertising positively affects perception of its value.This perception influences later consumer attitudes towards advertising on Facebook pages.A positive attitude towards this type of advertising positively affects purchase intent.However, advertising on a Facebook page that is not appreciated by consumers might affect the company's brand image and create a negative attitude in them which ultimately can lead to losing customers.

Limitations of the study
This study has some limitations.Additional studies are needed to overcome these limitations.Firstly, the sample size is small (210 participants).Secondly, respondents were only fans of the TT Facebook page who are already sympathetic with the operator.Third, the questionnaire was administered online in French, and it should have been translated into Arabic to ensure that respondents perfectly understand its sections.In addition, further research in this field can be done in other countries to point to cultural aspects to enrich the literature and to identify other factors affecting advertising value and consumer attitude towards advertising targeting fans of Facebook pages.

Future research
Future studies may include other factors that could affect value of online advertising, which in turn influences consumer attitudes toward online advertising and the use of this type of advertising to buy.For instance, researchers may find out whether there are other variables with a moderating effect on advertising value and consumer attitudes towards online advertising.They can also explore other factors that may influence the behavior of Internet users and their intention to purchase.

Table 1 : The sample
Journal of Marketing Research & Case Studies, DOI: 10.5171/2016.35778381.5% have a monthly income lower than 1,000 dinars, or about 445 Euros.We estimated the overall measurement model, with all its latent variables and items.Indeed, taking into account all the variables of the model is necessary to study the discriminant validity of the different scales.All items are measured on a five-point Likert scale.Items that have high standardized residuals were removed.The fit indices were ______________Salem Ben Brahim (2016),