causal model analysis of the e ect of facebook ad ...causal model analysis of the e ect of facebook...

Causal Model Analysis of the Effect ofFacebook Ad Constituents on

Advertisement Attitude

EunHee Kim1 and SeungYeob Yu∗2

1Dep. of Advertising & Public Relations & journalism,Mokwon University 88, Doan-dong, Seo-gu,

Daejeon-si, 35349, Republic of [email protected]

2Dept. of Advertising and Public Relations,Namseoul University 91, Daehak-ro, Seonghwan-eup,

Seobuk-gu, Cheonan-si, Chungnam, 31020,Republic of Korea

[email protected]∗2

Abstract

Background/Objectives:The present research aims toidentify the unique characteristics and constituents of Face-book advertisements, and the effect of these constituents onadvertising effect by utilizing causal model analysis.

Methods/Statistical analysis: A frequency analy-sis was performed to identify the general characteristicsof respondents. Also, an exploratory factor analysis wasconducted to identify Facebook ad constituents, and Cron-bach’s Alpha was obtained to test the reliability of eachfactor. Moreover, the validity of potential variables was ver-ified by using SPSS 20.0 program. Finally, AMOS 20.0 wasemployed to verify the goodness-of-fit of the confirmatoryfactor analysis and to test the finalized model.

Findings: The results of analysis can be summarizedas follows. As a result of exploratory factor analysis, fac-tors of Facebook advertisement were identified as Interest,

International Journal of Pure and Applied MathematicsVolume 120 No. 6 2018, 5331-5349ISSN: 1314-3395 (on-line version)url: http://www.acadpubl.eu/hub/Special Issue http://www.acadpubl.eu/hub/

5331

customized information, exposure, bystander reactivity, andproduct reviews info. It was found that the factor of interesthad a positive effect on Facebook ad fitness. Thus, accept-ing Hypothesis 1-1. It was also found that the factor of cus-tomized information also had a positive impact on Facebookad fitness. Thus, Hypothesis 1-2 was accepted.Bystanderreactivity, in turn, exerted a positive effect on Facebookad fitness. Therefore, Hypothesis 1-4 was accepted. Thefactor of interest, in turn, had a positive effect on Face-book ad attitude. Thus, Hypothesis 2-1 was accepted.Itwas found that customized information exerted a positiveeffect on Facebook ad attitude. Thus, Hypothesis 2-2 wasaccepted. Facebook ad fitness had a positive impact on adattitude. Thus, Hypothesis 3 was accepted.

Improvements/Applications: The results of the cur-rent analysis can be meaningful in that it obtained base datathat can be practically used by advertisers and ad producersin establishing strategies and plans to make Facebook ads.Also, the link between Facebook constituents and ad fitnessas well as its ad effect was identified.

Key Words : Facebook ad, Facebook ad constitutionfactor, Fitness, Ad attitude, Causal model analysis

1 Introduction

Facebook was ranked at Number 4 in the list of most valuablebrands in April 2017 according to Forbes, an influential magazine[1].Facebook, initially developed as a website for a dormitory atHarvard University in 2004, is now used, as of 2017, by more than1.9 billion users around the world [2].It has become a global leaderin social networking services and its brand and influential power isbeyond dispute, leading the trend in the 21st century.

Facebook offers network-oriented services. It started with theconcept of expanding offline relationships to online and is now usedby 2.1 billion users around the world. It does not merely containthe whereabouts and news from friends but also a variety of infor-mation dealing with various topics and interests, which would allhelp develop social networks. Facebook users now spontaneouslyget connected to one another to seek necessary information andcontents, and play the role of content producers in vitalizing social

International Journal of Pure and Applied Mathematics Special Issue

5332

networks [3].Facebook’s inexorable and rapid growth has also brought success

as an advertising platform as well as Web services. From Octoberto December in 2016, for example, sales from advertisements ac-counted for 98% of the total sales, and the number of users a daywas 1.23 billion, which was a 18% increase from the previous year[4].

Facebook is well acknowledged by marketing and advertisingpersonnel as ad contents. According to the survey conducted byFirebrand Group and Simply Measured in March 2016, in whichrespondents were asked to choose three most effective social plat-forms for advertising, 95.8 % of the marketers chose Facebook [5].Itclearly shows that Facebook is an indispensable platform for SNSmarketing.

Facebook’s success as an ad platform can be attributable to itsunique characteristics different from conventional internet and mo-bile advertisements. As a typical social networking service, it makesfull use of social relationships including the relationship with users,and the interaction among users. Thus, a Facebook ad would notremain as a tool of delivering message but can target a particu-lar group, using information of the users, to establish an affinityrelationship and build up trust.

An ad with the type of Sponsor Story may come in the contextof UCC (User Created Contents) by users’ spontaneous posting offeedbacks and information on products and stores though variouschannels including Fan pages. Facebook users also share their opin-ions as well as information, for example, by clicking on ‘I like it’. Allthese characteristics of Facebook ads have brought about changesin the paradigm of advertising.

In such a context, the present research aims to identify theunique characteristics and constituents of Facebook advertisements,and the effect of these constituents on advertising effect by utilizingcausal model analysis.

From the practical perspective, the identified effect of the con-stituents on advertising attitude is expected to help advertisers andad producers understand what elements should be considered inmaking Facebook advertisements.


5333

2 Facebook advertisements

Facebook has offered services with the core values of participation,sharing and opening. Its advertisement platform, similar to Pageand Google AdWords, has shown a remarkable growth, since it waslaunched in November 2007[6].Facebook ad platform is based onself-service, which means that advertisers register and manage adson independently by using SNS social information [7].The strategyof opening services to anybody and its characteristic of enabling ad-vertisers to target a particular group of consumers by using users’personal information have contributed to the success of its ad plat-form.

Facebook ads can be classified into various types: Facebookads, image ads, video ads, slide ads, canvas ads, link ads, collectionads, page ads, event response ads[8]. Such a variety in types mightreflect its differentiated constituents than conventional mobile ads.

The characteristics of Facebook ads can be summarized as fol-lows. First, in terms of exposure, a Facebook ad is exposed inthe ‘New Speed’ space, just like other contents posted by ordinaryusers. How, the phrase ‘sponsored’ is presented to tell them it isan ad. The right side of the picture is contents posted by users andadvertisements are exposed on the left side. It is often classifiedbelonging to the type of Native ads [9].

Second, as for the ad materials, Facebook ads can be madewith texts, images, and video files. In particular, for mobile devices,video ads can be most effective [9].Third, one can set up and managethe ad design, method of advertising and range of promotions on thehomepage. In other words, advertisers can make decisions on manyaspects of ads depending on their products and target consumers[7].

Fourth, the media channels Facebook ads are exposed to includeFacebook, Instagrarm and Facebook audience networks have part-nerships with Facebook for ad exposure. Fifth, Facebook ads arecharged with the criterion of CPC(cost per click). The CPC for eachad can not surpass an advertiser’s bidding price. Thus, advertiserscan control their total expenses by means of their bidding prices,which can perhaps help secure higher effect for lower expenses incomparison to other traditional media[9].Next, as far as targetingis concerned, customized targeting is possible, since advertisers can


5334

use big data from demographic information on users and users’ ex-periences. Such information is accessible, since Facebook users canbe identified as they are logged on [9].

One of the greatest benefits of Facebook as an ad platform wouldbe an enormous amount of spontaneous word-of-mouth (WOM)activities by many users in the network. The spread of informationcan be done through Likes, reply postings, and sharing [10]. So farwe have discussed the types and characteristics of Facebook as anad platform. Based on these characteristics, this research intendsto identify how the Facebook ad constituents affect the advertisingeffect.

3 Research hypotheses

Facebook builds a community though communication and sharing.Such a character naturally leads to rapid spread of informationprovided by businesses or individuals through users’spontaneousword-of-mouth activities in the network. Thus, Facebook as anad platform has various qualities to solve problems of conventionalchannels of advertising by enabling establishment of interactive net-works of friends from school, work or personal relationship and bynot depending on one-way delivery of information.

Also, Facebook can employ customized targeting method foreffective delivery of ad messages by matching customer data andFacebook user information. Thus, it can induce behavior and ac-tions by identifying values of customer behavior, depending onFacebook users’ interests [11].Such targeting method of Facebookcan provide users with appealing information and customized in-formation. Thus, it should be meaningful to explore the effect ofsuch characteristics on the effect of Facebook ads.

Facebook has presented a change in the way how ads are ex-posed. To take an example, the so-called ‘native ads’ are providedin the same way platform contents are exposed, thereby reducingconsumers’ repulsion against being exposed to ads. In other words,the consistency in the way contents and ads are presented can pro-duce high effect of Facebook advertisements. The current researchaims to see which of these characteristics of Facebook ads con-tributes to positive effect on ad fitness. Also, considering the simi-


5335

larity between the construct of Facebook pages and Facebook ads,we attempt to identify how such characteristics influence users’ adattitude. For that purpose, the following hypotheses were estab-lished.

Hypothesis 1: Facebook ad constituents will have a positiveeffect on ad fitness.

Hypothesis 1-1: The factor of interest will have a positiveeffect on ad fitness.

Hypothesis 1-2: The factor of customized information willhave a positive effect on ad fitness.

Hypothesis 1-3: The factor of exposure will have a positiveeffect on ad fitness.

Hypothesis 1-4: The factor of bystander reactivity will havea positive effect on ad fitness.

Hypothesis 1-5: The factor of product reviews informationwill have a positive effect on ad fitness.

Hypothesis 2: Facebook ad constituents will have a positiveeffect on ad attitude.

Hypothesis 2-1: The factor of interest will have a positiveeffect on ad attitude.

Hypothesis 2-2: The factor of customized information willhave a positive effect on ad attitude.

Hypothesis 2-3: The factor of exposure will have a positiveeffect on ad attitude.

Hypothesis 2-4: The factor of bystander reactivity will havea positive effect on ad attitude.

Hypothesis 2-5: The factor of product reviews informationwill have a positive effect on advertising attitude

Hypothesis 3: Ad fitness will have a positive effect on adver-tising attitude.

4 Research Model

The purpose of this study is to test hypotheses about the directinfluence of the constituents of Facebook advertisement on fitnessand ad attitude. In addition, we tried to confirm the indirect effectof fitness on ad attitude through hypothesis test. In other words,we tried to confirm the mediating effect. For this purpose, five fac-


5336

tors were identified as explaining factors of Facebook advertisementthrough exploratory factor analysis. That is, Interest, customizedinformation, Exposure, bystander reactivity, product reviews info.The research model for the hypothesis testing of this study is shownin Figure 1.

Figure 1: Research Model

5 Research Method

5.1 Research subjects

The participants of the current research consisted of a group of 350college students using Facebook: 137 male (39.1%) and 213 female(60.9%). Their average time using Facebook was 2 hours and 35minutes a day.


5337

5.2 Operational definition of variables

The set of Facebook ad constituents is defined as a group of uniquefeatures of the ads exposed through Facebook platform. A set of24 measurement items were adopted from Yu and Kim [12] model,as described below.

First, the factor of interest consisted of 7 measurement itemsin the form of the following statements: ‘Facebook ads are moreinteresting than those in other media’. ‘Facebook ads are more at-tracting than those in other media’, ‘I feel curious about the adsthat received many ‘Likes’ from friends and acquaintances’, ‘Face-book ads attract interest in products.’ ‘Facebook ads stimulate mycuriosity’, and ‘Facebook ads stimulate my curiosity, since I cannotsee them in other media than Facebook’.

The factor of customized information consisted of 4 measure-ment items, which were presented in the following statements: ‘Face-book ads provide me with necessary information in the form offeedbacks’, ‘It is easy and convenient to get information of prod-ucts from Facebook ads’, ‘In Facebook, ads of the products I aminterested are presented’, and ‘Facebook provides me with the in-formation of the products appropriate(optimal) for me’.

The factor of exposure consisted of the following six items: ‘InFacebook, I am naturally exposed to ads, while paging down theNew Speed space’, ‘I am forcibly exposed to ads in Facebook’, ‘Ioften run across ads involuntarily, with intention to click on dramasor reality programs’, ‘Regardless of my will, I am exposed to adsin popular pages of Facebook’, and ‘I do not do Facebook muchbecause of too many ads in it’.

The factor of bystander reactivity, in turn, consisted of the fol-lowing three items: ‘The contents in tags of friends and acquain-tances lead me to watch Facebook ads’, ‘I watch Facebook ads dueto the comments of friends and acquaintances’, and ‘Friends’ ‘Likes’lead me to watch Facebook ads’.

The factor of product reviews information was measured by thefollowing four items: ‘I can see product reviews by seeing Facebookads’, ‘I see Facebook ads, with the knowledge that they are ads fornearby restaurants or cafes’, ‘Users’ reviews of products appear invideo files’, and ‘I can see product reviews and product effects invideo files.’


5338

The factor of ad fitness refers to consistency and harmony be-tween Facebook pages and Facebook ads. In order to measure thisfactor, Jung[13]list of items was slightly modified for this research.The measurement items include the following four items: ‘Facebookads well match the structure and contents of Facebook’, ‘Facebookads are harmonious to the structure and contents of Facebook’,‘There is consistency between Facebook and Facebook ads in termsof structure and contents’, and ‘Facebook ads are adequate to thestructure and contents of Facebook’.

Facebook ad attitude refers to the degree to which consumerslike Facebook ads. To measure this, a slightly modified version ofBrackett and Carr’s[14]measurement items. It consisted of 5 itemslike the following: ‘Facebook ads are appealing’, ‘Facebook ads areimpressive’, ‘Facebook ads are attracting’, ‘Facebook ads provideuseful information’, and ‘Facebook ads are persuading’. Each of theitems was measured on a five point scale: 1 for strongly disagreeand 5 for strongly agree.

5.3 Analysis method

A frequency analysis was performed to identify the general charac-teristics of respondents. Also, an exploratory factor analysis wasconducted to identify Facebook ad constituents, and Cronbach’sAlpha was obtained to test the reliability of each factor. Moreover,the validity of potential variables was verified by using SPSS 20.0program. Finally, AMOS 20.0 was employed to verify the goodness-of-fit of the confirmatory factor analysis and to test the finalizedmodel.

6 Research Results

6.1 Reliability and validity of the measurementtool

An exploratory factor analysis and a reliability test were conductedfor the factors used in this research. Also, descriptive statisticalanalysis was performed to obtain the mean and standard deviationof the measurement items of the variables and to identify the degree


5339

of goodness. The results of such analyses are illustrated in Table1. The means of the items were obtained from 2.87 to 4.31 andthe standard deviation ranged from .702 to 1.111, which can beconsidered in good standing.

An exploratory factor analysis and reliability test of the factorsunder discussion found the following results. First, Kaiser-Meyer-Olkin(KMO) measure was obtained at .894. Bartlett’s sphericitytest (X2= 4694.084, df=231, p<.001) found the adequacy of sam-pling. And a couple of items were eliminated from a set of 24Facebook ad constituents, resulting in a set of 22 items categorizedinto five factors.

The first factor named ‘interest’ consisted of seven measurementitems (explained variance: 37.457%). The second factor consistedof four items and was called ‘customized information (explainedvariance: 13.519%). Factor 3 consisting of three items was called‘exposure’ (explained variance: 7.087%). The fourth factor con-sisted of three items and was called ‘bystander reactivity’ (explainedvariance: 6.948%). Finally, the fifth factor consisted of four itemsand was called ‘product reviews information (explained variance:4.983%). The factor of fitness consisted of four items and its re-liability was greater than .8. The factor of ad attitude, in turn,consisted of five items and its reliability was also greater than .8.


5340

6.2 Confirmatory factor analysis

A confirmatory factor analysis was conductedto identify the con-sistency of relationship structure between measurement items andpotential variables. A set of indexes including X2, CFI, NFI, TLI,GFI, RMR, RMSEA were extracted to verify the model fitness.Those evaluated as error items in the confirmatory factor analy-sis were eliminated to improve model fitness: measured variable 10in the potential variable of interest, number 14 of product reviewsinformation, numbers 4 and 5 of ad attitude. The figures in Ta-bles 2, 3, 4, and 5 illustrate the results of analysis of the modelfitness: X2=477.075(df =274, p=0.000), GFI=0.909, CFI=0.969,NFI=0.913, TLI=0.953, RMR=0.040, RMSEA=0.046(See Table2).

First, the t value of observation variables was all significant atthe level of p< 0.001. Also, the standardization coefficient of obser-vation variables ranged from 0.603 ∼ 0.921. Thus, the convergentvalidity and reliability of each item was verified good, since thecoefficient was greater than 0.5, which is the criterion. AVE (Av-erage Variance Extraction) and construct reliability were measuredto identify the convergent validity of the variables under discussion.First, AVE ranged from 0.568 to 0.771, which was greater than thecriterion of 0.5. The construct reliability was obtained from 0.796to 0.915, which was greater than the criterion of 0.7[15, 16]. Thus,


5341

convergent validity of the measurement items was secured(See Ta-ble 3).

The discriminant validity of the constituents was identified bycomparing their AVE index and the square of correlation coefficients[16].It was found that the AVE’s, which ranged from 0.568 to 0.771,were greater than their square of correlation coefficients, which wereobtained between 0.174 and 0.291. Thus, the discriminant validityof the factors was identified(See Table 4 and 5).


5342

6.3 Research hypothesis and Model verification

6.3.1 Model verification

The validity of the presented research model was verified to identifythe effect of Facebook ad constituents on ad fitness and ad attitude.The structural equation test of model fitness examined each of thesuggested path hypotheses and found the results illustrated in Table6. It can be concluded that the fitness indexes all satisfied thecriteria.


5343

6.3.2 Test of Hypotheses

The current research attempted to identify the effect of Facebookad constituents on Facebook ad fitness and ad attitude through astructural equation model analysis. The results of analysis can besummarized as follows(See Figure 2). It was found that the factorof interest had a positive effect on Facebook ad fitness. The stan-dardized path coefficient between the two variables was obtained at0.13(t=2.423, p< .05), thus accepting Hypothesis 1-1. It was alsofound that the factor of customized information also had a positiveimpact on Facebook ad fitness. The standardized path coefficientbetween the two variables was obtained at 0.46(t=4.824, p< .001).Thus, Hypothesis 1-2 was accepted.

The factor of exposure did not have a significant effect on Face-book ad fitness, thus rejecting Hypothesis 1-3. Bystander reactivity,in turn, exerted a positive effect on Facebook ad fitness. The stan-dardized path coefficient between the two variables was obtainedat 0.17(t=2.300, p< .05). Therefore, Hypothesis 1-4 was accepted.Product reviews information, in contrast, did not have a significantimpact on Facebook ad fitness, thereby rejecting Hypothesis 1-5.The factor of interest, in turn, had a positive effect on Facebook adattitude. The standardized path coefficient between the two vari-ables was obtained at 0.13(t=4.477, p< .001). Thus, Hypothesis2-1 was accepted.

It was found that customized information exerted a positiveeffect on Facebook ad attitude. The standardized path coefficientbetween the two variables was obtained at 0.44(t=4.561, p< .001).Thus, Hypothesis 2-2 was accepted. The factor of exposure, inturn, did not exert a positive effect on Facebook ad attitude. Thus,Hypothesis 2-3 was rejected. Bystander reactivity also did not havea positive effect on Facebook ad attitude, rejecting Hypothesis 2-4.The factor of product reviews information did not have any effect onad attitude, thus rejecting Hypothesis 2-5. Facebook ad fitness hada positive impact on ad attitude. The standardized path coefficientbetween the two variables was obtained at 0.15(t=3.243, p< .01),Thus, Hypothesis 3 was accepted.


5344

Figure 2: Finalized model

7 Concluding Remarks

Facebook has recently emerged as a new advertisement platformthanks to a variety of its beneficial features: a new type of ex-posure appropriate to users’ environment, individually customizedinformation for users, participation in marketing events, and adver-tising using communities of acquaintances. The current research es-tablished a research model to identify the effect of the constituentsof Facebook ads on ad fitness and users’ ad attitude.

The implications of the research results can be summarized asfollows. The set of unique characteristics of Facebook ads had apositive effect on ad fitness: interest, customized information, by-stander reactivity, product reviews information. The contents ofads can be interesting and attract curiosity from users. Also, prod-ucts and information can be customized to a particular group ofusers. And, tags, comments and responses by ‘Like’ can let users


5345

know the opinions of friends and acquaintances.These unique characteristics of Facebook ads might result from

the establishment of social networks and communities, thus differ-entiating them from other mobile ads. Therefore, marketing person-nel and advertising planners should better understand the uniquecharacteristics of Facebook ads and that users’ repulsion against adscan be decreased if the consistency and harmony between Facebookpages and Facebook ads would make them feel naturally exposedto the ads.

Second, it was found that interest and customized informationwere the constituents that would influence users’ ad attitude. By-stander reactivity, on the other hand, had a positive effect on adfitness, but not on ad attitude. It might mean that interesting andcuriosity-inducing contents of advertisements help enhance view-ers’ ad attention and persuasive power of ad messages.Also, it wasfound that the constituents of product reviews information and cus-tomized information of Facebook ads would help users gain usefulinformation. Thus, the ads would impress users that they were de-livered with persuasive power to them. What it might mean wouldbe that ad makers can enhance the effect of their Facebook ads byputting interesting and curiosity-inducing contents and providingnecessary customized information for their target users.

The results of the current analysis can be meaningful in that itobtained base data that can be practically used by advertisers andad producers in establishing strategies and plans to make Facebookads. Also, the link between Facebook constituents and ad fitnessas well as its ad effect was identified. This finding might be valuedas having presented academic significance.

The limitations of the current research are as follows. First,the research was carried out only with a group of college students,who arguably use Facebook the most. However, it is also truethat Facebook users range all age groups of all types of profession.Thus, further studies are expected to expand to various groups ofsubjects. Second, it should be admitted that types of Facebookads were not taken into consideration. So, future researches areexpected to note that afferent set of constituents can perhaps berelated with different types of ads.


5346

8 Acknowledgment

Funding for this paper was provided by Namseoul University.

References

[1] Business Post. (2017). Forbes Ranked 1st in Apple and 10th inSamsung Electronics for ”Most Valuable Brand”. Retrievedfromhttp://www.businesspost.co.kr/BP?command=naver&n-um =49857

[2] f newsroom. (2017). 2017 1Q Earnings Release Highlights.Retrieved from https://ko.newsroom.fb.com/?s=2017%EB%85%84+1%EB%B6%84%EA%B8%B0+%EC%8B%A4-%EC%A0%81

[3] BLOTER. (2017). Facebook’s New Missionand ”Facebook Group”. Retrieved fromhttps://www.bloter.net/archives/296762

[4] ChosunBiz(2017). Facebook, Quarterly Sales of 10 TrillionWon ... By mid-2017, 2 Billion Monthly Users. Retrievedfrom http://biz.chosun.com/site/data/html dir/2017/02/02/2017020200661.html#csidx6eee4a7cc609e52aa99e24f6af40366

[5] i-boss. (2016). Facebook Selected as The Best Platform forAdvertisement Efficiency. Retrieved from http://www.i-boss.co.kr/ab-2877-2277

[6] BLOTER. (2012). Until becoming a 111 Trillion WonCompany ... Facebook 8 Years. Retrieved fromhttp://www.bloter.net/archives/94516

[7] JinHee Doo, Jung Hyun Kim. (2012). The Influence of Face-book Ad Type on Advertising Effect Persuasion KnowledgeModel and Technology Acceptance Model. The Korean Jour-nal of Advertising and Public Relations, 14(2), 300-330.

[8] Facebook Business. (2017). Facebook Advertising Guide.Retrieved from https://www.facebook.com/business/ads-guide/


5347

[9] i-boss. (2017). Features of Facebook Ads. Retrieved fromhttp://www.i-boss.co.kr/ab-74577-388

[10] Gui Hong Min, HeunSeon Park, MiJeong Song, JinKyun Lee.(2015). The Effects of Types of Ads and the Intensity of Use onElectronic Word-of-Mouth in Facebook: The Moderating Roleof Independent Self-Construal. 1H.Retrieved from The KoreanJournal of Advertising,26(8), 7-33.

[11] Nasmedia. (2016). Media Issue Report 2016 Re-port by Issue in 1H.Retrieved from 1-46.https://www.slideshare.net/nasmedia/2016-2016071328f29

[12] Seungyeob Yu, Eunhee Kim. (2017). What constitutes a Face-book ad?.Retrieved from Consumer Integration Conference.

[13] Da ha Jung. (2014). A Study on the Native Advertising Ef-fect in Social Media (Master’s thesis). Graduate School, KoreaUniversity.

[14] Brackett, L. K. & B. N. Carr. (2001). Cyberspace advertis-ing vs. other media: Consumer vs. mature student attitudes.Journal of Advertising research, 41(5), 23-32.

[15] Bagozzi, R. P., & Yi, Y. J. (1988). On the evaluation of struc-tural equation models. Journal of Academy of Marketing Sci-ence, 16, 74-94. doi:10.1007/ BF02723327

[16] Fornell, C. and Larcker, D.F. (1981). Evaluating structuralequation models with unobservable variables and measure-ment error. Journal of Marketing Research, 18(1), 82-388.doi:10.2307/3151312


5348

causal model analysis of the e ect of facebook ad ...causal model analysis of the e ect of facebook...

Documents