The effect of disclosure of sponsorship in
a YouTube video on viewers’ attitudes
towards the video and towards the
spokesperson
Bachelor Thesis Communication Science Tamar Hellinga
University of Amsterdam 10784675
Teacher: Eline Smit Word count: 6179
Abstract
YouTube acts as an open platform for users to post videos on. Within these videos
promotional content can be present. Whether that is communicated as such or not refers to disclosure of sponsorship. This study researches the effect of disclosed sponsorship on viewers’ attitudes towards both the video and the spokesperson in it. On top of that, the moderating effects of interest in YouTube and knowledge of the spokesperson were tested. Using an existing video about recommended alcohol intake the stimulus material was made, creating a control condition, without ‘#ad’ written in it, and a manipulated condition, with ‘#ad’ written in the corner. Among the survey sample (N = 68) there was no significant effect for disclosure of sponsorship on video attitude nor on spokesperson attitude. Likewise,
regarding the moderating effects of interest in YouTube and knowledge of spokesperson no significant effects were found.
Introduction
YouTube hosts a space for over a billion users to upload, view and react to videos (YouTube, n.d.). The platform relies heavily on user-generated content (Freeman &
Chapman, 2007) and puts the ‘you’ central (van Dijck, 2009). The content that is created and uploaded by individual users can be seen as a form of free labour (Andrejevic, 2009). Not only do these videos create data for companies to respond to and work with, but they can also function as a promotion, whether or not that is the intention of the uploader. This blurred line between user and creator (van Dijck, 2009) could also imply a blurred line between user, creator and advertiser. Videos that may seem like a regular user-generated content video could in fact be sponsored (Freeman & Chapman, 2007).
There are people who are able to make money through making YouTube videos (Postigo, 2016), who in this study will be referred to as ‘YouTubers’. This income can come from the sites own advertising system, Google AdSense, but is more commonly earned through brand deals and promotion of products and services. YouTube creates new business opportunities (Cha, Kwak, Rodriguez, Ahn & Moon, 2007) that are interesting for both creators and advertisers. Advertisers can take advantage of the narrative method that many users employ (Pace, 2008). Beuker and Abbing (2010) show in their research that digital experience influences people on whether or not to buy a product. This could very well be applicable to YouTube, where people could easily be influenced by user-generated content. Promotional content on YouTube is characterized by a centrality of self-promotion, with brands playing the supporting instead of the central role (Smith, Fischer & Yongjian, 2012). On top of that, the content is more positive than promotional content on the social
networking sites Twitter and Facebook, which could be caused by the fact that YouTube is inherently less focused on conversation and link sharing than the other two sites (Smith et al., 2012). Expenditure on online advertising is increasing and allow strong customer relationships to be formed (Becker-Olsen, 2003). Therefore, whether or not a viewer knows of the spokesperson in the video might play a role with regard to their attitude, as knowledge could be related to feeling a relationship to the spokesperson.
Through the so-called sponsored content that creators can post on YouTube, uploaders of it are able to receive money or products from a company without intervention from the social medium. Different countries employ different regulations concerning sponsored content on social media. The Netherlands, for example, have employed a code that states that social media users are required to disclose sponsored content as such. However, there are no consequences attached to a user not obliging to this code (Stichting Reclame Code, 2009). By such, users can upload sponsored content that is not clearly labelled as sponsored content, which leaves viewers with misconceptions about the motives and sincerity of the video and spokesperson. Through this, the medium could be used for the wrong motives. YouTubers could use their big audience solely to earn money, maybe even without considering what is being promoted and how it influences the audience.
This issue is especially relevant regarding videos on health-related subjects. For viewers, receiving correct information about health is vital, as misleading information could have negative consequences (Brna, Dooley, Esser, Perry & Gordon, 2013; Delli, Livas, Vissink & Spijkervet, 2016; Haymes & Harries, 2016). The promotion of drinking could for example possibly lead to underage drinking, irresponsible behaviour towards alcohol or even alcohol addiction. For YouTubers, it is interesting to know how disclosure of sponsorship influences their audience. A negative influence could make creators unwilling to disclose sponsorship or make them find alternative ways of informing about sponsorship. Lastly, for advertisers it is important to know whether online promotional content on YouTube has a positive effect regarding attitudes or not, as this could be used to improve future advertising.
Apart from being socially relevant, this study aims to fill a gap in literature. While health communication on YouTube has been researched sufficiently (Hellinga, 2016), little is known about attitudes towards videos containing disclosed content and the spokespersons within them, specifically on YouTube and regarding health communication. Within this concept, viewers’ knowledge of the spokesperson and their interest in YouTube is also taken into account. Besides, as the online world is dynamic and continually developing, scientific research should attempt to keep up with it. Inspired by this the following research question has been formulated.
RQ: What is the difference in influence between content with sponsorship disclosure and content without this disclosure in a health-related video on viewers’ attitude towards the video and towards the spokesperson of the video?
Conceptual model
The proposed research question contains several concepts that require defining. These concepts are related to each other within a conceptual model (Figure 1). The main effects that the research question contains are the effects of disclosure of sponsorship on both attitude towards the video and attitude towards the spokesperson in it. Besides that, the model displays two moderating variables. The first is whether the viewer of the video knows of the spokesperson or not and the second is the degree of interest in YouTube of a viewer.
Disclosure of sponsorship
The first concept that needs defining, is the type of sponsored content that viewers are exposed to. This concept serves as the independent factor in this experiment. Disclosure of sponsorship is the explicit mention to the audience that commercial content is integrated into editorial content (Boerman, van Reijmersdal & Neijens, 2012). Two conditions will be
created, based on whether or not the sponsored content is clearly stated to be sponsored. The first condition consists of a video about recommended alcohol intake without any disclosure of sponsorship. The second condition consists of the exact same video, but has ‘#ad’ written in the top left corner and by doing so employs the use of disclosure of
sponsorship. The indication ‘#ad’ is common on YouTube and indicates that a video has been sponsored or is an advertisement for a product, service or company. In this study, the first condition serves as the control condition and the second as the manipulated condition.
Attitude towards video
Secondly, the concept of attitude towards the video will be measured as the first of two dependent variables. The concept of attitude can be described as an evaluation based on norms and perceptions (Hudson & Rosen, 1953). For this study, this definition is relevant and is used in combination with video. A viewers’ attitude towards the video therefore is the evaluation of the video, which is based on the viewers’ underlying norms and perceptions. This can be graded on a scale ranging from negative to positive. The score for attitude can lie anywhere between these two values. Previous research has shown that viewers have more negative attitudes towards blog posts that have disclosed sponsorship than towards blog posts without disclosure of sponsorship (van Reijmersdal, Fransen, van Noort, Opree, Vandeberg, Reusch, van Lieshout & Boerman, 2016; Colliander & Erlandsson, 2015). This could be explained by the fact that people create a feeling of resistance towards sponsored
content (van Reijmersdal et al, 2016). Based on this information, it is expected that a video with disclosed sponsorship leads to a less positive video attitude than a video without disclosed sponsorship, as displayed in Figure 1. This expectation manifests itself in the first hypothesis:
H1: Videos provided with ‘#ad’ produce a more negative video attitude than videos that are not provided with ‘#ad’.
Attitude towards spokesperson
The second dependent factor in the conceptual model for this experiment is the viewers’ attitude towards the spokesperson in the video. The same definition for attitude as for video attitude is used to describe this concept. Therefore, this concept can be defined as the evaluation of the spokesperson, based on underlying norms and perceptions. The spokesperson in this case is a YouTuber, Hannah Witton. Expectations regarding spokesperson attitude are in line with those for video attitude and are manifested in the following hypothesis:
H2: Videos provided with ‘#ad’ produce a more negative spokesperson attitude than videos that are not provided with ‘#ad’.
Viewer knows of spokesperson
Besides the independent and dependent factors, there are two factors expected to influence the relation between sponsored content and attitude. The first is whether or not the viewer knows of the spokesperson. The YouTuber in the video had a subscriber count of 302,2831 at the time of writing this study. It is thus possible that participants in this study know of her. Because of this, viewers could already have a certain relationship with the spokesperson. A relationship between visitor and blogger leads to a more positive video attitude than a lack of this relationship (Lee & Watkins, 2016). This concept is expected to influence the effect of disclosure of sponsorship on both video attitude and spokesperson attitude. The expectation is that if a person knows of the spokesperson, the effect of disclosed sponsorship is less negative on attitudes than if a person does not know of the spokesperson. On top of that, it is expected that the positive effect of non-disclosed sponsorship on attitudes is enforced by knowledge of the spokesperson. Based on this knowledge, the following four hypotheses have been formed:
H3a: For viewers who know of the spokesperson the effect of videos with ‘#ad’ on the attitude towards the video is less negative than for viewers who do not know of the spokesperson.
1
H3b: For viewers who know of the spokesperson the effect of videos without ‘#ad’ on the attitude towards the video is more positive than for viewers who do not know of the spokesperson.
H3c: For viewers who know of the spokesperson the effect of videos with ‘#ad’ on the attitude towards the spokesperson is less negative than for viewers who do not know of the spokesperson.
H3d: For viewers who know of the spokesperson the effect of videos without ‘#ad’ on the attitude towards the spokesperson is more positive than for viewers who do not know of the spokesperson.
Interest in YouTube
Lastly, the second factor that is expected to moderate the relation between sponsored content and attitude is a viewers’ interest in YouTube. This concept is measured by how many hours a day a respondent watches videos on YouTube. The more hours a person spends on YouTube, the higher the interest for YouTube according to this measure. As time spent on the medium influences attitude in a positive way (Lee & Watkins, 2016), the
expectation is that viewers with a higher interest in YouTube have a more positive attitude towards both the video and the spokesperson than viewers with a lower interest in YouTube. The expectation is hence that a high interest in YouTube weakens the negative effect of disclosure of sponsorship on attitudes. On top of that, it is expected that a high interest in YouTube strengthens the positive effect of non-disclosed sponsorship on attitudes. The following four hypotheses display these expectations:
H4a: For viewers with a high interest in YouTube the effect of videos with ‘#ad’ on the attitude towards the video is less negative than for viewers with a low interest in YouTube. H4b: For viewers with a high interest in YouTube the effect of videos without ‘#ad’ on the attitude towards the video is more positive than for viewers with a low interest in YouTube. H4c: For viewers with a high interest in YouTube the effect of videos with ‘#ad’ on the attitude towards the spokesperson is less negative than for viewers with a low interest in YouTube.
H4d: For viewers with a high interest in YouTube the effect of videos without ‘#ad’ on the attitude towards the spokesperson is more positive than for viewers with a low interest in YouTube.
Figure 1 Conceptual model
Methods
Stimulus material
The stimulus for this study consisted of an edited online video on recommended alcohol intake, first posted by Hannah Witton, a British YouTuber (Hannah Witton, 2016). This video was edited from 7:07 minutes to 2:50 minutes to ensure viewers would not be tempted to click away as this seems to be a reasonable length (see Appendix A). In this video, she talks about guidelines regarding alcohol intake in the United Kingdom and other countries. The stimulus material first shows her talking about the governmental website drinkaware.co.uk. She proceeds the video with a stop-motion with a voice-over by herself about units, which are indicative of recommended alcohol intake. After the stop-motion, the video cuts back to Hannah Witton standing in front of the camera. She continues the video by speaking about alcohol guidelines in countries other than the United Kingdom. The video ends similar to many other videos from YouTubers: she asks people to like the video, subscribe to her channel and comment below the video.
Two different conditions were created using this video. They differed regarding one aspect only: one video had #ad written in the top left corner and one did not. This video was used with permission from the uploader (see Appendix B). The original title and thumbnail, a still screen that is shown before clicking on a video, were also used. The sponsorship was fictive, as Hannah Witton states in the original video that the content was not sponsored by drinkaware.co.uk. Therefore, the two conditions were purely created for this research.
Research design
In this study, a between-subjects single-factor experimental design has been implemented. Within the experiment, one independent and two dependent factors were researched. On top of that, two moderating variables were investigated. This type of research, an experiment, fits with the research question as it aims to measure the difference between conditions and the effect of one factor on another (Gravetter & Forzano, 2012). The research instrument for this study consists of a survey executed via Qualtrics (see Appendix C).
Before being able to start the survey, participants were asked to agree to the informed consent that was presented to them. This informed consent consisted of disclaimers about anonymity, confidentiality of data and withdrawal of responses. By clicking an orange button to start the survey, they agreed to be informed about the nature and method of the study and to be of the age of eighteen or above.
The second page of the survey consisted of the questions ‘What is your age?’, ‘What is your gender?’ and ‘How much time do you spend daily on YouTube?’, of which the last aimed to measure interest in YouTube. The first was to be filled in by the respondent. For the second, respondents could choose between male and female. For the third question, the seven options were: no time at all, less than an hour a day, 1-2 hours a day, 2-3 hours a day, 3-4 hours a day, 4-5 hours a day and 5 or more hours a day.
After that, respondents were shown the stimulus material. In this, the first variable,
disclosure of sponsorship, was present. Either the control condition, the video without ‘#ad’, or the manipulated condition, the video with ‘#ad’, was shown. To ensure participants
watched the complete video, a timer was set to the amount of time it took to watch the video. Participants could not click a button to further the study until the timer, set to 170 seconds, ran out.
To check whether participants were aware of the manipulation they had received, a manipulation check was used, which consisted of the question ‘Did you see ‘#ad’ in the video you just watched?’. Participants could respond with either ‘Yes’ or ‘No’. Furthermore, to measure knowledge of the spokesperson participants were simply asked whether they knew of the spokesperson. They were also asked to give the name if they knew of her, to ensure the legitimacy of their previous answer.
Moving on, this survey used two dependent variables, namely attitude towards the video and attitude towards the spokesperson. Both were measured with existing scales for attitude. The first was taken from an article by Holbrook and Batra (1987), which measured attitude towards an advertisement. In order to make the scale fit in with this research, in all four of
the questions ‘ad’ was replaced with ‘video’. Other than that, exact replicates of the
questions in that scale were used, as the research instrument shows (see Appendix C). To measure attitude towards spokesperson, ten questions from an attitude scale from Ohanian (1990) were used. Although that scale used three different dimensions with fifteen questions, this research used only the trustworthiness and expertise dimension as they were deemed most relevant. This study disregards the attractiveness dimension, based on the knowledge that physical attractiveness is less important than social attractiveness (Lee & Watkins, 2016), which is represented by trustworthiness and expertise. Both scales were measured on a 7-point Likert scale to facilitate analyses.
For the scale regarding video attitude, respondents were asked to ‘Please indicate how you feel regarding the video’ using four questions. The first ranged from ‘I dislike the video’ to ‘I like the video’, the second from ‘I react favorably to the video’ to ‘I react unfavorably to the video’, the third from ‘I feel positive toward the video’ to ‘I feel negative toward the video’ and the fourth from ‘The video is bad’ to ‘The video is good’. These questions all ranged from 0 to 6, thus constituting a 7-point scale.
The second, measuring spokesperson attitude, was divided into two sections. The first section asked respondents to ‘Please indicate how you feel regarding the spokesperson in the video’. The answer scale ranged from ‘The spokesperson is not at all …’ to ‘The spokesperson is completely …’. The concepts that were to be filled in regarding this scale, were dependable, honest, reliable, sincere and trusthworthy. The second section was constructed in the same way, except here the concepts were expert, experienced,
knowledgable, qualified and skilled.These questions ranged from 0 to 6, which constitutes a 7-point scale.
Lastly, the final page of the survey debriefed respondents and thanked them for
participating. On top of that, there was the opportunity for respondents to leave questions or remarks at the end of the page.
Data collection
Before collecting data for this study, a pretest was executed in order to inquire the
effectiveness of the manipulation and to see whether the survey could be improved. For this, exactly twenty respondents were recruited who were instructed not to participate in the actual study as well to avoid participant bias due to knowledge about the aim of the survey. Participants were recruited via social media, in particular Facebook and Instagram. On both, the researcher posted a plea to friends, fellow students, and followers to fill in the survey used in this research study. Several posts were posted: one on the researchers’ personal
news feed to friends and family, two within two different Facebook groups for people living in the same student flat as the researcher, and one within a group for Communication Science students. The message participants were shown, consisted of the question to fill in a short survey in order for the researcher to finish her bachelors’ degree and contained a link that led to the survey in Qualtrics. On top of that, people close to the researcher, such as friends and family, have shared this message among their friends via social media. Beneficial to this direct method of recruiting is that participants feel more obliged to fill in the survey and pay full attention to it, as it is seen as a favour to the researcher. To facilitate and speed up the recruiting process, this convenience sampling method had been chosen as the preferred method for the collection of data.
Analyses
Analyses were executed in SPSS version 22. For the results of the executed pretest, several analyses were executed. Firstly, descriptive statistics regarding age and sex of the
participants were gathered using frequency tables. Due to a mistake in the design of the survey for the pretest, it was impossible to determine to which group participants were assigned. Therefore, a manipulation check could not be executed. This mistake was solved before activation and distribution of the actual survey.
As with the pretest, descriptive statistics regarding sex and age were gathered for the actual study. Apart from that, the valid response rate was measured and missing values were indicated as such. Lastly, the size of the two groups, the control group and the experimental group, was researched. Besides that, a manipulation check was performed using a Fisher exact test. This test is relevant, as the independent factor, whether participants were exposed to the condition with ‘#ad’ or without, is measured dichotomous. The dependent factor, whether or not the participants have said to have seen the manipulation or not, is also measured dichotomous. On top of that, the analysis aims to measure proportion, which makes this the most relevant option for this case.
Regarding analyses, the main effects of presence or absence of ‘#ad’ in a video on video attitude and on spokesperson attitude were measured the same way. Both were measured using a t-test for two means, using a significance of p ≤ 0,05. This analysis was chosen, because it measures the effect of a dichotomous variable, the disclosure of sponsorship, on a dependent variable measured at interval level, attitude towards the video and attitude towards the spokesperson. On top of that, there was a total number of participants higher than thirty per measure, which qualifies a t-test as method.
Lastly, the moderating effects of knowledge of spokesperson and of interest in YouTube were both measured using a multiple way variance analysis. This method is relevant, as the
independent variables, disclosure of sponsorship, knowledge of spokesperson and interest in YouTube, are all used as categorical variables. The dependent variables, video attitude and spokesperson attitude, are both measured at interval level. Whilst interest in YouTube was not measured at as a categorical variable in the study, it was transformed to be one during the analyses to facilitate the distinction between different groups. This variable was divided based on convenience between respondents with little to no interest, which included daily time spent on YouTube of less than one hour, and respondents with some to a lot of interest, which included daily time spent on YouTube of one or more hours.
Results
During the pretest 20 respondents participated. Of these, nine (45%) were male and 11 (55%) were female. The average age of these 20 respondents was 29.35 (SD = 15.12), with a minimum of 18 and a maximum of 61, bringing the range to 43.
Of the 118 respondents that started the actual survey only 68 completed it, bringing the response rate to 57.63%. The majority of these participants (n = 57) were female (83.8%), with just 11 participants being male (16.2%). Through Qualtrics respondents were randomly assigned to one of the two conditions. Of the completed surveys, 54.4% (n=37) was
assigned to the control condition, the video without ‘#ad’, and the remaining 45.6% (n = 31) was assigned to the manipulated condition, the video with ‘#ad’. The age amongst the survey population ranged from 18 to 80 with a mean of 28.60 (SD = 14.91).
The existing scales that were deemed reliable in previous literature, were shown to be reliable in this research as well. The first scale measured the attitude towards the video and showed an Eigenvalue of 2.90, which explained 72.52% of the total variance. It was deemed reliable, as the Cronbachs alpha for it was .87 (M = 15.34, SD = 4.49). The second scale, which measured attitude towards the spokesperson, was divided into two separate scales after a factor analysis proved that the factors could not be united into one component. The first of these, which measured attitude towards the spokesperson regarding trustworthiness had an Eigenvalue of 3.69 and explained 73.75% of the total variance. Its reliability score made it a reliable scale to be used in further analyses (α = .91, M = 17.00, SD = 6.41). The last scale that was used in analyses was one that measured attitude towards the
spokesperson regarding expertise. This scale had an Eigenvalue of 3.49 and explained 69.84% of the total variance. Once again, this scale was reliable and could be used for further analyses (α = .89, M = 11.10, SD = 5.95).
Before analyses were carried out, a manipulation check was executed. According to this, six respondents reported they had not seen ‘#ad’ although they were assigned to the ‘#ad’
condition (19.35%). Of the people that were assigned to the condition without disclosure of sponsorship, only one person reported to have seen ‘#ad’ (2.70%). The association
measure, χ², turned out to be significant (χ² = 40.15, p < .001).
After this, the main effects and interaction effects were tested, in line with the formulated hypotheses. Regarding video attitude, respondents in the control condition scored a slightly higher mean score (M = 3.93, SD = 1.14) than respondents in the manipulated condition (M = 3.73, SD = 1.11). This effect, however, showed to be of no significance, t(66) = .73, p = .469, CI 95% [-.35;.75], with equal variances assumed as Levene’s Test for Equality of Variances was not significant (F(66) = .05, p = .829). For spokesperson attitude regarding trustworthiness respondents in the control condition also scored a little higher (M = 3.65, SD = 1.34) than respondents in the manipulated condition (M = 3.10, SD = 1.16), but this once again was of no statistical significance, t(66) = 1.78, p = .080, CI 95% [-.07;1.16], again assuming equal variances as Levene’s Test for Equality of Variances was not significant (F(66) = .09, p = .770). For spokesperson attitude regarding expertise, respondents in the control condition scored slightly higher (M = 2.36, SD = 1.17) than respondents in the
manipulated condition (M = 2.05, SD = 1.22). In line with the previous results, this effect was not significant, t(66) = 1.07, p = .287, CI 95% [-.27;.89] assuming equal variances as
Levene’s Test for Equality of Variances was not significant (F(66) = .03, p = .874). The first interaction effect that was tested was that of interest in YouTube. To be able to compare different groups, the variable that measured interest in YouTube in hours per day spent on the site was categorized into two groups. The first is ‘little/no interest’, which consisted of respondents that reported to spend ‘no time at all’ to ‘less than an hour a day’ watching YouTube. This was a group of 46 respondents, making up 67.65% of the survey population. The other group, ‘some/a lot of interest’, was made up of respondents who reported to spend at least one hour a day on the website, which consisted of 22 respondents (32.35%).
The interaction effect of interest in YouTube with disclosure of sponsorship on attitude towards the video was measured using the variable with these two groups. According to the analyses, respondents with some/a lot of interest in YouTube scored higher on video attitude (M = 4.30, SD = 1.04) in the control condition than respondents with no/little interest in YouTube (M = 3.77, SD = 1.16). Likewise, respondents with some/a lot of interest in
YouTube scored higher on video attitude (M = 4.09, SD = 1.12) in the manipulated condition than respondents with no/little interest in YouTube (M = 3.53, SD = 1.09). This interaction effect turned out to be insignificant, F(1,67) = .01, p = .946.
Similar effects were found for the interaction of interest in YouTube with disclosure of
sponsorship on attitude towards spokesperson both regarding trustworthiness and expertise. Respondents with some/a lot of interest in YouTube scored higher (M = 4.16, SD = .97) than respondents with no/little interest in YouTube (M = 3.43, SD = 1.43) regarding
trustworthiness in the control condition, as well as in the manipulated condition (M = 2.91, SD = 1.11; M = 3.21, SD = 1.21). This effect was not significant, F(1,67) = 2.52, p = .117. Respondents with some/a lot of interest in YouTube scored higher (M = 2.65, SD = .80) than respondents with no/little interest in YouTube (M = 2.24, SD = 1.28) regarding expertise in the control condition, as well as in the manipulated condition (M = 2.11, SD = 1.10; M = 2.02, SD = 1.30). This effect was also not significant, F(1,67) = .28, p = .601.
This trend continues for the interaction effect of knowledge of spokesperson in the video with disclosure of sponsorship on attitudes towards the video and the spokesperson. Here, however, it is of importance to mention the size of groups within the variable that measures whether a respondent knew of the spokesperson or not. Only three respondents (4.41%) reported to know of the spokesperson Hannah Witton. Respondents who knew the
YouTuber scored higher on video attitude (M = 4.13, SD = 1.24) than respondents who did not know her (M = 3.91, SD = 1.15) in the control condition, as well as in the manipulated condition (M = 5.5, SD = 0; M = 3.67, SD = 1.08). This effect was not significant, F(1,67) = 1.34, p = .251. Respondents who knew of Hannah Witton scored higher regarding
trustworthiness (M = 3.90, SD = .42) than respondents who did not know of her (M = 3.63, SD = 1.37) in the control condition, as well as in the manipulated condition (M = 4.80, SD = 0; M = 3.05, SD = 1.14). Once again, this effect is not statistically significant, F(1,67) = .89, p = .349. Respondents who knew of the YouTuber also scored higher on expertise (M = 3.80, SD = .28) than respondents who did not know of her (M = 2.28, SD = 1.14) in the control condition, as well as in the manipulated condition (M = 4.40, SD = 0; M = 1.97, SD = 1.15). This effect was not significant, F(1,67) = .41, p = .527.
Discussion
Conclusions
All outcomes of the analyses turned out to not be significant. The main effects, which
measure the effect of disclosure of sponsorship on video attitude and spokesperson attitude, are not statistically significant and the first two hypotheses can therefore not be accepted. Results also show that interaction effects of disclosure of sponsorship and interest in YouTube on both video and spokesperson attitude were not significant. Hypotheses 3a, 3b, 3c and 3d thus can be refuted. Similarly, interaction effects of disclosure of sponsorship and
knowledge of the YouTuber on both video and spokesperson attitude were not significant. Hypotheses 4a, 4b, 4c and 4d can also be refuted.
Strengths and limitations
As all research, this study knows both strengths and limitations, which are important to recognize for future research and in order to place this study in its context. An important factor that influences the outcomes’ generalizability is the sample size. Whilst 118
respondents initially started the survey, only 68 actually finished it and could be taken into account regarding analyses. This is a small number in comparison to the population it aims to represent. On the other hand, this study does include a wide range of ages, which corresponds to the population it aims to represent. However, the ratio between men and women is unequal due to convenience sampling. Under the right conditions regarding time and reach, a more representative sample could be found in the future.
Since a large proportion of respondents did not complete the survey, the attrition rate was high. Qualtrics data showed most of these respondents stopped on the page where the video was displayed. This could be to do with the timer that was set on it. A possible
explanation for this is that people skipped ahead in the video and misunderstood they had to wait before being able to continue the survey. A solution would be to explicitly disclose to respondents that the timer lasted as long as the video and that the video therefore had to be watched completely.
The internal validity of the survey is at stake due to the inequality of groups within the analyses. Whilst one moderator aimed to compare viewers who knew of the YouTuber, Hannah Witton, and viewers who did not, the group of viewers who did know of her consisted of a mere three people. Such a small group could never represent the actual group of people who know of her, for which her subscriber count is an indicator. Precaution was taken with regard to this by asking the YouTuber to share the survey among followers. Unfortunately, she did not answer to this request and further attempts were not made. Another issue that influences the internal validity is the measurement for interest in
YouTube. Since literature did not provide a method for measuring this, it was measured as time spent on the medium. This, however, does not necessarily correlate to the interest in YouTube as someone who spends little time on the medium could still have a high interest for YouTube and vice versa.
Lastly, a limitation to this study, in particular to the survey, is that the disclosure of sponsorship was perhaps not salient enough. A few respondents noted at the end of the survey that they could have missed the ‘#ad’ in the video as they were watching the video on their mobile phone, which did not show the whole width of the video in portrait mode. It is
therefore advisable to either disclose to only fill in the survey in landscape mode or on a tablet or computer, or to make the sponsorship disclosure more salient.
Implications
Attached to the outcomes of this study are some implications, with regard to science, YouTube content creators and advertisers. Concerning the first, science, this study serves as a contribution to research. Whereas current literature mainly focuses on ‘older’ online and offline media, such as blogs and television, regarding the effects of disclosure of
sponsorship (Boerman, 2014; Hwang & Jeong, 2016), this study focuses on a newer online platform, namely YouTube. This is of importance, as the online world is rapidly developing and new platforms are continuously surfacing.
This study shows that disclosure of sponsorship has no significant effect on attitude towards the video and the spokesperson in the video. This is important information for both
YouTubers and advertisers. For YouTubers it means that disclosed sponsorship does not necessarily negatively influence their audience. It is therefore advisable to comply with guidelines regarding this, that state that it is expected from online content creators to disclose sponsorship. Although attitude towards the brand or advertisement within the sponsored content is not explicitly measured, the same effect is expected. For advertisers it is therefore also recommended to comply with guidelines concerning this issue.
Suggestions for future research
Elaborating on the importance of focusing on newer online media is the first suggestion for future research. This is to study disclosure of sponsorship and its effects on media popular at the moment of research. Examples are Snapchat and Instagram, platforms on which quite some sponsored content can be found, which could potentially influence users differently across different platforms. It is therefore suggested that more elaborate research is done on more types of media, so results for different platforms can be compared.
Secondly, this study focused on one way of disclosing sponsorship, namely through adding ‘#ad’ during a video. However, there are many more ways to disclose sponsorship (Hwang & Jeong, 2016). It could be interesting to research whether different ways of disclosure, for example verbally versus textually, have different effects regarding attitudes. Another thing that could be measured is the viewers’ intention to (re)visit a channel connected to a video in which sponsorship is disclosed versus one in which it is not disclosed.
Whilst this experiment was a quantatative study, a qualitative research regarding the topic could also be interesting. It could be researched why videos either with or without disclosed sponsorship are liked or disliked. Insights into this could help YouTubers and advertisers behind their decision to fully disclose sponsorship or not.
Lastly, a concrete method for measuring participants’ interest in YouTube could not have been found in literature. It might be useful to have such a measurement for future research regarding the online platform.
References
Andrejevic, M. (2009). Exploiting YouTube: Contradictions of user-generated labor. In P. Snickars & P. Vonderau (Eds.), The YouTube Reader (pp. 406-423). Retrieved from http://www.youtubereader.com/images/youtubereader.pdf
Becker-Olsen, K. L. (2003). And now, a word from our sponsor--a look at the effects of sponsored content and banner advertising. Journal of Advertising, 32, 17-32.
doi:10.1080/00913367.2003.10639130
Beuker, R., & Abbing, E. R. (2010). Two faces of social media: Brand communication and brand research. Design Management Review, 21, 54-60.
doi:10.1111/j.1948-7169.2010.00051.x
Boerman, S. C. (2014). "This program contains product placement": Effects of sponsorship disclosure on television viewers’ responses (Doctoral dissertation, University of Amsterdam, Amsterdam, the Netherlands). Retrieved from
https://pure.uva.nl/ws/files/1965589/138870_thesis.pdf
Boerman, S. C., van Reijmersdal, E. A., & Neijens, P. C. (2012). Sponsorship disclosure: Effects of duration on persuasion knowledge and brand responses. Journal of
Communication, 62, 1047-1064. doi:10.1111/j.1460-2466.2012.01677.x
Brna, P. M., Dooley, J. M., Esser, M. J., Perry, M. S., & Gordon, K. E. (2013). Are YouTube seizure videos misleading? Neurologists do not always agree. Epilepsy & Behavior, 29, 305-307. doi:10.1016/j.yebeh.2013.08.003
Cha, M., Kwak, H., Rodriguez, P., Ahn, Y. Y., & Moon, S. (2007). I tube, you tube, everybody tubes: Analyzing the world's largest user generated content video system. Proceedings of the 7th ACM SIGCOMM conference on Internet measurement, USA, 1-14.
doi:10.1145/1298306.1298309
Colliander, J., & Erlandsson, S. (2015). The blog and the bountiful: Exploring the effects of disguised product placement on blogs that are revealed by a third party. Journal of
Marketing Communications, 21, 110-124. doi:10.1080/13527266.2012.730543
Delli, K., Livas, C., Vissink, A., & Spijkervet, F. K. (2016). Is YouTube useful as a source of information for Sjögren's syndrome? Oral diseases, 22, 196-201. doi:10.1111/odi.12404
Freeman, B., & Chapman, S. (2007). Is “YouTube” telling or selling you something? Tobacco content on the YouTube video-sharing website. Tobacco Control, 16, 207-210.
doi:10.1136/tc.2007.020024
Gravetter, F. J. & Forzano, L. B. (2012). Research methods for the behavorial sciences. Hampshire, England: Cengage Learning EMEA.
Haymes, A. T., & Harries, V. (2016). 'How to stop a nosebleed': An assessment of the quality of epistaxis treatment advice on YouTube. The Journal of laryngology and otology, 130, 749-754. doi:10.1017/S0022215116008410
Hellinga, T. (2016). Literature review on the quality of health-related information on YouTube. Unpublished manuscript, Department of Communication Science, University of Amsterdam, Amsterdam, The Netherlands.
Holbrook, M. B. & Batra, R. (1987). Assessing the role of emotions as mediators of consumer responses to advertising. Journal of consumer research, 14(3), 404-420. Retrieved from http://www.jstor.org/stable/pdf/2489501.pdf
Hudson, R. A., & Rosen, H. (1953). On the definition of attitude: Norms, perceptions, and evaluations. Public Opinion Quarterly, 17, 141-146. doi:10.1086/266444
Hwang, Y., & Jeong, S. H. (2016). “This is a sponsored blog post, but all opinions are my own”: The effects of sponsorship disclosure on responses to sponsored blog
posts. Computers in Human Behavior, 62, 528-535. doi:10.1016/j.chb.2016.04.026 Lee, J. E., & Watkins, B. (2016). YouTube vloggers' influence on consumer luxury brand perceptions and intentions. Journal of Business Research, 69, 5753-5760.
doi:10.1016/j.jbusres.2016.04.171
Ohanian, R. (1990). Construction and validation of a scale to measure celebrity endorsers' perceived expertise, trustworthiness, and attractiveness. Journal of advertising, 19, 39-52. doi:10.1080/00913367.1990.10673191
Pace, S. (2008). YouTube: An opportunity for consumer narrative analysis? Qualitative Market Research: An International Journal, 11, 213-226. doi:10.1108/13522750810864459 Postigo, H. (2016). The socio-technical architecture of digital labor: Converting play into YouTube money. Media & Society, 18, 332-349. doi:10.1177/1461444814541527
Smith, A. N., Fischer, E., & Yongjian, C. (2012). How does brand-related user-generated content differ across YouTube, Facebook, and Twitter? Journal of Interactive Marketing, 26, 102-113. doi:10.1016/j.intmar.2012.01.002
Stichting Reclame Code (2009). Reclamecode Social Media (RSM). [Web page]. Retrieved from https://www.reclamecode.nl/nrc/pagina.asp?paginaID=289%20&deel=2
van Dijck, J. (2009). Users like you? Theorizing agency in user-generated content. Media, culture, and society, 31, 41-58. doi:10.1177/0163443708098245
van Reijmersdal, E. A., Fransen, M. L., van Noort, G., Opree, S. J., Vandeberg, L., Reusch, S., van Lieshout, F. & Boerman, S. C. (2016). Effects of disclosing sponsored content in blogs how the use of resistance strategies mediates effects on persuasion. American Behavioral Scientist, 60, 1458-1474. doi:10.1177/0002764216660141
Witton, H. (2016, January 26). Drinking Responsibly | Hannah Witton [Video file]. Retrieved from https://www.youtube.com/watch?v=XrwPfbagxEk
Youtube (n.d.). About YouTube [Web page]. Retrieved from https://www.youtube.com/yt/about/
Appendices
Appendix A: Video references
Control condition: Hellinga, T. (2016, October 15). Drinking responsibly [Video file]. Retrieved from https://www.youtube.com/watch?v=GRIPlRGB7KU
Manipulated condition: Hellinga, T. (2016, October 15). Drinking responsibly [Video file]. Retrieved from https://www.youtube.com/watch?v=gfOccunu2JY&t=1s
