Page 1 MKTG204 Integrated Marketing Communications, Session 1 2017 Consumer insights survey results for Task 2B 1. About this document The following document has been prepared for students enrolled in MKTG204 during Session 1, 2017 at Macquarie University. This document provides students with the information required to complete Task 2B (Consumer insights report). The following results come from selected data, collected by MKTG204 students from the Assessment Task 2A (Consumer insights data collection). Please refer to the Assessment Criteria document for details on the Task 2B assessment and its marking rubric. The Assessment Criteria document and the template for completing Assessment Task 2B can both be found on iLearn. 2. Copyright This document is a copyright material. Reproduction or sharing of this material without obtaining prior permission from the unit convenor and deputy convenor is strictly prohibited. 3. Research aim, questions and significance This research aims to gain insights into the influence of social media posts (i.e. Instagram) with branded information about dining and lifestyle on consumer responses and marketing communications outcomes. This research asks: 1) if there is any effects of Instagram posts on consumer responses and marketing communications outcomes depending on Instagrammer sexes, message framing types and consumer segments; 2) what the role of envy is in social media marketing. This research provides practical implications on creative strategies/executions of marketing communications campaigns for restaurant and lifestyle product/services. 4. Research method This research employed a 2 factor (male vs. female Instagrammer) x 2 factor (male vs. female consumer) x 2 factor (neutral vs. bragging message framings) between-subject experimental online survey (see these visual ad stimuli in Figure 1 below). This means each participant saw one Instagram post in a randomised fashion. Before exposure, all participants rated their hunger. All participants were told to view an Instagram post of a food blogger and see if the experience is desirable and that they could view it as long as they wished just like how they viewed Instagram posts. After exposure, all participants were asked the same questions to investigate the influence of visual/verbal cues before providing personal information. Page 2 Figure 1. Visual stimuli used in the experiment and numbers of participants who were shown each experimental condition (in a randomised fashion) Message framing types 1a. Neutral message (n = 49) 1b. Bragging message (n = 57) Instagrammer’s sexes Female Instagrammer (n = 97) 2a. Neutral message (n = 50) 2b. Bragging message (n = 53) Male Instagrammer (n = 103) Note: The sizes of the Instagram posts are reduced here for a layout purpose. Larger sizes of these visual stimuli are provided in the Appendix. One participant only saw one image (randomly displayed). The number of n in the bracket indicates the total number of participants who saw the image or factor. For example, Instagram post 1a was shown to 49 participants in total and Instagram posts 1a and 1b (i.e. Female Instagrammer’s posts) were shown to 97 participants in total. Page 3 5. Key constructs, definitions and operationalisations Table 1. Key constructs investigated in this article with definitions and operationalisations explained Key constructs Definitions Operationalisations* Instagrammer’s sexes Depiction of one male vs. one female Instagrammer on Instagram posts This is an experimental factor. Similar fictitious name were used: Daniel/Danielle Ford. Message framing types Depiction of two different messages on Instagram posts This is an experimental factor: Neutral message vs. bragging message (i.e. excessively proud and boastful talk) about one’s dining experience/a lifestyle Consumer sexes Individuals’ sexes identified by participants This is an experimental factor: Men vs. Women Passion about foods Consumer evaluation of their love of foods Food is ……… my passion “0 Not at all”, “1 Not very much”, “2 Somewhat”, “3 Very much”, “4 Totally” Foodies (See note below) Consumers who have passion about foods above the average. Non-foodies = 0 Foodies = 1 Brand attitude (See note below) Consumer overall evaluation of Alain Ducasse Restaurant After viewing the Instagram post, how good or bad do you think the Alain Ducasse Restaurant is?: “-2 Very bad”, “-1 Bad”, “0 Neither good nor bad”, “1 Good”, “2 Very good” Instagram post attitude (See note below) Consumer overall evaluation of the Instagram post. How much do you like or dislike the ad?: “-2 Dislike very much”, “-1 Dislike”, “0 Neither like nor dislike”, “1 Like”, “2 Like very much” Envy Consumer evaluation of their jealousy towards the Instagrammer While viewing this Instagram post, did you initially feel a sense of envy (or jealousy) towards this person? “0 Not at all credible”, “1 Slightly”, “2 Moderately”, “3 Very”, “4 Extremely” Source liking Consumer evaluation of the overall likability of the Instagrammer How much do you like or dislike this person? “-2 Dislike very much”, “-1 Dislike”, “0 Neither like nor dislike”, “1 Like”, “2 Like very much” Hunger Consumer subjective evaluation of feeling deprived of food How hungry are you feeling right now? “0 not at all hungry”, “1 Slightly hungry”, “2 Moderately hungry”, “3 Very hungry”, “4 Extremely hungry” Note: Index of foodie vs. non-foodie consumers is derived from calculating the mean value of passion about foods. If consumers rate their passion about foods above the mean, then they are defined as foodies (i.e. recoded to a dummy variable, 1 otherwise 0). Brand attitude is a key marketing communications outcome. Instagram post attitude in social media communication here is similar to Ad attitude in advertising communication. Page 4 6. Quality control before data collection We ensured the quality of data prior to data collection by following Podsakoff, MacKenzie, Lee & Podsakoff’s (2003) procedural remedies. We counterbalanced question order where it would not disrupt the flow of the survey. We used both radio button and slider scale formats in a randomized fashion. Rotated scale options were used and randomized. In addition, all scale points were verbally and numerically anchored to reduce response biases of scales anchored only at endpoints among some respondents who may exhibit extreme response style (Dolnicar & Grün, 2007). 7. Quality control after data collection In this survey, 200 participants (51% males) were included in the analyses (excluding participants who used mobile devices for controlling the effect of screen size). Participants’ age ranged from 18 to 59. Note: More participants participated in Task 1A data collection. However, only the above number of participants was selected here to demonstrate the effect and the role of envy in social media marketing. We provide statistical tests where we think it is useful for students to generalize the findings to the target population e.g., using a bootstrap sampling method. This is also to simply statistical analysis techniques required so that it is not too complex for students to understand. 8. How to cite this document Pitt, J., Singh, C., and Ang, L. (2017). MKTG204 Investigating the effects of social media posts with branded information on consumer responses: Assessment Task 2B, Session 1, 2017 Consumer insights survey results. North Ryde: Macquarie University. Note: There is no need to include the tables/graphs within this document onto your report. Simply make a reference e.g., Pitt et al. (2017, table x or figure x, p. x) for in-text references and provide the full reference in your References list. Only include tables, graphs or figures if they come from your own synthesis. 9. Analyses and results There are two parts in this section: Part 1. Frequency statistics on participants’ ages/age ranges, passion about foods and foodies and brand attitude; and Part 2. Experimental study to test the effects of visual and verbal elements and consumer segments on their responses. Page 5 Part 1: Frequency statistics 9.1. Participants’ ages Participants’ ages range from 18 to 59 years. However, most participants’ ages range from 20 to 22 years. Figure 2 below shows a summary of participants’ ages in percentage. Figure 2. A summary of participants’ ages in percentage Note: We also grouped the participants into two age groups: 18 to 25 vs. 26 to 59 years to simplify a comparison in further analyses. Page 6 9.2. Passion about foods On average, participants are somewhat passionate about foods (Mean = 2.35). Figure 2 shows a summary of participants’ passion about foods. Table 2 shows summary frequency statistics with a bootstrapping sampling of 5000 samples at 95% confidence interval to generalize the sampled findings to the target population. Figure 2. A summary of participants’ passion about foods in percentage Table 2. Summary frequency statistics on passion about foods with a bootstrapping sampling of 5000 samples at 95% confidence interval Passion about foods Frequency Percent Valid Percent Cumulative Percent Bootstrap for Percent Bias Std. Error BCa 95% Confidence Interval Lower Upper Magnitude of passion about foods 0 7 3.5 3.5 3.5 0.0 1.3 1.5 5.5 1 33 16.5 16.5 20.0 0.0 2.6 12.0 21.0 2 69 34.5 34.5 54.5 0.1 3.4 28.5 41.0 3 66 33.0 33.0 87.5 0.0 3.4 27.5 39.0 4 25 12.5 12.5 100.0 0.0 2.3 8.5 16.5 Total 200 100.0 100.0 0.0 0.0 Page 7 Note: Bootstrapping is a technique from which the sampling distribution of a statistic is estimated by taking repeated samples from the data set. From this, confidence intervals and significance tests can be computed. There is a difference between the percent column vs. the BCa 95% Confidence Interval column. The percent column shows the percentages of sampled participants. Whereas, the BCa95% Confidence Interval column shows the estimated percentage ranges for the entire target population. 9.3. Foodie consumers Among the surveyed participants, approx. 46% are foodies, determined by their passion about foods. Figure 3 shows a comparison of non-foodie vs. foodie participants in percentage. Figure 2. A summary of non-foodie vs. foodie consumers in percentage 9.4. Overall brand attitude Table 3 shows a summary frequency statistics of participants’ evaluation of Alain Ducasse Restaurant overall after seeing the Instagram post (regardless of experimental conditions) with a bootstrapping sampling of 5000 samples at 95% confidence interval to generalize the sampled findings to the target population. Page 8 Table 3. A summary frequency statistics on brand attitude with a bootstrapping sampling of 5000 samples at 95% confidence interval Brand attitude Frequency Percent Valid Percent Cumulative Percent Bootstrap for Percenta Bias Std. Error BCa 95% Confidence Interval Lower Upper Magnitude of brand attitude rating -2 1 0.5 0.5 0.5 0.0 0.5 .0b 1.5 -1 3 1.5 1.5 2.0 0.0 0.9 .0b 3.5 0 53 26.5 26.5 28.5 -0.1 3.1 21.5 32.0 1 78 39.0 39.0 67.5 0.1 3.4 32.5 45.5 2 65 32.5 32.5 100.0 0.0 3.3 27.0 38.5 Total 200 100.0 100.0 0.0 0.0 a. Unless otherwise noted, bootstrap results are based on 5000 bootstrap samples b. Some results could not be computed from jackknife samples, so this confidence interval is computed by the percentile method rather than the BCa method. Note: Bootstrapping is a technique from which the sampling distribution of a statistic is estimated by taking repeated samples from the data set. From this, confidence intervals and significance tests can be computed. There is a difference between the percent column vs. the BCa 95% Confidence Interval column. The percent column shows the percentages of sampled participants. Whereas, the BCa95% Confidence Interval column shows the estimated percentage ranges for the entire target population. Part 2: Experimental study 9.5. The effects message framing types, Instagrammer’ sexes and consumer segments by sexes, age groups and non-foodie and foodie consumer groups Next, we began investigating the effect of message framing types, Instagrammer’ sexes and consumer segments by sexes, age groups and non-foodie and foodie consumer responses. We conducted a series of Independent samples t-tests comparing consumer responses (as test variables) between these grouping variables (one at a time). The results showed that message framing types, Instagrammer’ sexes and consumer segment by age groups did not significantly effect envy, Instagram post attitude, brand attitude and source liking (all ps > 0.05). However, women rated the Instagram posts and Alain Ducasse Restaurant (regardless of Instagram post conditions) more positively than men (Mdiff. = 0.35, t(198) = 2.56, p = 0.01; and Mdiff. = 0.37, t(198) = 3.29, p = 0.002, in the respective order). In addition, foodie consumers felt more envy towards the Instagrammers than non-foodie consumers (Mdiff. = 0.36, t(198) = 2.28, p = 0.024). The mean values are provided in Tables 4, 5, 6, 7 and 8 with significant different test results highlighted. We also investigated if the effects of these grouping variables are dependent on each other and the results showed non-significant interaction effects. Hence, detailed mean values of overlapped groupings are not provided to keep this document concise. Page 9 Table 4. Mean values and significance test of the mean differences of each construct by message framing types Construct Neutral message framing (Instagram posts 1a and 2a) Bragging message framing (Instagram posts 1b and 2b) Envy 0.86 0.84 Instagram post attitude 0.31 0.41 Brand attitude 0.99 1.04 Source liking 0.06 0.09 *Significant at p <= 0.05 Note: Mean is the measure of central tendency. It is the average score, which is calculated by adding up all of the scores from all sampled participants and then divided by the total number of sampled participants (n). Mean difference is the difference between two mean values, which is calculated by Mean a minus Mean b. t is a test statistic to test whether the two mean values are significantly different from zero (in this context). When the t value is + or -1.96, it means the difference is significant at 95% Confidence interval. p is a test statistic to test how it is evidently weak or strong to reject the null hypothesis (i.e. in this context the null hypothesis is that Mean a – Mean b = 0 or no difference). When the p value is smaller than or equal to 0.05 (i.e. <= 0.05), it means there is strong evidence against the null hypothesis (i.e. there is a significant difference). When the p value is larger than 0.05 (i.e. >0.05), it means there is a weak evidence to reject the null hypothesis (i.e. there is no significant difference). If the p value ranges from 0.051 to 0.100, it means there is a moderate evidence to reject the null hypothesis (i.e. the difference is marginally significant). Page 10 Table 5. Mean values and significance test of the mean differences of each construct by Instagrammer’s sex depicted in the posts Construct Female Instagrammer (Instagram posts 1a and 1b) Male Instagrammer (Instagram posts 2a and 2b) Envy 1.12 1.09 Instagram post attitude 0.28 0.44 Brand attitude 1.05 0.98 Source liking 0.82 0.74 *Significant at p <= 0.05 Note: Mean is the measure of central tendency. It is the average score, which is calculated by adding up all of the scores from all sampled participants and then divided by the total number of sampled participants (n). Mean difference is the difference between two mean values, which is calculated by Mean a minus Mean b. t is a test statistic to test whether the two mean values are significantly different from zero (in this context). When the t value is + or -1.96, it means the difference is significant at 95% Confidence interval. p is a test statistic to test how it is evidently weak or strong to reject the null hypothesis (i.e. in this context the null hypothesis is that Mean a – Mean b = 0 or no difference). When the p value is smaller than or equal to 0.05 (i.e. <= 0.05), it means there is strong evidence against the null hypothesis (i.e. there is a significant difference). When the p value is larger than 0.05 (i.e. >0.05), it means there is a weak evidence to reject the null hypothesis (i.e. there is no significant difference). If the p value ranges from 0.051 to 0.100, it means there is a moderate evidence to reject the null hypothesis (i.e. the difference is marginally significant). Page 11 Table 6. Mean values and significance test of the mean differences of each construct by consumers’ sex (regardless of Instagram post conditions) Construct Men Women Envy 0.75 0.96 Instagram post attitude 0.19* 0.54* Brand attitude 0.83* 1.20* Source liking -0.03 0.18 Table 7. Mean values and significance test of the mean differences of each construct by age grouping (regardless of Instagram post conditions) Construct 18 to 25 Years 25 to 60 years Envy 0.91 0.72 Instagram post attitude 0.36 0.37 Brand attitude 0.99 1.07 Source liking 0.05 0.13 Table 8. Mean values and significance test of the mean differences of each construct by non-foodie and foodie consumers (regardless of Instagram post conditions) Construct Non-foodie consumers Foodie consumers Envy 0.69* 1.04* Instagram post attitude 0.32 0.41 Brand attitude 1.06 0.97 Source liking 0.08 0.07 *Significant at p <= 0.05 Note: Mean is the measure of central tendency. It is the average score, which is calculated by adding up all of the scores from all sampled participants and then divided by the total number of sampled participants (n). Mean difference is the difference between two mean values, which is calculated by Mean a minus Mean b. t is a test statistic to test whether the two mean values are significantly different from zero (in this context). When the t value is + or -1.96, it means the difference is significant at 95% Confidence interval. p is a test statistic to test how it is evidently weak or strong to reject the null hypothesis (i.e. in this context the null hypothesis is that Mean a – Mean b = 0 or no difference). When the p value is smaller than or equal to 0.05 (i.e. <= 0.05), it means there is strong evidence against the null hypothesis (i.e. there is a significant difference). When the p value is larger than 0.05 (i.e. >0.05), it means there is a weak evidence to reject the null hypothesis (i.e. there is no significant difference). If the p value ranges from 0.051 to 0.100, it means there is a moderate evidence to reject the null hypothesis (i.e. the difference is marginally significant). Page 12 9.6. The influence of Envy, Instagram post attitude, Envy x Instagram post attitude, Source liking and Hunger on Brand attitude Next, we conducted a series of multiple regression analyses with ENTER approach. First, we loaded brand attitude as the dependent variable and all other variables were loaded as predictors in the order appeared above. These predictors included the interaction between Envy x Instagram post attitude. This is to see if the influence of envy on brand attitude is dependent on Instagram post attitude. The results showed that the interaction term, source liking and hunger did not significantly influence brand attitude (β = 0.045, -0.023, -0.002 respectively, all ps > 0.05). However, Instagram post attitude was found to significantly influence brand attitude (β = 0.298, p < 0.001) and envy was found to have a marginally significant influence on brand attitude (β = 0.093, p = 0.08). Then, we removed the interaction term, source liking and hunger from the predictors list to free the parameters and reran the model again. The results showed that both envy and Instagram post attitude significantly influenced brand attitude. Table 9 shows the multiple regression results. The model showed that it significantly explained the variation in Brand attitude (R2 = approx. 0.20, p < 0.001). Table 9. The influence of envy and Instagram post attitude on brand attitude (All participants) Model Unstandardized Coefficients Standardized Coefficients t Sig. B Std. Error Beta (Constant) 0.801 0.069 11.672 0.000 Envy 0.113 0.049 0.150 2.315 0.022 Instagram post attitude 0.328 0.055 0.389 6.004 0.000 Note: Multiple regression is a model in which an outcome (i.e. Brand attitude (Y) in this context) is predicted by a linear combination of two or more predictor variables i.e. Envy (X1) and Instagram post attitude (X2). That is: Yi = b0 + b1X1i + b2X2i + … + bnXni) + Ɛi. B is an unstandardized value of coefficient of determination or the proportion of variance in the outcome variable (i.e. Brand attitude in this context) explained by a predictor variable. For example, the B value for Envy  Brand attitude is 0.113. This means when Envy increases by one unit (i.e. scale point), it is estimated that Brand attitude would increase by 0.113. t and p are test statistics – in this multiple regression context, the values show whether the B value is significantly different from zero. For example, the p value for Envy  Brand attitude is 0.022. This means there is a strong evidence to reject the null hypothesis (i.e. B ≠ 0). Put simply, Envy could significantly predict Brand attitude when the other predictors were simultaneously considered. 9.7. The influence of Envy and Instagram post attitude on Brand attitude comparing between nonfoodie vs. foodie consumer groups Finally, we reran the last multiple regression model comparing the influence of envy and Instagram post attitude on brand attitude between non-foodie and foodie consumers. This is because the earlier analysis shows that foodie consumers felt more envy towards the Instagrammers. Hence, it is anticipated that the influence of envy on brand attitude may be different depending on the consumer segment. Table 10 shows the multiple regression results by the different consumer segments. Page 13 Table 10. The influence of envy and Instagram post attitude on brand attitude (By non-foodie vs. foodie consumer segments) Model Unstandardized Coefficients Standardized Coefficients t Sig. B Std. Error Beta Non foodies Foodies Non foodies Foodies Non foodies Foodies Non foodies Foodies Non foodies Foodies (Constant) 0.885 0.673 0.087 0.111 10.145 6.062 0.000 0.000 Envy 0.099 0.147 0.071 0.069 0.124 0.202 1.389 2.138 0.168 0.035 Instagram post attitude 0.318 0.345 0.074 0.081 0.383 0.404 4.271 4.271 0.000 0.000 Note: Multiple regression is a model in which an outcome (i.e. Brand attitude (Y) in this context) is predicted by a linear combination of two or more predictor variables i.e. Envy (X1) and Instagram post attitude (X2). That is: Yi = b0 + b1X1i + b2X2i + … + bnXni) + Ɛi. B is an unstandardized value of coefficient of determination or the proportion of variance in the outcome variable (i.e. Brand attitude in this context) explained by a predictor variable. For example, the B value for Envy  Brand attitude for foodie consumers is 0.147. This means when Envy increases by one unit (i.e. scale point), it is estimated that Brand attitude would increase by 0.147 among foodie consumers. t and p are test statistics – in this multiple regression context, the values show whether the B value is significantly different from zero. For example, the p value for Envy  Brand attitude for foodie consumers is 0.035. This means there is a strong evidence to reject the null hypothesis (i.e. B ≠ 0). Put simply, Envy could significantly predict Brand attitude when the other predictors were simultaneously considered among foodie consumers. 9.8. Questions to help students reflect and synthesize these findings The following questions are designed to intrigue students’ thinking and learning process. These are to help students reflect and synthesize the findings. These questions are not meant to be a suggested question-answer structure/format of the Task 2B report writing.  Is there likely to be a direct effect of Instagram post factors (i.e. message framing type and Instagrammer’s sexes) on consumer responses?  Is there any factors found to influence envy?  If so, why do you think this happens?  Does envy negatively or positively influence brand attitude?  If so, is the influence different by different consumer segments?  Considering all the evidence here and what you might have found elsewhere, how can you describe the nature and the role of envy in social media marketing?  Do you need to consider different marketing communication strategies for different consumer segments?  Do you think envy can be more evoked using communication/advertising executional tactics you have learned in MKTG204? If so, how?  Is marketing communications research important to campaign developments?  If so, how? Page 14 Appendix Instagram post 1a Instagram post 1b Page 15 Instagram post 2a Instagram post 2b