Abstract

Introduction

Literature review

Brand loyalty

Theoretical background

Potential moderators

Methodological moderators

Questionnaire type

Economic and social development moderators

Human development index

Level of globalization

Country Brand ranking

Connectivity level

Methods

Search strategy

Inclusion criteria

Coding procedure

Meta-analysis procedure

Results

Discussion

Implications for practitioners

Limitations and future studies

Reference

Abstract

     This research investigates the robustness of the relationship between social media marketing activities (SMMA) and brand loyalty (BL). This relationship is examined based on six contextual factors classified into three categories: methodological, economic, and social variables. A quantitative meta-analysis is conducted to test the study hypotheses, and 11 effect size from the article published between 2010 and 2019 are examined for the final sample size (n = ۳,۵۳۵). The meta-analysis results indicate a positive correlation between SMMA and BL at the cumulative level of 0.49, demonstrating a medium to large effect size. Potential moderating effects have been investigated for three types of study characteristic variables; the methodological (sample type and survey type), social (human development index [HDI] and level of globalization), and economic variables (country brand ranking and connectivity level). These relationships are moderated by sample type, survey type, and the country’s brand ranking, among other factors. However, the levels of globalization and connectivity, and the HDI, have no significant moderating effects on the SMMA–BL link. More importantly, the study indicates the factors that moderate these relationships. No prior meta-analysis of this fundamental relationship has been conducted

Introduction

     In today’s competitive global market, managers strive to improve marketing effectiveness by identifying new marketing metrics linked to business performance. Marketing campaigns on social media platforms (e.g., Facebook, Instagram, and Twitter) are considered to be one of the metrics used to promote corporate products, build strong brand awareness, and strengthen the relationships between companies and customers (Boyd & Ellison, 2007; Chan & Guillet, 2011; Erdo
gmus¸&C¸ic¸ek, 2012). Marketing through social media is defined as the process of using social media platforms to create, communicate, deliver, and exchange corporate offerings that have value for corporate stakeholders (Tuten & Solomon, 2016). Furthermore, social media is a communication tool: it enables users and their peers to communicate and add networks of friends, particularly among their peer groups (Filo et al., 2015; Ho, 2014; Whitla, 2009; Zhang & Daugherty, 2009).

Results

     Our bivariate meta-analysis’s outcomes exhibited a strong positive correlation between SMMA and BL at the cumulative level of 0.49, demonstrating a medium to the large effect size of the SMMA–BL linkage based on Cohen’s rule of thumb. As shown in Table 3, the confidence interval (CI) does not comprise the value zero (95%-CI [.32, .65]), so SMMA has a significant effect on BL. Hence, our assumption about the positive impact of SMMA on BL is strongly supported. Thus, Hypothesis 1 is accepted.

     Tables 3 and 4 report our findings of the subgroup analysis and metaregression. Our selected moderators (methodological, social, and economic variables) could clarify 73% of the variance in the relationship between SMMA and BL. For the moderating effect of sample type, the result of our subgroup analysis revealed that studies that comprised nonstudent-based samples reported large effect size (q ¼ .۶۰) compared to those that used student-based samples, which stated small effect size (q ¼ .۳۳), based on Cohens’ rule of thumb. The r-to-Z conversion result showed significant differences between the studies with student-based and nonstudent-based samples. Moreover, our meta-regression analysis exhibited a significant moderating effect of sample type on the SMMA–BL linkage. Thus, Hypothesis 2 is accepted. The corrected correlation coefficient of the studies that used an offline-based survey showed a more substantial value (q ¼ .۵۷) than that of the studies that used an online-based survey (q ¼ .۴۵). The outcome of our initial r-to-Z conversion and the meta-regression showed a significant moderating effect of survey type on the relationship between SMMA and BL. Thus, Hypothesis 3 is accepted.