Results of Research Model

In order to conduct multilevel analyses, this research first conducts correlation analyses to understand multilevel relationships among corporate branding, brand psychological ownership, brand citizenship behavior, and brand equity. Then, this research conducts analyses of HLM. Based on Hofmann (1997), procedures of HLM have to examine four different models which include null model, random coefficients regression model, intercepts-as-outcomes model, and slopes-as-outcomes model.

Among these models, slopes-as-outcomes model is not conducted by this research because this research doesn’t investigate moderating effects of organizational variables on individual variables. That is, random coefficients regression models are

conducted in individual level analyses and intercepts-as-outcomes models are conducted in multilevel analyses.

5.7.2 Correlations

As reported in Table 5-6, brand psychological ownership is significantly related to brand citizenship behavior (r=0.795***, P<0.01). Corporate branding is significantly associated with brand psychological ownership (r=0.545***, P<0.01), and related to brand citizenship behavior with a marginal significance (r=0.305*, P<0.1). Brand citizenship behavior is significantly related to brand equity (r=0.447***, P<0.01). These correlation results were consistent with the hypotheses proposed by this study. This study further investigated the relationships between factors of corporate branding, brand psychological ownership, and brand organizational citizenship behaviors.

Table 5-6 Means, Standard Deviation, and Correlations of Research Constructs

Variables Mean S.D. (1) (2) (3) (4) (5)

Individual Level

(1) Brand psychological ownership 3.6316 0.71294 1

(2) Brand citizenship behavior 3.991 0.72539 0.795*** 1

(3)Gender 1.5856 0.49346 0.070 0.74 1

(4)Age 1.5685 0.87271 0.067 0.16 0.022 1

(5)Education 2.6301 0.67364 -0.044 -0.117 0.034 -0.12 1

Organizational Level

(1) Corporate branding 4.0012 0.53659 1

(2) Brand psychological ownership 3.5970 0.38610 0.545*** 1

(3) Brand citizenship behavior 4.0041 0.45728 0.305* 0.725*** 1

(4) Brand equity 3.7853 0.33018 0.269 0.526*** 0.447*** 1

(5) Type 0.8710 0.34078 0.111 0.159 0.166 -0.54 1

5.7.3 Null Model Analyses

This research utilizes null models to prove two phenomena. First, employees’

cognitions and behaviors are different from different franchisee organizations. Second, employees’cognitions and behaviors may be affected by individual level variables and contextual variables (i.e. corporate branding). It is more appropriate to investigate multilevel relationships if values of τ00reach significant level (Hofmann, 1997). Null models in which no predictors are evaluated on either the individual level or organizational level were evaluated by this study. According to the results in Table

5-7, the residual variances of the intercepts of brand psychological ownership (τ00=0.101, p<.001), including brand self-efficacy (τ00=0.085, p<.001), brand

accountability (τ00=0.043, p<.001), and identification and belongingness of brand (τ00=0.148, p<.001), are all significant. The residual variances of the intercepts of

brand citizenship behavior (τ00=0.124, p<.001), including helping behaviors of brand (τ00=0.069, p<.001), consideration and enhancement of brand (τ00=0.074, p<.001),

and sportsmanship and endorsement of brand (τ00=0.091, p<.001), are all significant.

That is, there exists heterogeneity of relationships explored in the proposed model among different organizations. Therefore, it is more appropriate to investigate the relationships among corporate branding, brand psychological ownership, and brand citizenship behavior through multilevel analyses.

5.7.3.1 Null Models

Level-1

Y_ij=β_0j+ε_ij Level-2 β_0j= γ₀₀+u_0j

Note: i =individuals, j =organizations; Y ij refers to brand psychological ownership, brand self-efficacy, brand accountability, identification and belongingness of brand, brand citizenship behavior, helping behavior of brand, consideration and enhancement of brand, and sportsmanship and endorsement of brand.

Table 5-7 Null Model

***P< 0.01, ** P<0.05, ** P<0.1

5.7.4 Individual Level Analysis- Random Coefficients Regression Model

In the individual level analysis, hypothesis 1 is proposed to investigate the

Fixed Effect Random Effect

Dependent Variables

Intercept γ00

S. E. P τ00 ² P

Brand psychological ownership 3.587*** 0.068 0.000 0.101*** 0.41 0.000

Brand self-efficacy 3.712*** 0.067 0.000 0.085 *** 0.368 0.000

Brand accountability 3.912*** 0.058 0.000 0.043*** 0.418 0.000

Identification and belongingness of brand 3.818*** 0.082 0.000 0.148*** 0.41 0.000

Brand citizenship behavior 3.969*** 0.075 0.000 0.124*** 0.420 0.000

Helping behavior of brand 4.063*** 0.067 0.000 0.069*** 0.494 0.000

Consideration and enhancement of brand 4.011*** 0.065 0.000 0.074*** 0.387 0.000

Sportsmanship and endorsement of brand 3.741*** 0.069 0.000 0.091*** 0.401 0.000

As reported in Table 5-8, brand psychological ownership positively affects brand citizenship behavior (β10=0.73, p<0.01), indicating that hypothesis 1 is supported.

The individual level model is showed as follows.

5.7.4.1 Random Coefficients Regression Model

Level-1

BCB_ij=β_0j+β_1j*(BPO_ij)+ β_2j*(gender_ij) +β_3j*(age_ij)+ β_4j*(education_ij)+ε_ij Level-2

β_0j= γ₀₀+ u_0j β_1j= γ₁₀+ u_1j β_2j= γ₂₀+ u_2j β_3j= γ₃₀+ u_3j β_4j= γ₄₀+ u_4j

Note: i =individuals, j =organizations

5.7.5 Multilevel Analyses- Intercepts-as-outcomes Model

In cross-level analyses, hypothesis 2 is proposed to investigate the relationship between corporate branding and brand psychological ownership. Hypothesis 3 is proposed to investigate the relationship between corporate branding and brand citizenship behavior. Hypothesis 4 is proposed to investigate the relationship between brand citizenship behavior and brand equity. As reported in Table 5-8, corporate

branding positively affects brand psychological ownership (β01=0.374, p<0.01) and brand citizenship behavior (β01=0.287, p<0.05), revealing that hypotheses 2 and 3 are supported. Aggregated brand CB positively affects equity (γ01=0.279, p<0.01), which indicating that hypothesis 4 is supported. Multilevel models are represented as follows.

5.7.5.1 Intercepts-as-outcomes Model

Multilevel Model-1

Level-1

BPO_ij=β_0j+ β_1j*(gender_ij) +β_2j*(age_ij)+ β_3j*(education_ij)+ε_ij Level-2

β_0j= γ₀₀+γ₀₁(CB_j) +γ₀₂(type_j) +u_0j β_1j= γ₁₀+ u_1j

β_2j= γ₂₀+ u_2j β_3j= γ₃₀+ u_3j

Multilevel Model-2

Level-1

BCB_ij=β_0j+ β_1j*(gender_ij) +β_2j*(age_ij)+ β_3j*(education_ij)+ε_ij Level-2

β_0j= γ₀₀+γ₀₁(CB_j) +γ₀₂(type_j) +u_0j

β_1j= γ₁₀+ u_1j β_2j= γ₂₀+ u_2j β_3j= γ₃₀+ u_3j

Multilevel Model-3

Level-1

BCB_ij=β_0j+β_1j*(BPO_ij) +β_2j*(gender_ij) +β_3j*(age_ij)+ β_4j*(education_ij)+ε_ij Level-2

β_0j= γ₀₀+γ₀₁(CB_j) +γ₀₂(type_j) +u_0j β_1j= γ₁₀+ u_1j

β_2j= γ₂₀+ u_2j β_3j= γ₃₀+ u_3j β_4j= γ₄₀+ u_4j

Multilevel Model-4

Level-1

Brand equity_ij=β_0j+ ε_ij Level-2

β_0j= γ₀₀+γ₀₁(BCB_j) +γ₀₂(type_j) +u_0j

Note: i =individuals, j =organizations

Type: 1=food-drink organizations, 0=retailer organizations

Table 5-8 Hierarchical Linear Modeling Results of the Proposed Model

aDeviance is a measure of model fit. Deviance =-2* log-likelihood of the full maximum-likelihood estimate.

***P< 0.01, ** P<0.05, *<0.1

Organizations n=31; Supervisors n=250; Employees n=283; Customers n=577 5.7.6 Cross-level Mediating Effect

To explore the importance of brand psychological ownership, this study further investigated the cross-level mediating effect of brand psychological ownership between corporate branding and brand CB through the four analytical steps of Baron and Kenny (1986). The first step is to confirm the effect of brand psychological ownership on brand citizenship behavior. The second step is to confirm the effect of corporate branding on brand citizenship behavior. The third step is to examine the

Models Model-1 Model-2 Model-3 Model-4

Dependent variable

Intercept 3.609*** 3.76*** 3.79*** 3.78***

Brand psychological ownership 0.73***

Gender 0.085 0.047 -0.019

Age 0.066 0.018 -0.043

Education -0.126* -0.166** -0.08

Organizational level

Corporate branding 0.374*** 0.287** 0.017

Brand citizenship behavior 0.279***

Type 0.123 0.146 0.054 -0.106

R2 0.11 0.06 0.63 0.03

Deviance^a 540.71 532.97 320.57 821.097

effect of corporate branding on brand psychological ownership. The fourth step is to examine whether the effect of corporate branding on brand CB became non-significant or reduced when both corporate branding and brand psychological ownership are jointly utilized as predictors of brand CB. If that is true, the cross-level mediating effect of brand psychological ownership is confirmed. This research conducted these three-step analyses. As reported in Table 5-8, the results of Model-1 revealed that corporate branding significantly affected brand psychological ownership (BPO) and the deviance was 540.71. Second, the results of Model-2 showed that corporate branding significantly affected brand CB and the deviance was 532.97.

Third, the results of Model-3 showed that brand PO significantly affected brand CB, and the deviance was 320.57. However, corporate branding didn’t affect brand CB significantly in Model-3. From the variation of deviance in three models, the value of deviance change is 212.4 (decreased from 532.97 to 320.57) after a mediating variable (i.e., brand psychological ownership) was added. The level of change reaches significant level of 0.005 (212.4＞X²(1) _0.005=7.879). Furthermore, the effect of corporate branding (γ01) on brand CB reduced from 0.287 to 0.017. Based on the aforementioned results, brand psychological ownership fully mediates the relationship between corporate branding and brand CB, revealing that brand psychological ownership is a cross-level mediator in the multilevel relation between corporate

branding and brand citizenship behavior. Hypothesis 5 is supported.