Upload
vuongtram
View
219
Download
0
Embed Size (px)
Citation preview
73
CHAPTER – 4
ANALYSIS OF DATA
4.1 Introduction
This chapter provides the empirical findings on presenting the detailed analyses of
the data collected and the statistical results of the study. The chapter is divided into
three main sections.
The first section provides information on the profiles of the respondents and
descriptive statistics.
The second section provides preliminary analysis, which includes normality test,
reliability and validity.
The third section discusses hypothesis tests.
4.2 Profile of Respondents
Total 968 respondents were contacted. Data were collected from respondents in two
ways. They were requested to participate in the survey through a fill physical printout
of the questionnaire as well as questionnaire put online through online research
website. Sent request of online questionnaire through webpage link
by the way of
email and social networking website like Facebook, LinkedIn. The responses received
through online were 285 respondents (Annexture-3) were 19 responses were rejected
due to lack of sufficient information. The other approach was through a face-to-face
interview, approached 750 respondents and hence 702 respondents have accepted the
request and 48 have rejected the request. The reasons for rejecting the survey were
lack of interest, lack of time, lack of willingness etc. After collection of data, total 968
(702 Physical copy + 266 online) completely filled questionnaires were available for
the analysis. A detailed description of the respondents of this study is explained in
table 4.1.
In the questionnaire, six types of demographic information collected from the
respondents, namely, gender, age, education qualification, occupation and annual
family income. Out of the total respondents 55.3 % (535) were male and 44.7 percent
(433) were female. In case of age of respondents, 35 percent (339) of the respondents
Link for online questionnaire: https://qtrial.qualtrics.com/SE/?SID=SV_esuvSyJm4GGbzYp
74
were in the age group of 12-18 years and 19-25 years, was followed by 43.7 percent
(423) respondents who were more than 26-35 years old, while 21.2 percent (206) of
the respondents were in the age group 36 -50 and more than 50 years. In terms of
annually family income out of 968 respondents, 10.1 percent (98) of respondents‟
family income were below ₹ 1 lac. Total 30.4 percent (294) respondents described
their income between ₹ 1 lac to ₹ 3 lac. Respondents whose family income between ₹
7 Lac to 12 Lac was 16.8 percent (163) from the sample. The remaining 4 percent
(39) of the respondents reported their annually family income as greater than ₹ 12 lac.
Table: 4.1 Demographic profiles of respondents
Profile of Respondents
Characteristics Measuring Variables Frequency Percentage
Gender
Male 535 55.3
Female 433 44.7
Total 968 100
Age
12 to18 years 29 3
19 to 25 years 310 32
26 to 35 years 423 43.7
36 to 50 years 193 19.9
More than 50 years 13 1.3
Total 968 100
Educational
Qualification
Up to HSC 121 12.5
Graduate 454 46.9
Post Graduate 372 38.4
Other 21 2.2
Total 968 100
Occupation
Business 88 9.1
Service 393 40.6
Professional 75 7.7
Student 226 23.3
Housewife 182 18.8
Other 04 0.4
Total 968 100
Annual Family
Income
Below Rs. 1 Lac 98 10.1
1 lac to 3 lac 294 30.4
3 Lac to 7 Lac 374 38.6
7 Lac to 12 Lac 163 16.8
More than 12 Lac 39 4.0
Total 968 100
Source: Primary data collected by the researcher
75
In terms of education qualification, the majority of respondents had a
graduation and post graduation level degree. The sample consisted of 38.4 percent
(372) post graduate respondents. Total 46.9 percent (454) respondents were graduates,
whereas 12.5 percent (121) respondents had studied up to higher secondary and
hardly 2.2 percent (21) respondents had other qualification means the diploma holder,
I.T.I holder etc. The noticeable participation of graduate and post graduate
respondents was in this research.
In case of occupation, the 40.6 percent (393) of the respondents were
employees (Service) and 23.3 percent (226) were students of the total respondents. 9.1
percent (81) respondents were the occupation of business and seventy five (7.7%)
respondents were professionals. Whereas 18.8 % respondents (182) ware house wife
and merely four respondents were comes under another category means all are found
unemployed. The data were collected from major cities of Gujarat state. These cities
are Ahmedabad (250), Vadodara (175), Rajkot (175), Bhavnagar (90), Surat (160),
Mehsana (60), Himatnagar (58) etc.
76
4.3 Source of information
Respondents were asked to source, where usually they got information regarding to
product or services. The question was asked in to multiple choice answers among
given various options.
Table: 4.2 Frequency of sources of information
Source of Information Frequency Percentage
Television 884 91.32
Magazine 355 36.67
Newspaper 733 75.72
Hoardings 408 42.15
Internet 556 57.44
In store display 262 27.07
Radio 167 17.25
Neighbor 194 20.04
Friend 538 55.58
Colleague 251 25.93
Direct mail 123 12.71
Social networking 353 36.47
Other 33 3.41
Source: Primary data collected by the researcher
From above table 4.2, it has been found that 91.32% (884) respondents got
information from television while second most preferable source was newspaper.
Total thirteen sources were given, and among them four sources named television,
newspaper, internet and friends were playing a major role (more than 50%) for
accessing information. Here noticeable source is social media (36.47 %). Due to the
penetration and popularity of FM radio stations, out of the total 17.25% of
respondents got information, where as hoarding was 42.15 %. At last magazine
(36.67%) and neighbor (20.04%) identified as a source of information.
77
4.4 Identification of Product categories endorsed by celebrities
Table: 4.3 Identification of Product categories endorsed by celebrity
FMCG
Category Product category Yes % No %
Household Care Detergent Powder/Soap 683 70.6 285 29.4
Home cleaning 517 53.4 451 46.6
Personal Care
Oral Care 731 75.5 237 24.5
Skin Care 796 82.2 172 17.8
Hair Care 807 83.4 161 16.6
Foods
Bakery products 575 59.4 393 40.6
Snack food 652 67.4 316 32.6
Chocolates 700 72.3 268 27.7
Beverages Tea 633 65.4 335 34.6
Soft drinks 712 73.6 256 26.4
Insurance Life Insurance 529 54.6 439 45.4
Source: Primary data collected by the researcher
Respondents were asked to identify celebrity endorsement across various product
categories which were given in to tabular form in the questionnaire. If they remember
any of the celebrity endorsement in relevant product categories then ticks on „Yes‟
otherwise tick on „No‟. From the data interpretation, it is found that 83.4% (807)
respondents identify as celebrity endorsed Hair care related products. The major
categories identified by respondents were detergent powder/soap, oral care, skin care,
chocolates and soft drink.
78
4.5 Identification of Product categories endorsed by non-celebrity
Table: 4.4 Identification of Product categories known by the respondents by
Non-celebrity
FMCG
Category Product category Yes % No %
Household Care Detergent Powder/Soap 574 59.6 394 40.7
Home cleaning 597 61.7 371 38.3
Personal Care
Oral Care 605 62.5 363 37.5
Skin Care 532 55.0 436 45.0
Hair Care 586 60.5 382 39.5
Foods
Bakery products 577 59.6 391 40.4
Snack food 623 64.4 345 35.6
Chocolates 642 66.3 326 33.7
Beverages Tea 591 61.1 377 38.9
Soft drinks 513 53.0 455 47.0
Insurance Life Insurance 663 68.5 305 31.5
Source: Primary data collected by the researcher
In case of non-celebrity endorsement table 4.4, respondents were asked to identify
non-celebrity endorsement across various product categories given in to tabular form
in the questionnaire. If they remember any of the non-celebrity endorsement in
relevant category than tick on „Yes‟ otherwise „No‟. From data interpretation, it is
found that 68.5 (663) respondents identify as non-celebrity endorsed in the Insurance
category. The major categories identified by respondents were Home cleaning, Oral
care, Hair care, Snack food, chocolates, Tea and Insurance.
79
4.6 Liking of advertisement endorsed by celebrity and Non-celebrity
Table: 4.5 Liking of advertisement of product categories
Category Advertisement Frequency Percentage
Detergent
powder/soap
Celebrity (Salman Khan) 380 39.3
Non Celebrity 588 60.7
Home cleaning Celebrity (Hussain) 590 61.0
Non Celebrity 378 39.0
Oral Care Celebrity (Shahrukh Khan) 471 48.7
Non Celebrity 497 51.3
Skin Care Celebrity (Aishwarya Rai) 666 68.8
Non Celebrity 302 31.2
Hair Care Celebrity (Katrina Kaif) 697 72.0
Non Celebrity 271 28.0
Bakery products Celebrity (Amitabh Bachchhan) 342 35.3
Non Celebrity 626 64.7
Snack food Celebrity (Parineeti Chopra) 694 71.7
Non Celebrity 274 28.3
Chocolates Celebrity (Katrina Kaif) 455 47.0
Non Celebrity 513 53.0
Tea Celebrity (Saif Ali Khan) 522 53.9
Non Celebrity 446 46.1
Soft drinks Celebrity (MS Dhoni) 684 70.7
Non Celebrity 284 29.3
Life Insurance Celebrity (Sachin Tendulkar) 342 35.3
Non Celebrity 626 64.7
Mobile Handset Celebrity (Priyanka Chopra) 674 69.6
Non Celebrity 294 30.4
Source: Primary data collected by the researcher
Respondents have to give their choice for advertisement from the pair of
advertisement (celebrity and non-celebrity) for each product category. The selection
of advertisement for this research is based on screening of all the advertisements,
including print advertisements as well as a television commercial. The list of
advertisements given in annexture-1, from this pair, I have selected those
advertisements having a maximum frequency of appearance in the TVC and print
media. According to Sherman (1985), approximately 20 percent of all television
advertisements include famous people and approximately 10 percent of the money
spent on television advertisements are used on celebrity endorsements. Table 4.5,
80
showing choice of advertisement from the pair of advertisement given in
questionnaire.
For non-celebrity endorsed products, Bakery products have highest liking of
64.7%, followed by life insurance (64.7%). Where as In the Detergent powder/soap
category 60.7 % responded likes. The celebrity advertisement likes by respondents
under categories of Home cleaning (61%, famous TV star Hussein endorse Harpic
brand toilet cleaner), Skin care (68.8%, Aishwarya Rai endorse L‟oreal skin cream).
Further Hair care category is highest for liking of celebrity endorsed advertisement
(72%, Ketrina Kaif endorse Pentene brand shampoo), Snack food (71.7%, Parneeti
Chopra endorse Kurkure), Tea (53.9%, Saif Ali khan endorse Tajmahal Tea), Soft
drink (70.7%, M.S. Dhoni endorse Pepsi) and Mobile handset (69.6%, Priyanka
Chopra endorse the Sony brand).
4.7 Descriptive Statistics
The mean and standard deviation were calculated for all the intervals scaled items.
These are reported in table 4.6 along with the skewness and kurtosis of these items.
The results indicated that the responses to the variables had a good dispersion on the
scales.
Table 4.6 Descriptive Statistics of celebrity and Non-celebrity’s characteristics
Constructs Items N Mean Std.
Deviation Skewness Kurtosis
Celebrity‟s
Physical
Attractiveness
Beautiful 968 6.01 1.410 -1.837 3.125
Elegant 968 5.36 1.367 -0.984 0.938
Sexy 968 5.22 1.748 -0.879 -0.183
Pleasant 968 5.50 1.528 -1.062 0.591
Celebrity‟s
Trustworthiness
Dependable 968 5.16 1.703 -0.924 0.122
Honest 968 5.42 1.725 -1.047 0.122
Reliable 968 5.40 1.595 -0.981 0.286
Sincere 968 5.32 1.661 -0.852 -0.025
Believable 968 5.21 1.747 -0.865 -0.242
Having A Good
Reputation 968 5.66 1.660 -1.296 0.812
Celebrity‟s
Expertise
Experienced 968 5.57 1.646 -1.190 0.644
Knowledgeable 968 5.50 1.612 -1.112 0.513
Qualified 968 5.26 1.577 -0.898 0.119
Skilled 968 5.47 1.508 -1.052 0.448
Professional 968 5.51 1.537 -1.111 0.769
81
Constructs Items N Mean Std.
Deviation Skewness Kurtosis
Celebrity‟s
Likability
Familiar 968 5.60 1.514 -1.211 1.002
Similar 968 5.27 1.450 -0.973 0.648
I can relate 968 4.95 1.705 -0.764 -0.085
Appropriate 968 5.10 1.562 -0.762 -0.070
Logical 968 4.95 1.581 -0.589 -0.193
Non-celebrity‟s
Physical
Attractiveness
Beautiful 968 5.41 1.692 -0.990 -0.026
Elegant 968 5.07 1.635 -0.596 -0.029
Sexy 968 4.82 1.823 -0.460 -0.833
Pleasant 968 5.10 1.727 -0.655 -0.512
Non-celebrity‟s
Trustworthiness
Dependable 968 4.82 1.751 -0.645 -0.537
Honest 968 4.94 1.792 -0.591 -0.698
Reliable 968 4.80 1.798 -0.459 -0.825
Sincere 968 4.88 1.832 -0.491 -0.879
Believable 968 4.85 1.841 -0.492 -0.877
Having A Good
Reputation 968 4.82 1.841 -0.446 -0.967
Non-celebrity‟s
Expertise
Experienced 968 4.92 1.866 -0.589 -0.826
Knowledgeable 968 5.02 1.769 -0.565 -0.787
Qualified 968 4.86 1.707 -0.403 -0.836
Skilled 968 5.03 1.735 -0.537 -0.764
Professional 968 4.79 1.808 -0.424 -0.888
Non-celebrity‟s
Likability
Familiar 968 4.81 1.868 -0.556 -0.810
Similar 968 4.66 1.679 -0.492 -0.646
I can relate 968 4.44 1.751 -0.342 -0.849
Appropriate 968 4.64 1.680 -0.351 -0.796
Logical 968 4.51 1.728 -0.270 -0.956
Source: Primary data collected by the researcher
Distribution of skewness is either symmetric or skewed. In a distribution, the values
on either side of the centre of the distribution are the same. The negative data
indicating a negative skew.
Kurtosis is measure of relative peakedness or flatness of the curve defined by
the frequency distribution. If the kurtosis is positive, than the distribution is more
peaked than a normal distribution. A negative value means that the distribution is
flatter than a normal distribution.
82
Table 4.7 Descriptive Statistics of variables of effectiveness of celebrity.
Constructs Items
(Variables) N Mean
Std.
Deviation Skewness Kurtosis
Celebrity
endorsement
effectiveness
(Q-7)
V1 968 4.30 0.933 -1.381 1.528
V2 968 3.20 1.210 -0.398 -0.751
V3 968 3.93 1.119 -0.807 -0.285
V4 968 3.40 1.263 -0.354 -0.872
V5 968 3.16 1.336 -0.259 -1.110
V6 968 3.23 1.259 -0.218 -0.955
V7 968 3.15 1.187 -0.305 -0.837
V8 968 3.25 1.083 -0.232 -0.446
V9 968 3.24 1.193 -0.220 -0.741
V10 968 2.93 1.317 0.058 -1.133
V11 968 3.80 1.184 -0.779 -0.254
V12 968 3.48 1.200 -0.487 -0.636
V13 968 3.58 1.138 -0.520 -0.472
V14 968 3.44 1.261 -0.561 -0.628
V15 968 3.39 1.336 -0.467 -0.926
V16 968 3.58 1.104 -0.413 -0.605
V17 968 2.77 1.328 0.010 -1.230
V18 968 3.09 1.295 -0.115 -1.034
V19 968 3.12 1.354 -0.127 -1.149
V20 968 3.29 1.307 -0.331 -0.959
V21 968 2.87 1.250 0.028 -0.953
V22 968 2.80 1.289 0.036 -1.036
V23 968 3.11 1.381 -0.187 -1.200
V24 968 3.78 1.155 -0.709 -0.367
Source: Primary data collected by the researcher
The means of the sixty eight items were ranged satisfactorily from 6.01 to 2.77 with
standard deviations ranging from 0.933 to 1.868 on the 7-point and 5-point scales. As
it can be observed from the table 4.6, characteristics of celebrity, Non-celebrity and its
endorsement effectiveness are neutral. The one item under celebrity‟s physical
attractiveness has values greater than 6.0 which is the neutral value.
4.8 Measurement of Reliability and Validity
It is essential to examine the reliability and validity of the twenty four interval-scaled
items that measured the five constructs. The measurements of reliability and validity
are important steps in ensuring that the variables measure the relevant construct
exactly and accurately. The reliability is defined as an assessment of the degree of
consistency between multiple measurements of a variable (Hair et al., 2010).
83
According to Sekaran (2003), the internal consistency of measures is an indication of
the homogeneity of the items which measure the same construct. The validity is the
extent to which a set of measuring variables accurately represents the concept of
interest (Hair et al., 2010). This test confirms if the multiple variables developed for a
construct rightly measure that construct. The measuring variables were developed and
tested in the literature. It remained necessary to test their reliability and validity,
especially in a new dataset. This might be helpful in validating the previous findings.
In this study reliability was tested using Cronbach‟s alpha and exploratory factor
analysis.
4.8.1 Reliability
Cronbach‟s alpha was used to test the reliability of the interval-scaled variables. It is
the most popular test of inter-item consistency, reliability, which is useful for
multipoint- scaled variables (Sekaran, 2003). The cutoff of reliability α is 0.70 (Hair
et al., 2010), hence the reliabilities ≥ 0.70 are acceptable. The results showed that all
eight constructs obtained acceptable reliabilities, which is more than 0.7, except one
case of celebrity‟s physical attractiveness had 0.673 which was accepted. The
constructs were described in the table 4.7 with the value of cronbach‟s alpha, which
indicates that the test of reliability approves the scale. From the below table 4.2 values
of cronbach‟s alpha it can be concluded that all the constructs of influencing
characteristics of celebrity and non-celebrity‟s Physical attractiveness,
Trustworthiness, Expertise and Likability were found reliable. The reliability of the
constructs also indicated that further analysis is possible.
84
Table: 4.8 Reliability and Factor analysis for Celebrity and Non-celebrity.
Factor analysis for Celebrity
Constructs Items Cronbach’s Alpha Factor loading
Celebrity‟s
Physical
Attractiveness
Beautiful
0.673
0.780
Elegant 0.764
Sexy 0.610
Pleasant 0.708
Celebrity‟s
Trustworthiness
Dependable
0.873
0.575
Honest 0.856
Reliable 0.834
Sincere 0.823
Believable 0.818
Having A Good
Reputation 0.781
Celebrity‟s
Expertise
Experienced
0.858
0.753
Knowledgeable 0.860
Qualified 0.803
Skilled 0.852
Professional 0.729
Celebrity‟s
Likability
Familiar
0.833
0.757
Similar 0.791
I can relate 0.794
Appropriate 0.812
Logical 0.719
Factor analysis for Non-celebrity
Non-celebrity‟s
Physical
Attractiveness
Beautiful
0.840
0.822
Elegant 0.866
Sexy 0.775
Pleasant 0.830
Non-celebrity‟s
Trustworthiness
Dependable
0.917
0.732
Honest 0.876
Reliable 0.863
Sincere 0.876
Believable 0.861
Having A Good
Reputation 0.832
Non-celebrity‟s
Expertise
Experienced
0.911
0.819
Knowledgeable 0.892
Qualified 0.864
Skilled 0.873
Professional 0.853
Non-celebrity‟s
Likability
Familiar
0.894
0.795
Similar 0.856
I can relate 0.856
Appropriate 0.868
Logical 0.824
Source: Primary data collected by the researcher
85
4.8.2 Validity
A measurement scale has validity, if it measures what it is supposed to measure
(Aaker et al., 2009). Churchill et al., (2010) argue that at a minimum, a researcher
should establish that new measures developed have face validity, construct validity,
discriminant validity, convergent validity and nomological validity. Face validity is
the extent to which the content of the items is consistent with the construct definition,
based solely on the researcher‟s judgment. The face validity is invoked when it can be
argued that the measurement scale apparently reflects the content of the concept, it is
trying to measure (Aaker et al., 2009). In the pre-testing of the measures, the face
validity of the measurement scales was established. The construct validity is the
extent to which a set of measured variables actually represent the theoretical latent
construct they are designed to measure. It is made up of four components: convergent
validity, discriminant validity, nomological validity and construct reliability. The
convergent validity is the extent to which indicators of a specific construct “converge”
or share a high proportion of variance in common.
The discriminant validity establishes that a construct is truly distinct from
other constructs. The nomological validity is tested by examining whether or not the
correlations between the constructs in the measurement theory make sense. The
construct reliability is a measure of reliability and internal consistency based on the
square of the total of factor loadings for a construct (Churchill et al., 2010; Aaker et
al., 2009).
The factor analysis techniques can achieve their purposes from either an
exploratory or confirmatory perspective (Hair et al., 2010). The exploratory factor
analysis is an effective method to assess if the dimensions of a concept are strongly
associated with each other to represent that concept (Hair et al., 2010; Sekaran, 2003).
The exploratory factor analysis is often considered to be more appropriate and
conducted before confirmatory analysis. In this validity analysis, exploratory factor
analysis was adopted to confirm whether the different variables were loaded on the
right constructs. When applying exploratory factor analysis, there were three main
steps. The first step was to assess the appropriateness of applying this technique. The
second step was to conduct the factor extraction, while the third was to interpret factor
loadings among the multiple variables. In this study, four types of constructs/factors
(i) Celebrity‟s physical attractiveness, (ii) Trustworthiness, (iii) Expertise and (iv)
86
Likability were tested with the help of total twenty items and using a 7 - point scale.
Therefore, factor analysis was conducted separately for all four main measurement
scales. Same scale was applicable to non-celebrity.
4.8.3 Exploratory Factor Analysis
Under descriptive analysis of research, the exploratory factor analysis was conducted
using the principal component method and Varimax rotation at the initial stage of
analysis using SPSS (Hair et al., 2010; Sekaran, 2003). The twenty items that derived
from the literature review and mentioned in chapter three, here table 4.9 showing
exploratory factor analyses. The exploratory factor analysis was applied first on
physical attractiveness, Trustworthiness, Expertise and Likability of celebrity.
In this study, the result of Bartlett‟s test of Sphericity (Chi-Square value 9565,
degree of freedom (df) 190, p value 0.000) and KMO (0.935) was found significant
and which indicate that the data were appropriate for factor analysis. In the analysis
only the factors having latent roots or eigenvalue greater than 1 were considered
significant. In this research, an orthogonal rotational method, namely, Varimax was
adopted. This method was chosen because, compared to other rotated methods, it is a
more useful means of reducing the variables to a smaller set of uncorrelated variables
for subsequent use in the regression technique, as well as to give a clearer separation
of the factors (Hair et al., 2010). According to Hair et al. (2010) the factor loadings
exceeding ±0.60 are considered practically significant.
As shown in table 4.9, each of the variables obtained a factor loading greater
than 0.6. The four factors were extracted during this stage of exploratory factor
analysis, accounted for 61.37 percent of the total variance. These four factors were
physical attractiveness, Trustworthiness, Expertise and Likability.
Table: 4.9 Exploratory Factor Analyses for celebrity
KMO and Bartlett's Test
Kaiser-Meyer-Olkin Measure of Sampling
Adequacy. 0.935
Bartlett's Test of
Sphericity
Approx. Chi-Square 9565.550
df 190
Sig. 0.000
87
Celebrity Attitude Factor Loadings Factor
Beautiful 0.780 PHYSICAL
ATTRACTIVENESS
Elegant 0.764
Sexy 0.610
Pleasant 0.708
Dependable 0.575
TRUSTWORTHINESS
Honest 0.856
Reliable 0.834
Sincere 0.823
Believable 0.818
Having A Good Reputation 0.781
Experienced 0.753
EXPERTISE
Knowledgeable 0.860
Qualified 0.803
Skilled 0.852
Professional 0.729
Familiar 0.757
LIKABILITY
Similar 0.791
I can relate 0.794
Appropriate 0.812
Logical 0.719
Source: Primary data collected by the researcher
The second exploratory factor analysis was conducted for non-celebrity, comprising
of four constructs/items and twenty variables same as celebrity. As shown in table
4.10 the result of Bartlett‟s test of sphericity (Chi Square value 14760, df 190, p value
0.000) and KMO (0.965) were found significant and which indicated that the data
were appropriate for further factor analysis. The eigenvalue greater than 1 was
considered significant, and varimax rotation was conducted which resulted in factor
loadings greater than ±0.60. Total four factors were extracted during this stage of
exploratory factor analysis, accounted for 63.81 percent of the total variance.
Table: 4.10 Exploratory Factor Analyses for Non-celebrity.
KMO and Bartlett's Test
Kaiser-Meyer-Olkin Measure of Sampling
Adequacy. 0.965
Bartlett's Test of
Sphericity
Approx. Chi-Square 14760.125
df 190
Sig. 0.000
88
Non-celebrity Attitude Factor Loadings Factor
Beautiful 0.822 PHYSICAL
ATTRACTIVENESS
Elegant 0.866
Sexy 0.775
Pleasant 0.830
Dependable 0.732
TRUSTWORTHINESS
Honest 0.876
Reliable 0.863
Sincere 0.876
Believable 0.861
Having A Good Reputation 0.832
Experienced 0.819
EXPERTISE
Knowledgeable 0.892
Qualified 0.864
Skilled 0.873
Professional 0.853
Familiar 0.795
LIKABILITY
Similar 0.856
I can relate 0.856
Appropriate 0.868
Logical 0.824
Source: Primary data collected by the researcher
4.9 Multiple Regression Analysis of Factors affecting Celebrity Endorsement and
Celebrity Endorsement Effectiveness
In reference of first objective of research, the multiple regression analysis was
employed. Hypotheses developed regarding consumer perception for celebrity
endorsements in advertisements .The first four hypotheses designed with respect to
means score of twenty four statements related to effectiveness of celebrity
endorsement in advertisement (Question-7) as a dependent variable, the analysis
firstly investigates whether the data have met basic assumptions for applying multiple
regression analysis. Then presents the multiple regression analysis for celebrity‟s
characteristics. These are physical attractiveness, trustworthiness, expertise and
likability. Hair et al. (2010) considered linearity of relationships, constant variance of
residuals, and the normality of residual distribution essential assumptions underlying
multiple regression analysis, which can ensure that the results obtained from multiple
regression analysis are truly representative of the sample and that the best results are
deemed to be obtained. The partial regression plots and the constant variance support
the linearity. All these assumptions were all satisfactorily confirmed. The assumptions
89
of linearity, constant variance and normality were statistically supported to employ
multiple regression analysis.
Dependent variable (Y): Effectiveness of celebrity endorsement in advertisement
Independent Variables (X): Factors affecting Celebrity Endorsement / Celebrity‟s
characteristics
For independent variables Physical Attractiveness, Trustworthiness, Likability and
expertise are taken one by one for regression analysis
According to Churchill et al. (2010) and Aaker et al. (2009) the equation in multiple
regression analysis has the following form:
Y = α + β1X1 + β2X2 + … + βnXn + ε
Where,
Y = the level of the dependent, or predicted, variable
α = the intercept or constant
β1 = the slope coefficient of the independent variable X1
X1 = the level of the independent variable X1
β2 = the slope coefficient of the independent variable X2
X2 = the level of the independent variable X2
βn = the slope coefficient of the independent variable Xm
Xn = the level of the independent variable Xm
n = the number of independent variables in the equation
ε = the random error associated with the prediction of Y
4.9.1 Hypothesis testing for celebrity endorsement effectiveness
H01a: There is no significant relationship between physical attractiveness of celebrity
and mean score of celebrity endorsement effectiveness.
H01b: There is no significant relationship between trustworthiness of celebrity and
mean score of celebrity endorsement effectiveness.
H01c: There is no significant relationship between expertise of celebrity and celebrity
mean score of endorsement effectiveness.
90
H01d: There is no significant relationship between likability of celebrity and celebrity
mean score of endorsement effectiveness.
H01e: There is no significant relationship between physical attractiveness of non-
mean score of celebrity and celebrity endorsement effectiveness.
H01f: There is no significant relationship between trustworthiness of non-celebrity
and mean score of celebrity endorsement effectiveness.
H01g: There is no significant relationship between expertise of non-celebrity and
mean score of celebrity endorsement effectiveness.
H01h: There is no significant relationship between likability of non-celebrity and
mean score of celebrity endorsement effectiveness.
As shown in table 4.11 under the title of analysis of variance, this percentage was
statistically significant (F = 11.662, p<0.05). There was no threat of multicollinearity
among the predictor variables as the value of tolerance for each of the independent
variables was more than 0.2 and the VIF (Variance Inflation Factor) was less than 5
(Churchill et al., 2010).
Table: 4.11 Analysis of variance for celebrity endorsement effectiveness
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 17.307 8 2.163
11.662 0.000 Residual 177.902 959 0.186
Total 195.210 967
Source: Primary data collected by the researcher
As seen in the table 4.12, the intercept (identified as the constant) was significant at
the level of 0.05 (B=2.285; t=22.088). From all eight factors, only one regression
coefficients was statistically not significant (p < 0.05). This coefficient implied that
there is a significant relationship exists among celebrity‟s likability characteristics.
91
Table 4.12: Regression analyses for celebrity endorsement effectiveness
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.795 32.046 0.000 2.795
Celebrity's Physical
Attractiveness
0.017 0.042 1.182 0.238 0.764 1.308
Celebrity's
Trustworthiness
-0.022 -0.063 -1.277 0.202 0.386 2.592
Celebrity's Expertise -0.023 -0.064 -1.325 0.186 0.411 2.435
Celebrity's Likability 0.132 0.356 7.773 0.000 0.453 2.206
Non-celebrity's
Physical
Attractiveness
-0.015 -0.046 -1.095 0.274 0.533 1.875
Non-celebrity's
Trustworthiness
0.000 -0.001 -0.010 0.992 0.247 4.049
Non-celebrity's
Expertise
0.020 0.069 1.073 0.284 0.233 4.296
Non-celebrity's
Likability
-0.006 -0.020 -0.364 0.716 0.303 3.302
So, the hypothesis is there is a significant relationship between celebrity‟s likability
and means score of celebrity endorsement effectiveness. The positive beta shows
positive relationship.
4.9.2 Summary of Multiple Regression Analysis.
Table 4.13 Summary of hypotheses for celebrity endorsement effectiveness
Hypothesis statement Result
H01a
There is no significant relationship between physical
attractiveness of celebrity and mean score of celebrity
endorsement effectiveness.
Not rejected
H01b
There is no significant relationship between trustworthiness
of celebrity and mean score of celebrity endorsement
effectiveness.
Not rejected
H01c
There is no significant relationship between expertise of
celebrity and mean score of celebrity endorsement
effectiveness.
Not rejected
H01d
There is no significant relationship between likability of
celebrity and mean score of celebrity endorsement
effectiveness.
Rejected
92
Summary of multiple regression from above table 4.13, it is concluded that weather
celebrity or non-celebrity endorsement, Physical attractiveness, Trustworthiness and
Expertise characteristics might not play very important role but in case of celebrity‟s
likability was significantly influence the overall effectiveness of endorsement.
H01e
There is no significant relationship between physical
attractiveness of non-celebrity and mean score of celebrity
endorsement effectiveness.
Not rejected
H01f
There is no significant relationship between trustworthiness
of non-celebrity and mean score of celebrity endorsement
effectiveness.
Not rejected
H01g
There is no significant relationship between expertise of non-
celebrity and mean score of celebrity endorsement
effectiveness.
Not rejected
H01h
There is no significant relationship between likability of non-
celebrity and mean score of celebrity endorsement
effectiveness.
Not rejected
93
4.10 t-Tests for effectiveness of celebrity on product preferences
t- Test is used to test a hypothesis stating that the mean scores on some variable will
be significantly different for two independent samples or groups. It is used when the
number of observations (sample size) is small and the population standard deviation is
unknown (Zikumnd pp. 506). Compares the means of one variable for two groups of
cases. Descriptive statistics for each group and Levene‟s test for equality of variances
are provided, as well as both equal- and unequal-variance t-values and a 95%
confidence interval for the difference in means. Here, t-test applied for celebrity and
non-celebrity endorser‟s characteristics and mean score of celebrity endorsement
effectiveness. The hypothesis test for the variables were conducted using t statistics
examination in case of two categorical groups (Churchill et al., 2010) and one way
ANOVA employed in case of more than two category groups (Malhotra, 2007;
Churchill et al., 2010). The hypothesis test for Gender, Age group, Education
qualification, Occupation and Family monthly income were conducted for
characteristics of celebrity and non-celebrity. The t test was conducted for gender and
one way ANOVA was employed by Age Group, Education, Qualification,
Occupation and Family monthly income.
4.10.1 Hypothesis for effectiveness of celebrity on product preferences
H02a: There is no significant difference between likening advertisement of fabric care
product and mean score of celebrity endorsement effectiveness.
Table 4.14: t-test for likening advertisement of fabric care product and mean
score of celebrity endorsement effectiveness.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 0.140 0.709 1.299 966 0.194
Equal variances not assumed 1.282 772.910 0.200
Source: Primary data collected by the researcher
As shown in Table 4.14: the t-test results indicate that there is no significant
difference found in likening of celebrity and non-celebrity endorser for advertisement
of fabric care product. Under the title of Levene‟s test for equality variances, the F
94
value was 0.140 with a Significance value of 0.709. The p value was not significant at
0. 194, indicating the perceived difference of celebrity endorsement effectiveness and
likening advertisement of fabric care product was not significant (p > 0.05). Thus,
hypothesis H02a was not rejected.
H02b: There is no significant difference between likening of advertisement of home
care product and mean score of celebrity endorsement effectiveness.
Table 4.15: t-test for likening advertisement of home care product and mean
score of celebrity endorsement effectiveness.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 7.355 0.007 0.962 966 0.336
Equal variances not assumed 1.282 772.910 0.986
Source: Primary data collected by the researcher
As shown in Table 4.15: the t-test results indicate that there is no significant
difference found in likening of celebrity and non-celebrity endorser for advertisement
of home care product. Under the title of Levene‟s test for equality variances, the F
value was 7.355 with a Significance value of 0.007. The p value was not significant at
0.336, indicating the perceived difference of celebrity endorsement effectiveness and
likening advertisement of home care product was not significant (p > 0.05). Thus,
hypothesis H02b was not rejected.
H02c: There is no significant difference between likening advertisement of oral care
product and mean score of celebrity endorsement effectiveness.
Table 4.16: t-test for likening advertisement of oral care product and mean score
of celebrity endorsement effectiveness.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 0.748 0.387 1.809 966 0.071
Equal variances not assumed 1.811 965.809 0.070
Source: Primary data collected by the researcher
95
As shown in Table 4.16: the t-test results indicate that there is no significant
difference found in likening of celebrity and non-celebrity endorser for advertisement
of oral care product. Under the title of Levene‟s test for equality variances, the F
value was 0.748 with a Significance value of 0.387. The p value was not significant at
0.071, indicating the perceived difference of celebrity endorsement effectiveness and
likening advertisement of oral care product was not significant (p > 0.05). Thus,
hypothesis H02c was not rejected.
H02d: There is no significant difference between likening advertisement of skin care
product and mean score of celebrity endorsement effectiveness.
Table 4.17: t-test for likening advertisement of skin care product and mean score
of celebrity endorsement effectiveness.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 0.762 0.383 0.572 966 0.567
Equal variances not assumed 1.811 965.809 0.555
Source: Primary data collected by the researcher
As shown in Table 4.17: the t-test results indicate that there is no significant
difference found in likening of celebrity and non-celebrity endorser for advertisement
of skin care product. Under the title of Levene‟s test for equality variances, the F
value was 0.762 with a Significance value of 0.383. The p value was not significant at
0.567, indicating the perceived difference of mean score of celebrity endorsement
effectiveness and likening advertisement of skin care product was not significant (p >
0.05). Thus, hypothesis H02d was not rejected.
96
H02e: There is no significant difference between likening advertisement of hair care
product and mean score of celebrity endorsement effectiveness.
Table 4.18: t-test for likening advertisement of hair care product and mean score
of celebrity endorsement effectiveness.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 2.239 0.135 -1.971 966 0.049
Equal variances not assumed 1.811 965.809 -2.034
Source: Primary data collected by the researcher
As shown in Table 4.18: the t-test results indicate that there is a significant difference
found in likening of celebrity and non-celebrity endorser for advertisement of hair
care product. Under the title of Levene‟s test for equality variances, the F value was
2.239 with a Significance value of 0.135. The p value was significant at 0.049,
indicating the perceived difference of celebrity endorsement effectiveness and
likening advertisement of hair care product was significant (p < 0.05). Thus,
hypothesis H02e was rejected.
H02f: There is no significant difference between likening advertisement of bakery
product and mean score of celebrity endorsement effectiveness.
Table 4.19: t-test for likening advertisement of bakery product and mean score
of celebrity endorsement effectiveness.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 0.102 0.750 1.757 966 0.079
Equal variances not assumed 1.811 965.809 1.742
Source: Primary data collected by the researcher
As shown in Table 4.19: the t-test results indicate that there is no significant
difference found in likening of celebrity and non-celebrity endorser for advertisement
of bakery product. Under the title of Levene‟s test for equality variances, the F value
97
was 0.102 with a Significance value of 0.750. The p value was not significant at
0.079, indicating the perceived difference of celebrity endorsement effectiveness and
likening advertisement of bakery product was not significant (p > 0.05). Thus,
hypothesis H02f was not rejected.
H02g: There is no significant difference between likening advertisement of snack
food product and mean score of celebrity endorsement effectiveness.
Table 4.20: t-test for likening advertisement of snack food product and mean
score of celebrity endorsement effectiveness.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 0.622 0.430 -1.661 966 0.097
Equal variances not assumed 1.811 965.809 -1.635
Source: Primary data collected by the researcher
As shown in Table 4.20: the t-test results indicate that there is no significant
difference found in likening of celebrity and non-celebrity endorser for advertisement
of snack food product. Under the title of Levene‟s test for equality variances, the F
value was 0.622 with a Significance value of 0.430. The p value was not significant at
0.097, indicating the perceived difference of celebrity endorsement effectiveness and
likening advertisement of snack food product was not significant (p > 0.05). Thus,
hypothesis H02g was not rejected.
H02h: There is no significant difference between likening advertisement of chocolate
product and mean score of celebrity endorsement effectiveness.
Table 4.21: t-test for likening advertisement of chocolate product and mean score
of celebrity endorsement effectiveness.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 9.546 0.002 1.825 966 0.068
Equal variances not assumed 1.811 965.809 1.806
Source: Primary data collected by the researcher
98
As shown in Table 4.21: the t-test results indicate that there is no significant
difference found in likening of celebrity and non-celebrity endorser for advertisement
of chocolate product. Under the title of Levene‟s test for equality variances, the F
value was 9.546 with a Significance value of 0.002. The p value was not significant at
0.068, indicating the perceived difference of celebrity endorsement effectiveness and
likening advertisement of chocolate product was not significant (p > 0.05). Thus,
hypothesis H02h was not rejected.
H02i: There is no significant difference between likening advertisement of tea and
mean score of celebrity endorsement effectiveness.
Table 4.22: t-test for likening advertisement of tea and mean score of celebrity
endorsement effectiveness.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 0.257 0.613 1.087 966 0.277
Equal variances not assumed 1.811 965.809 1.091
Source: Primary data collected by the researcher
As shown in Table 4.22: the t-test results indicate that there is no significant
difference found in likening of celebrity and non-celebrity endorser for advertisement
of tea. Under the title of Levene‟s test for equality variances, the F value was 0.257
with a Significance value of 0.613. The p value was not significant at 0.277,
indicating the perceived difference of celebrity endorsement effectiveness and
likening advertisement of tea was not significant (p > 0.05). Thus, hypothesis H02i
was not rejected.
99
H02j: There is no significant difference between likening advertisement of soft drink
product and mean score of celebrity endorsement effectiveness.
Table 4.23: t-test for likening advertisement of soft drink product and mean
score of celebrity endorsement effectiveness.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 0.038 0.846 1.566 966 0.118
Equal variances not assumed 1.811 965.809 1.571
Source: Primary data collected by the researcher
As shown in Table 4.23: the t-test results indicate that there is no significant
difference found in likening of celebrity and non-celebrity endorser for advertisement
of soft drink product. Under the title of Levene‟s test for equality variances, the F
value was 0.038 with a Significance value of 0.846. The p value was not significant at
0.118, indicating the perceived difference of celebrity endorsement effectiveness and
likening advertisement of soft drink product was not significant (p > 0.05). Thus,
hypothesis H02j was not rejected.
H02k: There is no significant difference between likening advertisement of life
insurance and mean score of celebrity endorsement effectiveness.
Table 4.24: t-test for likening advertisement of life insurance and mean score of
celebrity endorsement effectiveness.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 3.745 0.053 2.586 966 0.010
Equal variances not assumed 1.811 965.809 2.510
Source: Primary data collected by the researcher
As shown in Table 4.24: the t-test results indicate that there is a significant difference
found in likening of celebrity and non-celebrity endorser for advertisement of life
insurance. Under the title of Levene‟s test for equality variances, the F value was
0.038 with a Significance value of 0.846. The p value was significant at 0.010,
100
indicating the perceived difference of celebrity endorsement effectiveness and
likening advertisement of life insurance was significant (p < 0.05). Thus, hypothesis
H02k was rejected.
H02l: There is no significant difference between likening advertisement of mobile
handset and mean score of celebrity endorsement effectiveness.
Table 4.25: t-test for likening advertisement of mobile handset and mean score of
celebrity endorsement effectiveness.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 0.853 0.356 -2.294 966 0.022
Equal variances not assumed 1.811 965.809 -2.343
Source: Primary data collected by the researcher
As shown in Table 4.25: the t-test results indicate that there is a significant difference
found in likening of celebrity and non-celebrity endorser for advertisement of mobile
handset. Under the title of Levene‟s test for equality variances, the F value was 0.853
with a Significance value of 0.356. The t value was significant at 0.022, indicating the
perceived difference of celebrity endorsement effectiveness and likening
advertisement of mobile handset was significant (p < 0.05). Thus, hypothesis H02k
was rejected.
101
4.10.2 Summary of effectiveness of celebrity on product preferences
Table 4.26: Summary of effectiveness of celebrity on product preferences.
Source: Primary data collected by the researcher
Hypothesis statement F-value P value Result
H02a There is no significant difference between likening
advertisement of fabric care product and mean score of
celebrity endorsement effectiveness.
0.140 0.194 Not
Rejected
H02b There is no significant difference between likening of
advertisement home care product and mean score of
celebrity endorsement effectiveness.
7.355 0.336 Not
Rejected
H02c There is no significant difference between likening
advertisement of oral care product and mean score of
celebrity endorsement effectiveness.
0.748 0.071 Not
Rejected
H02d There is no significant difference between likening
advertisement of skin care product and mean score of
celebrity endorsement effectiveness.
0.762 0.567 Not
Rejected
H02e There is no significant difference between likening
advertisement of hair care product and mean score of
celebrity endorsement effectiveness.
2.239 0.049 Rejected.
H02f There is no significant difference between likening
advertisement of bakery product and mean score of
celebrity endorsement effectiveness.
0.102 0.079 Not
Rejected
H02g There is no significant difference between likening
advertisement of snack food product and mean score of
celebrity endorsement effectiveness.
0.622 0.097 Not
Rejected
H02h There is no significant difference between likening
advertisement of chocolate and mean score of celebrity
endorsement effectiveness.
9.546 0.068 Not
Rejected
H02i There is no significant difference between likening
advertisement of tea and mean score of celebrity
endorsement effectiveness.
0.257 0.277 Not
Rejected
H02j There is no significant difference between likening
advertisement of soft drink product and mean score of
celebrity endorsement effectiveness.
0.038 0.118 Not
Rejected
H02k There is no significant difference between likening
advertisement of life insurance and mean score of
celebrity endorsement effectiveness
3.745 0.010 Rejected.
H02l There is no significant difference between likening
advertisement of mobile handset and mean score of
celebrity endorsement effectiveness.
0.853 0.022 Rejected.
102
4.11 t-Tests for Demographic Factors
H03a: There is no significance difference between gender group and celebrity‟s
physical attractiveness.
Table: 4.27 t-test for gender group and celebrity’s physical attractiveness.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 3.257 0.071 -1.180 966 0.238
Equal variances not assumed -1.195 958.328 0.232
Source: Primary data collected by the researcher
As shown in Table 4.27 the t-test results indicate that there is no significant difference
between a gender of the respondents and celebrity‟s physical attractiveness. Under the
title of Levene‟s test for equality variances, the F value was 3.257 with a Significance
value of 0.071. This value indicated that the variances of celebrity‟s physical
attractiveness were not equally distributed both groups of male and female. The t
value (-1.180) was not significant at 0.238, indicating the mean difference between
the male and female groups by celebrity‟s physical attractiveness was not significant
(p > 0.05). Thus, hypothesis H03a was not rejected.
H03b: There is no significance difference between gender group and celebrity‟s
trustworthiness.
Table: 4.28 t-tests for gender group and celebrity’s trustworthiness.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 3.073 0.080 -4.451 966 0.000
Equal variances not assumed -4.470 938.177 0.000
Source: Primary data collected by the researcher
As shown in Table 4.28 the t-test results indicate that there is a significant difference
between a gender of the respondents and celebrity‟s trustworthiness. Under the title of
Levene‟s test for equality variances, the F value was 3.073 with a Significance value
of 0.080. The t value (-4.451) was significant at 0.000, indicating the mean difference
103
between the male and female groups by celebrity‟s trustworthiness was significant (p
< 0.05). It indicates that the respondents view regarding celebrity‟s trustworthiness
was associated by gender wise. Thus, hypothesis H03b was rejected.
H03c: There is no significance difference between gender group and celebrity‟s
expertise.
Table: 4.29 t-tests for gender group and celebrity’s expertise.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 2.296 0.130 0.130 966 0.000
Equal variances not assumed -4.000 937.27 0.000
Source: Primary data collected by the researcher
As shown in Table 4.29 the t-test results indicate that there is a significant difference
between a gender of the respondents and celebrity‟s expertise. Under the title of
Levene‟s test for equality variances, the F value was 2.296 with a Significance value
of 0.130. This value indicated that the variances of celebrity‟s expertise were not
equally distributed both groups of male and female. The t value (0.130) was
significant at 0.000, indicating the mean difference between the male and female
groups by celebrity‟s expertise was significant (p < 0.05). It indicates that the
respondents view regarding celebrity‟s expertise was associated by gender wise. Thus,
hypothesis H03c was rejected.
H03d: There is no significance difference between gender group and celebrity‟s
likability.
Table: 4.30 t-test for Gender and celebrity’s likability.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 0.686 0.408 -4.448 966 0.000
Equal variances not assumed -4.465 936.964 0.000
Source: Primary data collected by the researcher
104
As shown in Table 4.30 the t-test results indicate that there is a significant difference
between a gender of the respondents and celebrity‟s likability. Under the title of
Levene‟s test for equality variances, the F value was 0.686 with a Significance value
of 0.408. This value indicated that the variances of celebrity‟s likability were not
equally distributed both groups of male and female. The t value (-4.448) was
significant at 0.000, indicating the mean difference between the male and female
groups by celebrity‟s likability was significant (p < 0.05). It indicates that the
respondents view regarding celebrity‟s likability was associated with gender. Thus,
hypothesis H03d was rejected.
H03e: There is no significance difference between gender group and non-celebrity‟s
physical attractiveness.
Table: 4.31 t-test for Gender group and non-celebrity’s physical attractiveness.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 1.428 0.232 -2.215 966 0.027
Equal variances not assumed -2.203 902.814 0.028
Source: Primary data collected by the researcher
As shown in Table 4.31 the t-test results indicate that there is a significant difference
between a gender of the respondents and non-celebrity‟s physical attractiveness
characteristics. Under the title of Levene‟s test for equality variances, the F value was
1.428 with a Significance value of 0.232. This value indicated that the variances of
non-celebrity‟s physical attractiveness were not equally distributed both groups of
male and female. The t value (-2.215) was significant at 0.027, indicating the mean
difference between the male and female groups by non-celebrity‟s physical
attractiveness was significant (p < 0.05). It indicates that the respondents view
regarding non-celebrity‟s physical attractiveness was associated gender wise. Thus,
hypothesis H03e was rejected.
105
H03f: There is no significance difference between gender group and non-celebrity‟s
trustworthiness.
Table: 4.32 t-test for Gender group and non-celebrity’s trustworthiness.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 6.539 0.011 -2.034 966 0.042
Equal variances not assumed -2.014 884.830 0.044
Source: Primary data collected by the researcher
As shown in Table 4.32 the t-test results indicate that there is a significant difference
between a gender of the respondents and non-celebrity‟s trustworthiness. Under the
title of Levene‟s test for equality variances, the F value was 6.539 with a Significance
value of 0.011. This value indicated that the variances of non-celebrity‟s
trustworthiness were not equally distributed both groups of male and female. The t
value (-2.034) was significant at 0.042, indicating the mean difference between the
male and female groups by non-celebrity‟s trustworthiness was significant (p <
0.05).n Thus, hypothesis H03f was rejected.
H03g: There is no significance difference between gender group and non-celebrity‟s
expertise.
Table: 4.33 t-test for Gender group and Non-celebrity’s expertise.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 20.185 0.000 -1.588 966 0.113
Equal variances not assumed -1.563 856.584 0.118
Source: Primary data collected by the researcher
As shown in Table 4.33 the t-test results indicate that there is no significant difference
between a gender of the respondents and non-celebrity‟s expertise. Under the title of
Levene‟s test for equality variances, the F value was 20.185 with a Significance value
of 0.000. This value indicated that the variances of non-celebrity‟s expertise were
equally distributed both groups of male and female. The t value (-1.588) was not
significant at 0.113, indicating the mean difference between the male and female
groups by Non-celebrity‟s expertise was not significant (p > 0.05). It indicates that the
106
respondents view regarding non-celebrity‟s expertise was not associated gender wise.
Thus, hypothesis H03g was not rejected.
H03h: There is no significance difference between gender group and non-celebrity‟s
likability.
Table: 4.34 t-test for Gender group and non-celebrity’s likability.
Levene's Test for
Equality of
Variances
Sig. t-test
value Df p value
Equal variances assumed 9.295 0.002 -1.040 966 0.299
Equal variances not assumed -1.030 884.706 0.304
Source: Primary data collected by the researcher
As shown in Table 4.34 the t-test results indicate that there is no significant difference
between a gender of the respondents and non-celebrity‟s likability characteristics.
Under the title of Levene‟s test for equality variances, the F value was 9.295 with a
Significance value of 0.002. This value indicated that the variances of non-celebrity‟s
likability were equally distributed both groups of male and female. The t value (-
1.040) was not significant at 0.299, indicating the mean difference between the male
and female groups by non-celebrity‟s likability was not significant (p > 0.05). It
indicates that the respondents view regarding non-celebrity‟s likability was not
associated with gender wise. Thus, hypothesis H03h was not rejected.
107
4.11.1 Summary of t-Tests for Demographic Factors by celebrity and non-
celebrity’s characteristics:
The hypothesis test of the gender demographic variable and celebrity / non-celebrity‟s
characteristics and factor of effectiveness in the advertisement was concluded.
Table: 4.35 Summary of t-Tests for demographic factors by celebrity and non-
celebrity’s characteristics.
Hypothesis statement F-value P value Result
H03a There is no significance difference between gender
group and celebrity‟s physical attractiveness. 3.257 0.238
Not
Rejected
H03b There is no significance difference between gender
group and celebrity‟s trustworthiness. 3.073 0.000 Rejected
H03c There is no significance difference between gender
group and celebrity‟s expertise. 2.296 0.000 Rejected
H03d There is no significance difference between gender
group and celebrity‟s likability. 0.686 0.000 Rejected
H03e There is no significance difference between gender
group and non-celebrity‟s physical attractiveness. 1.428 0.027 Rejected
H03f There is no significance difference between gender
group and non-celebrity‟s trustworthiness. 6.539 0.042 Rejected
H03g There is no significance difference between gender
group and non-celebrity‟s expertise. 20.185 0.113
Not
Rejected
H03h There is no significance difference between gender
group and non-celebrity‟s likability. 9.295 0.299
Not
Rejected
108
4.12 ANOVA test for Age group and celebrity / non-celebrity’s characteristics
H04a: There is no significance difference in mean scores of different age group in
case of celebrity‟s physical attractiveness.
Table 4.36: ANOVA test for age groups and celebrity’s physical attractiveness.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 31.062 4 7.766
6.817 0.000 Within Groups 1097.016 963 1.139
Total 1128.079 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.36, The F value with a
significance value of 0.000 indicated mean differences among the age groups were
significant (p < 0.05). Thus, hypothesis H04a was not accepted for given dataset. This
shows age group wise different perception among celebrity‟s physical attractiveness.
H04b: There is no significance difference in mean scores of different age group in
case of celebrity‟s Trustworthiness.
Table 4.37: ANOVA test for age group and celebrity’s trustworthiness.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 11.008 4 2.752
1.596 0.173 Within Groups 1660.776 963 1.725
Total 1671.784 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.37, The F value with a
significance value of 0.173 indicated mean differences among the age groups were
not significant (p > 0.05). Thus, hypothesis H04b was accepted for dataset. This shows
age group wise perception for celebrity‟s trustworthiness was same.
109
H04c: There is no significance difference in mean scores of different age group in
case of celebrity‟s Expertise.
Table 4.38: ANOVA test for age group and celebrity’s expertise.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 7.515 4 1.879
1.184 0.316 Within Groups 1527.661 963 1.586
Total 1535.176 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.38, The F value with a
significance value of 0.316 indicated mean differences among the age groups were
not significant (p > 0.05). Thus, hypothesis H04c was accepted for given dataset. This
shows age group wise perception for celebrity‟s expertise was same.
H04d: There is no significance difference in mean scores of different age group in
case of celebrity‟s Likability.
Table 4.39: ANOVA test for age group and celebrity’s likability.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 12.364 4 3.091
2.118 0.077 Within Groups 1405.486 963 1.459
Total 1417.851 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.39, The F value with a
significance value of 0.077 indicated mean differences among the age groups were
not significant (p > 0.05). Thus, hypothesis H04d was accepted for given dataset. This
shows age wise perception for celebrity‟s likability was same.
110
H04e: There is no significance difference in mean scores of different age group in
case of non-celebrity‟s physical attractiveness.
Table 4.40: ANOVA test for age group and non-celebrity’s physical
attractiveness.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 7.142 4 1.785
0.893 0.468 Within Groups 1925.744 963 2.000
Total 1932.886 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.40, The F value with a
significance value of 0.468 indicated mean differences among the age groups were
not significant (p > 0.05). Thus, hypothesis H04e was accepted for given dataset. This
shows age group wise perception for non-celebrity‟s physical attractiveness was
same.
H04f: There is no significance difference in mean scores of different age group in case
of non-celebrity‟s Trustworthiness.
Table 4.41: ANOVA test for age group and non-celebrity’s trustworthiness.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 16.694 4 4.174
1.810 0.125 Within Groups 2220.915 963 2.306
Total 2237.610 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.41, The F value with a
significance value of 0.125 indicated mean differences among the age groups were
not significant (p > 0.05). Thus, hypothesis H04f was accepted for given dataset. This
shows age group wise perception for non-celebrity‟s trustworthiness was same.
111
H04g: There is no significance difference in mean scores of different age group in
case of non-celebrity‟s Expertise.
Table 4.42: ANOVA test for age group and non-celebrity’s expertise.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 11.234 4 2.809
1.205 0.307 Within Groups 2245.429 963 2.332
Total 2256.663 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.42, The F value with a
significance value of 0.307 indicated mean differences among the age groups were
not significant (p > 0.05). Thus, hypothesis H04g was accepted for given dataset. This
shows age group wise perception for non-celebrity‟s expertise was same.
H04h: There is no significance difference in mean scores of different age group in
case of non-celebrity‟s Likability.
Table 4.43: ANOVA test for age group and non-celebrity’s likability.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 22.914 4 5.729
2.703 0.029 Within Groups 2041.226 963 2.120
Total 2064.140 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.43, The F value with a
significance value of 0.029 indicated mean differences among the age groups were
significant (p < 0.05). Thus, hypothesis H04h was not accepted for given dataset. This
shows age group wise different perception among non-celebrity‟s likability.
112
4.13 ANOVA test for Educational groups and Celebrity / non-celebrity’s
characteristics.
H05a: There is no significance difference in mean scores of different educational
group in case of celebrity‟s physical attractiveness.
Table 4.44: ANOVA test for educational groups and celebrity’s physical
attractiveness.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 28.203 3 9.401
8.240 0.000 Within Groups 1099.876 964 1.141
Total 1128.079 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.44, The F value with a
significance value of 0.000 indicated mean differences among the educational groups
were significant (p < 0.05). Thus, hypothesis H05a was not accepted for given dataset.
This shows educational groups wise perception for celebrity‟s physical attractiveness
was different.
H05b: There is no significance difference in mean scores of different educational
group in case of celebrity‟s Trustworthiness.
Table 4.45: ANOVA test for educational groups and celebrity’s trustworthiness.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 37.324 3 12.441
7.338 0.000 Within Groups 1634.460 964 1.695
Total 1671.784 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.45, The F value with a
significance value of 0.000 indicated mean differences among the educational groups
were significant (p < 0.05). Thus, hypothesis H05b was not accepted for given dataset.
This shows educational groupd wise perception for celebrity‟s trustworthiness was
different.
113
H05c: There is no significance difference in mean scores of different educational
group in case of celebrity‟s Expertise.
Table 4.46: ANOVA test for educational groups and celebrity’s expertise.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 50.358 3 16.786
10.898 0.000 Within Groups 1484.819 964 1.540
Total 1535.176 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.46, The F value with a
significance value of 0.000 indicated mean differences among the educational groups
were significant (p < 0.05). Thus, hypothesis H05c was not accepted for given dataset.
This shows educational groups wise perception for celebrity‟s expertise was different.
H05d: There is no significance difference in mean scores of different educational
group in case of celebrity‟s Likability.
Table 4.478: ANOVA test for educational groups and celebrity’s likability.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 8.733 3 2.911
1.991 0.114 Within Groups 1409.118 964 1.462
Total 1417.851 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.47, The F value with a
significance value of 0.114 indicated mean differences among the educational groups
were not significant (p > 0.05). Thus, hypothesis H05d was accepted for given dataset.
This shows educational group wise perception for celebrity‟s likability was not
different.
114
H05e: There is no significance difference in mean scores of different educational
group in case of non-celebrity‟s physical attractiveness.
Table 4.48: ANOVA test for educational groups and non-celebrity’s physical
attractiveness.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 4.858 3 1.619
0.810 0.489 Within Groups 1928.028 964 2.000
Total 1932.886 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.48, The F value with a
significance value of 0.489 indicated mean differences among the educational groups
were not significant (p > 0.05). Thus, hypothesis H05e was accepted for given dataset.
This shows educational group wise perception for non-celebrity‟s physical
attractiveness was not different.
H05f: There is no significance difference in mean scores of different educational
group in case of non-celebrity‟s Trustworthiness.
Table 4.49: ANOVA test for educational groups and non-celebrity’s
trustworthiness.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 9.416 3 3.139
1.358 0.254 Within Groups 2228.194 964 2.311
Total 2237.610 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.49, The F value with a
significance value of 0.254 indicated mean differences among the educational groups
were significant (p > 0.05). Thus, hypothesis H05f was accepted for given dataset.
This shows educational group wise perception for non-celebrity‟s trustworthiness was
not different.
115
H05g: There is no significance difference in mean scores of different educational
group in case of non-celebrity‟s Expertise.
Table 4.50: ANOVA test for educational groups and non-celebrity’s expertise.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 4.881 3 1.627
0.696 0.554 Within Groups 2251.782 964 2.336
Total 2256.663 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.50, The F value with a
significance value of 0.554 indicated mean differences among the educational groups
were not significant (p > 0.05). Thus, hypothesis H05g was accepted for given dataset.
This shows educational group wise perception for non-celebrity‟s expertise was not
different.
H05h: There is no significance difference in mean scores of different educational
group in case of non-celebrity‟s Likability.
Table 4.51: ANOVA test for educational groups and non-celebrity’s likability.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups .742 3 0.247
0.116 0.951 Within Groups 2063.398 964 2.140
Total 2064.140 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.51, The F value with a
significance value of 0.951 indicated mean differences among the educational groups
were not significant (p > 0.05). Thus, hypothesis H05h was accepted for given dataset.
This shows educational group wise perception for non-celebrity‟s likability was not
different.
116
4.14 ANOVA test for occupational groups and Celebrity / non-celebrity’s
characteristics.
H06a: There is no significance difference in mean scores of different occupational
group in case of celebrity‟s physical attractiveness.
Table 4.52: ANOVA test for occupational groups and celebrity’s physical
attractiveness.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 27.849 5 5.570
4.870 0.000 Within Groups 1100.230 962 1.144
Total 1128.079 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.52, The F value with a
significance value of 0.000 indicated mean differences among the occupational
groups were significant (p < 0.05). Thus, hypothesis H06a was not accepted for given
dataset. This shows occupational group wise perception for celebrity‟s physical
attractiveness was different.
H06b: There is no significance difference in mean scores of different occupational
group in case of celebrity‟s Trustworthiness.
Table 4.53: ANOVA test for occupational groups about celebrity’s
trustworthiness.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 54.242 5 10.848
6.452 0.000 Within Groups 1617.542 962 1.681
Total 1671.784 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.53, The F value with a
significance value of 0.000 indicated mean differences among the occupational
groups were significant (p < 0.05). Thus, hypothesis H06b was not accepted for given
dataset. This shows occupational group wise perception for celebrity‟s trustworthiness
was different.
117
H06c: There is no significance difference in mean scores of different occupational
group in case of celebrity‟s Expertise.
Table 4.54: ANOVA test for occupational groups and celebrity’s expertise.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 43.115 5 8.623
5.560 0.000 Within Groups 1492.061 962 1.551
Total 1535.176 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.54, The F value with a
significance value of 0.000 indicated mean differences among the occupational
groups were significant (p < 0.05). Thus, hypothesis H06c was not accepted for given
dataset. This shows occupational group wise perception for celebrity‟s expertise was
different.
H06d: There is no significance difference in mean scores of different occupational
group in case of celebrity‟s Likability.
Table 4.55: ANOVA test for occupational groups and celebrity’s likability.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 29.822 5 5.964
4.134 0.001 Within Groups 1388.029 962 1.443
Total 1417.851 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.55, The F value with a
significance value of 0.001 indicated mean differences among the occupational
groups were significant (p < 0.05). Thus, hypothesis H06d was not accepted for given
dataset. This shows occupational group wise perception for celebrity‟s likability was
different.
118
H06e: There is no significance difference in mean scores of different occupational
group in case of non-celebrity‟s physical attractiveness.
Table 4.56: ANOVA test for occupational groups and non-celebrity’s physical
attractiveness.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 14.667 5 2.933
1.471 0.197 Within Groups 1918.219 962 1.994
Total 1932.886 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.56, The F value with a
significance value of 0.197 indicated mean differences among the occupational
groups were not significant (p > 0.05). Thus, hypothesis H06e was accepted for given
dataset. This shows occupation wise perception for non-celebrity‟s physical
attractiveness was not different.
H06f: There is no significance difference in mean scores of different occupational
group in case of non-celebrity‟s Trustworthiness.
Table 4.57: ANOVA test for occupational groups and non-celebrity’s
trustworthiness.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 20.306 5 4.061
1.762 0.118 Within Groups 2217.304 962 2.305
Total 2237.610 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.57, The F value with a
significance value of 0.118 indicated mean differences among the occupational
groups were not significant (p > 0.05). Thus, hypothesis H06f was accepted for
dataset. This shows occupation wise perception for non-celebrity‟s trustworthiness
was not different.
H06g: There is no significance difference in mean scores of different occupational
group in case of non-celebrity‟s Expertise.
119
Table 4.58: ANOVA test for occupational groups and non-celebrity’s expertise.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 17.817 5 3.563
1.531 0.177 Within Groups 2238.846 962 2.327
Total 2256.663 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.58, The F value with a
significance value of 0.177 indicated mean differences among the occupational
groups were not significant (p > 0.05). Thus, hypothesis H06g was accepted for
dataset. This shows occupation wise perception for non-celebrity‟s expertise was not
different.
H06h: There is no significance difference in mean scores of different occupational
group in case of non-celebrity‟s Likability.
Table 4.59: ANOVA test for occupational groups and non-celebrity’s likability.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 16.359 5 3.272
1.537 0.176 Within Groups 2047.781 962 2.129
Total 2064.140 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.59, The F value with a
significance value of 0.176 indicated mean differences among the occupational
groups were not significant (p > 0.05). Thus, hypothesis H06h was accepted for given
dataset. This shows occupational group wise perception for non-celebrity‟s likability
was not different.
120
4.15 ANOVA test for Income groups and Celebrity / non-celebrity’s
characteristics.
H07a: There is no significance difference in mean scores of different income group in
case of celebrity‟s physical attractiveness.
Table 4.60: ANOVA test for income groups and celebrity’s physical
attractiveness.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 22.337 4 5.584
4.863 0.001 Within Groups 1105.741 963 1.148
Total 1128.079 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.60, The F value with a
significance value of 0.001 indicated mean differences among the income groups
were significant (p < 0.05). Thus, hypothesis H07a was not accepted for given dataset.
This shows income group wise perception for celebrity‟s physical attractiveness was
different.
H07b: There is no significance difference in mean scores of different income group in
case of celebrity‟s Trustworthiness.
Table 4.61: ANOVA test for income groups and celebrity’s trustworthiness.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 51.923 4 12.981
7.717 0.000 Within Groups 1619.861 963 1.682
Total 1671.784 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.61, The F value with a
significance value of 0.000 indicated mean differences among the income groups
were significant (p < 0.05). Thus, hypothesis H07b was not accepted for given dataset.
This shows income group wise perception for celebrity‟s trustworthiness was
different.
H07c: There is no significance difference in mean scores of different income group in
case of celebrity‟s Expertise.
121
Table 4.62: ANOVA test for income groups and celebrity’s expertise.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 39.900 4 9.975
6.424 0.000 Within Groups 1495.276 963 1.553
Total 1535.176 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.62, The F value with a
significance value of 0.000 indicated mean differences among the income groups
were significant (p < 0.05). Thus, hypothesis H07c was not accepted for given dataset.
This shows income group wise perception for celebrity‟s expertise was different.
H07d: There is no significance difference in mean scores of different income group in
case of celebrity‟s Likability.
Table 4.63: ANOVA test for income groups and celebrity’s likability.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 42.427 4 10.607
7.426 0.000 Within Groups 1375.424 963 1.428
Total 1417.851 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.63, The F value with a
significance value of 0.000 indicated mean differences among the income groups
were significant (p < 0.05). Thus, hypothesis H07d was not accepted for given dataset.
This shows income group wise perception for celebrity‟s likability was different.
H07e: There is no significance difference in mean scores of different income group in
case of non-celebrity‟s physical attractiveness.
Table 4.64: ANOVA test for income groups and non-celebrity’s physical
attractiveness.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 2.070 4 0.517
0.258 0.905 Within Groups 1930.817 963 2.005
Total 1932.886 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.64, The F value with a
significance value of 0.905 indicated mean differences among the income groups
122
were not significant (p >0.05). Thus, hypothesis H07e was accepted for given dataset.
This shows income group wise perception for non-celebrity‟s physical attractiveness
was not different.
H07f: There is no significance difference in mean scores of different income group in
case of non-celebrity‟s Trustworthiness.
Table 4.65: ANOVA test for income groups and non-celebrity’s trustworthiness.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 13.903 4 3.476
1.505 0.199 Within Groups 2223.707 963 2.309
Total 2237.610 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.65, The F value with a
significance value of 0.199 indicated mean differences among the income groups
were not significant (p >0.05). Thus, hypothesis H07f was accepted for given dataset.
This shows income group wise perception for non-celebrity‟s trustworthiness was not
different.
H07g: There is no significance difference in mean scores of different income group in
case of non-celebrity‟s Expertise.
Table 4.66: ANOVA test for income groups and non-celebrity’s expertise.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 10.816 4 2.704
1.159 0.327 Within Groups 2245.847 963 2.332
Total 2256.663 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.66, The F value with a
significance value of 0.327 indicated mean differences among the income groups
were not significant (p >0.05). Thus, hypothesis H07g was accepted for given dataset.
This shows income group wise perception for non-celebrity‟s expertise was not
different.
123
H07h: There is no significance difference in mean scores of different income group in
case of non-celebrity‟s Likability.
Table 4.67: ANOVA test for income groups and non-celebrity’s likability.
ANOVA
Sum of Squares df Mean Square F Sig.
Between Groups 15.587 4 3.897
1.832 0.121 Within Groups 2048.553 963 2.127
Total 2064.140 967
Source: Primary data collected by the researcher
The results of the one-way ANOVA are shown in table 4.67, The F value with a
significance value of 0.121 indicated mean differences among the income groups
were not significant (p >0.05). Thus, hypothesis H07h was accepted for given dataset.
This shows income group wise perception for non-celebrity‟s likability was not
different.
Table 4.68: Summary of ANOVA test.
Hypothesis statement F P value Result
H04a
There is no significance difference in mean scores of
different age group in case of celebrity‟s physical
attractiveness.
6.817 0.000 Not
Accepted
H04b There is no significance difference in mean scores of
different age group in case of celebrity‟s Trustworthiness 1.596 1.730 Accepted
H04c There is no significance difference in mean scores of
different age group in case of celebrity‟s Expertise. 1.184 0.316 Accepted
H04d There is no significance difference in mean scores of
different age group in case of celebrity‟s Likability. 2.118 0.077 Accepted
H04e
There is no significance difference in mean scores of
different age group in case of non-celebrity‟s physical
attractiveness.
0.893 0.468 Accepted
H04f
There is no significance difference in mean scores of
different age group in case of non-celebrity‟s
Trustworthiness.
1.810 0.125 Accepted
H04g There is no significance difference in mean scores of
different age group in case of non-celebrity‟s Expertise. 1.205 0.307 Accepted
124
H04h There is no significance difference in mean scores of
different age group in case of non-celebrity‟s Likability. 2.703 0.029
Not
Accepted
H05a
There is no significance difference in mean scores of
different educational group in case of celebrity‟s physical
attractiveness.
8.240 0.000 Not
Accepted
H05b
There is no significance difference in mean scores of
different educational group in case of celebrity‟s
Trustworthiness.
7.338 0.000 Not
Accepted
H05c There is no significance difference in mean scores of
different educational group in case of celebrity‟s Expertise. 10.898 0.000
Not
Accepted
H05d There is no significance difference in mean scores of
different educational group in case of celebrity‟s Likability. 1.991 0.114 Accepted
H05e
There is no significance difference in mean scores of
different educational group in case of non-celebrity‟s
physical attractiveness.
0.810 0.489 Accepted
H05f
There is no significance difference in mean scores of
different educational group in case of non-celebrity‟s
Trustworthiness.
1.358 0.254 Accepted
H05g
There is no significance difference in mean scores of
different educational group in case of non-celebrity‟s
Expertise.
0.696 0.544 Accepted
H05h
There is no significance difference in mean scores of
different educational group in case of non-celebrity‟s
Likability.
0.116 0.951 Accepted
H06a
There is no significance difference in mean scores of
different occupational group in case of celebrity‟s physical
attractiveness.
4.870 0.000 Not
Accepted
H06b
There is no significance difference in mean scores of
different occupational group in case of celebrity‟s
Trustworthiness.
6.452 0.000 Not
Accepted
H06c There is no significance difference in mean scores of
different occupational group in case of celebrity‟s Expertise. 5.560 0.000
Not
Accepted
125
Source: Primary data collected by the researcher
H06d There is no significance difference in mean scores of different
occupational group in case of celebrity‟s Likability. 4.134 0.001
Not
Accepted
H06e
There is no significance difference in mean scores of different
occupational group in case of non-celebrity‟s physical
attractiveness.
1.471 0.197 Accepted
H06f There is no significance difference in mean scores of different
occupational group in case of non-celebrity‟s Trustworthiness. 1.762 0.118 Accepted
H06g There is no significance difference in mean scores of different
occupational group in case of non-celebrity‟s Expertise. 1.531 0.177 Accepted
H06h There is no significance difference in mean scores of different
occupational group in case of non-celebrity‟s Likability. 1.537 0.176 Accepted
H07a There is no significance difference in mean scores of different
income group in case of celebrity‟s physical attractiveness. 4.863 0.001
Not
Accepted
H07b There is no significance difference in mean scores of different
income group in case of celebrity‟s trustworthiness. 7.717 0.000
Not
Accepted
H07c There is no significance difference in mean scores of different
income group in case of celebrity‟s Expertise. 6.424 0.000
Not
Accepted
H07d There is no significance difference in mean scores of different
income group in case of celebrity‟s Likability. 7.426 0.000
Not
Accepted
H07e
There is no significance difference in mean scores of different
income group in case of non-celebrity‟s physical
attractiveness.
0.258 0.905 Accepted
H07f There is no significance difference in mean scores of different
income group in case of non-celebrity‟s Trustworthiness. 1.505 0.199 Accepted
H07g There is no significance difference in mean scores of different
income group in case of non-celebrity‟s Expertise. 1.159 0.327 Accepted
H07h There is no significance difference in mean scores of different
income group in case of non-celebrity‟s Likability. 1.832 0.121 Accepted
126
4.16 Multiple Regression Analysis for Purchase Behavior and characteristics of
celebrity and non-celebrity:
The previous section tested the hypothesis for demographic variables by
characteristics of celebrity and non-celebrity. This part of the analysis presents the
process of the multiple regression analysis on purchase behavior across different
product categories. Multiple regression analysis was employed to examine the
relationships in the hypotheses developed regarding celebrity and non celebrity‟s
physical attractiveness, trustworthiness, expertise and likability characteristics.
Another hypothesis relating to purchase behavior and advertisement effectiveness.
The products are Detergent, Home cleaning, Oral care, Skin care, Hair care, Bakery
products, Snack food, Chocolates, Tea, Soft drink and Life insurance.
4.16.1 Hypothesis test for purchasing behavior and Celebrity’s factors
H08a: There is no significant relationship between purchasing behaviors and
celebrity‟s characteristics in case of detergent products.
Dependent Variable (Y): Purchasing behavior of products.
Independent Variable (X): Factors affecting celebrity endorsement.
For purchasing behavior of detergent products, as shown in the table 4.69 under the
title of analysis of variance, this percentage was statistically significant (F = 4.623,
p<0.05). There was no threat of multicollinearity among the predictor variables as the
value of tolerance for each of the independent variables was more than 0.2 and the
VIF (Variance Inflation Factor) was less than 5 (Churchill et al.,2010).
Table: 4.69 Analysis of variance for detergent products.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 13.219 4 3.305
4.623 0.001 Residual 688.397 963 0.715
Total 701.616 967
Source: Primary data collected by the researcher
As seen in the table 4.70, the intercept (identified as the constant) was significant at
the level of 0.05 (B = 2.480; t = 15.388). From all four, only one regression
127
coefficients were statistically significant (p < 0.05). This coefficient implied that there
is a significant relationship exists between celebrity‟s expertise characteristics.
Table: 4.70 Regression analysis of detergent products.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.480 15.388 0.000
Celebrity's Physical
Attractiveness 0.030 0.038 1.064 0.287 0.784 1.275
Celebrity's
Trustworthiness 0.018 0.028 0.562 0.574 0.398 2.510
Celebrity's Expertise -0.111 -0.165 -3.368 0.001 0.427 2.343
Celebrity's Likability 2.480 -0.004 -0.095 0.924 0.475 2.106
Source: Primary data collected by the researcher
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
Among the four factors, likability had the highest beta score (β = 0.030, p >0.05),
which is not signifying. While, trustworthiness is not significant characteristics. (β = -
0.004, p >0.05). The negative beta of likability indicated that these construct had a
negative relationship with purchasing behavior of detergent product.
H08b: There is no significant relationship between purchasing behaviors and
celebrity‟s characteristics in case of home cleaning products.
For purchasing behavior of home cleaning products, as shown in the table 4.71 under
the title of analysis of variance, this percentage was statistically significant (F=1.997,
p>0.05). There was a threat of multicollinearity among the predictor variables as the
value of tolerance for each of the independent variables was more than 0.2 and the
VIF (Variance Inflation Factor) was less than 5 (Churchill et al., 2010).
Table: 4.71 Analysis of variance for home cleaning products.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 5.146 4 1.286
1.997 0.093 Residual 620.288 963 0.644
Total 625.434 967
Source: Primary data collected by the researcher
128
As seen in the table 4.72, the intercept (identified as the constant) was significant at
the level of 0.05 (B=2.5460; t=16.643). From all four factors, none of regression
coefficients were statistically significant (p > 0.05). These coefficients implied that
there is no significant relationship exists between celebrity‟s characteristics. So, the
entire hypothesis was not rejected.
Table: 4.72 Regression analysis of home cleaning product.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.546 16.643 0.000
Celebrity's Physical
Attractiveness 0.015 0.020 0.563 0.573
0.784 1.275
Celebrity's
Trustworthiness -0.018 -0.030 -0.586 0.558
0.398 2.510
Celebrity's Expertise -0.039 -0.061 -1.246 0.213 0.427 2.343
Celebrity's Likability -0.012
-0.018 -0.378 0.706 0.475 2.106
Source: Primary data collected by the researcher
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
Among the four constructs are not signifying.
H08c: There is no significant relationship between purchasing behaviors and
celebrity‟s characteristics in case of oral care products.
For purchasing behavior of oral care products, as shown in the table 4.73 under the
title of analysis of variance, this percentage was statistically significant (F = 3.142,
p<0.05). There was no threat of multicollinearity among the predictor variables as the
value of tolerance for each of the independent variables was more than 0.2 and the
VIF (Variance Inflation Factor) was less than 5 (Churchill et al.,2010).
Table: 4.73 Analysis of variance for oral care products.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 9.961 4 2.490
3.142 0.014 Residual 763.163 963 0.792
Total 773.124 967
Source: Primary data collected by the researcher
129
As seen in the table 4.74, the intercept (identified as the constant) was significant at
the level of 0.05 (B=2.551; t=15.038). From all four characteristics, only one
regression coefficients were statistically significant (p < 0.05). This coefficient
implied that there is a significant relationship exists between celebrity‟s
trustworthiness characteristic. So, the hypothesis that the celebrity‟s only one
characteristic having relationship.
Table: 4.74 Regression analysis of oral care products.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.551 15.038 0.000
Celebrity's Physical
Attractiveness -0.002 -0.003 -0.078 0.938 0.784 1.275
Celebrity's
Trustworthiness -0.073 -0.108 -2.120 0.034 0.398 2.510
Celebrity's Expertise -0.010 -0.014 -0.286 0.775 0.427 2.343
Celebrity's Likability 0.006 0.008 0.180 0.857 0.475 2.106
Source: Primary data collected by the researcher
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
Among the four constructs, likability had the highest beta score (β = 0.006, p >0.05),
not signifying, on the purchasing behavior of the oral care product. While
Trustworthiness is signifying construct (β = -0.073, p <0.05). The negative beta of
likability indicated that these construct had a negative relationship with the oral care
product.
H08d: There is no significant relationship between purchasing behaviors and
celebrity‟s characteristics in case of skin care products.
For purchasing behavior of skin care products, as shown in the table 4.75 under the
title of analysis of variance, this percentage was statistically significant (F=5.257,
p<0.05). There was no threat of multicollinearity among the predictor variables as the
value of tolerance for each of the independent variables was more than 0.2 and the
VIF (Variance Inflation Factor) was less than 5 (Churchill et al.,2010).
130
Table: 4.75 Analysis of variance for skin care product.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 17.987 4 4.497
5.257 0.000 Residual 823.727 963 0.855
Total 841.715 967
Source: Primary data collected by the researcher
As seen in the table 4.76, the intercept (identified as the constant) was significant at
the level of 0.05 (B=2.645; t=15.005). From all four characteristics, only one
regression coefficients were statistically significant (p < 0.05). These coefficients
implied that there is a significant relationship exists between celebrity‟s
trustworthiness characteristics.
Table: 4.76 Regression analysis of skin care product.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.645 15.005 0.000
Celebrity's Physical
Attractiveness 0.005 0.005 0.149 0.882 0.784 1.275
Celebrity's
Trustworthiness -0.095 -0.133 -2.638 0.008 0.398 2.510
Celebrity's Expertise -0.014 -0.019 -0.386 0.699 0.427 2.343
Celebrity's Likability -0.001 -0.002 -0.037 0.970 0.475 2.106
Source: Primary data collected by the researcher
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
Among the four factors, physical attractiveness had the highest beta score (β= 0.005, p
>0.05), which was not signifying and not influenced purchasing behavior of skin care
product. While Trustworthiness is signifying construct (β = -0.095, p <0.05). The
negative beta of likability indicated that these characteristic had a negative
relationship with the skin care product.
H08e: There is no significant relationship between purchasing behaviors and
celebrity‟s characteristics in case of hair care products.
For purchasing behavior of hair care products, as shown in the table 4.77 under the
title of analysis of variance, this percentage was statistically significant (F=4.942,
p<0.05). There was no threat of multicollinearity among the predictor variables as the
131
value of tolerance for each of the independent variables was more than 0.2 and the
VIF (Variance Inflation Factor) was less than 5 (Churchill et al.,2010).
Table: 4.77 Analysis of variance for hair care product.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 16.626 4 4.156
4.942 0.001 Residual 809.883 963 0.841
Total 826.508 967
Source: Primary data collected by the researcher
As seen in the table 4.78, the intercept (identified as the constant) was significant at
the level of 0.05 (B=2.476; t=14.168). From all four factors, none of regression
coefficients were statistically significant (p > 0.05). These coefficients implied that
there is no significant relationship exists between celebrity‟s all four characteristics.
So all the four hypothesis were not rejected
Table: 4.78 Regression analysis of hair care product.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.476 14.168 0.000
Celebrity's Physical
Attractiveness 0.045 0.053 1.473 0.141 0.784 1.275
Celebrity's
Trustworthiness -0.050 -0.071 -1.404 0.161 0.398 2.510
Celebrity's Expertise -0.031 -0.042 -0.865 0.387 0.427 2.343
Celebrity's Likability -0.049 -0.064 -1.373 0.170 0.475 2.106
Source: Primary data collected by the researcher
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
Among the four factors, physical attractiveness had the highest beta score (β = 0.045,
p >0.05), and not signifying, while Trustworthiness, expertise and likability were also
not signifying construct (β = -0.050, -0.031, -0.049 p >0.05). The negative beta of
likability indicated that these construct had a negative relationship with the hair care
product.
H08f: There is no significant relationship between purchasing behaviors and
celebrity‟s characteristics in case of bakery products.
132
For purchasing behavior of bakery product, as shown in the table 4.79 under the title
of analysis of variance, this percentage was statistically significant (F = 3.484,
p<0.05). There was no threat of multicollinearity among the predictor variables as the
value of tolerance for each of the independent variables was more than 0.2 and the
VIF (Variance Inflation Factor) was less than 5.
Table: 4.79 Analysis of variance for bakery product.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 10.518 4 2.630
3.484 0.008 Residual 726.788 963 0.755
Total 737.306 967
Source: Primary data collected by the researcher
As seen in the table 4.80, the intercept (identified as the constant) was significant at
the level of 0.05 (B = 2.352; t = 14.203). From all four factors, only one regression
coefficients were statistically significant (p < 0.05). This coefficient implied that there
is a significant relationship exists between celebrity‟s likability characteristics.
Table: 4.80 Regression analysis of bakery product.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.352 14.203 0.000
Celebrity's Physical
Attractiveness 0.053 0.066 1.822 0.069 0.784 1.275
Celebrity's
Trustworthiness -0.050 -0.075 -1.484 0.138 0.398 2.510
Celebrity's Expertise 0.037 0.054 1.093 0.275 0.427 2.343
Celebrity's Likability -0.071 -0.098 -2.109 0.035 0.475 2.106
Source: Primary data collected by the researcher
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
Among the four characteristics, physical attractiveness had the highest beta score (β =
0.053, p >0.05), this was not signifying and not influenced purchasing behavior of the
bakery products. While likability is signifying construct (β = -0.071 p <0.05). The
negative beta of likability indicated that this characteristic had a negative relationship
with the bakery product.
133
H08g: There is no significant relationship between purchasing behaviors and
celebrity‟s characteristics in case of snack food.
For purchasing behavior of snack product, as shown in the table 4.81 under the title of
analysis of variance, this percentage was statistically not significant (F=2.326,
p>0.05). There was a threat of multicollinearity among the predictor variables as the
value of tolerance for each of the independent variables was more than 0.2 and the
VIF (Variance Inflation Factor) was less than 5.
Table: 4.81 Analysis of variance for snack food.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 6.795 4 1.699
2.326 0.055 Residual 703.270 963 0.730
Total 710.065 967
Source: Primary data collected by the researcher
As seen in the table 4.82, the intercept (identified as the constant) was significant at
the level of 0.05 (B=2.519; t=15.468). From all four factors, none of regression
coefficients were statistically significant (p > 0.05). These coefficients implied that
there is no significant relationship exists between celebrity‟s factors. So all the four
hypotheses were not rejected
Table: 4.83 Regression analysis for snack food.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.519 15.468 0.000
Celebrity's Physical
Attractiveness 0.003 0.003 0.095 0.925 0.784 1.275
Celebrity's
Trustworthiness -0.044 -0.067 -1.317 0.188 0.398 2.510
Celebrity's Expertise 0.002 0.003 0.069 0.945 0.427 2.343
Celebrity's Likability -0.031 -0.043 -0.934 0.351 0.475 2.106
Source: Primary data collected by the researcher
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
Among the four characteristics, expertise characteristic had highest beta score (β =
0.003, p >0.05), which was not signifying and not influenced on purchasing behavior
of snack food. While Trustworthiness, and likability were also not signifying
134
characteristics (β = -0.044, -0.031 p >0.05). The negative beta indicated that these
construct had a negative relationship with the snack food.
H08h: There is no significant relationship between purchasing behaviors and
celebrity‟s characteristics in case of chocolate product.
For purchasing behavior of chocolate product, as shown in the table 4.83 under the
title of analysis of variance, this percentage was statistically not significant
(F=1.554, p>0.05). There was a threat of multicollinearity among the predictor
variables as the value of tolerance for each of the independent variables was more
than 0.2 and the VIF (Variance Inflation Factor) was less than 5.
Table: 4.83 Analysis of variance for chocolate product.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 4.292 4 1.073
1.554 0.185 Residual 664.898 963 0.690
Total 669.190 967
Source: Primary data collected by the researcher
As seen in the table 4.84, the intercept (identified as the constant) was significant at
the level of 0.05 (B=2.338; t=14.762). From all four factors, none of regression
coefficients were statistically significant (p > 0.05). These coefficients implied that
there is no significant relationship exists between celebrity‟s characteristics. So all the
four hypotheses were not rejected.
Table: 4.84 Regression analysis for chocolate product.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.338 14.762 0.000
Celebrity's Physical
Attractiveness -0.019 -0.024 -0.665 0.506 0.784 1.275
Celebrity's
Trustworthiness -0.044 -0.070 -1.374 0.170 0.398 2.510
Celebrity's Expertise 0.023 0.034 0.697 0.486 0.427 2.343
Celebrity's Likability -0.018 -0.026 -0.560 0.575 0.475 2.106
Source: Primary data collected by the researcher
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
135
Among the four characteristics, expertise characteristic had the highest beta score
(β=0.023, p >0.05) which was not signifying and not influenced on purchasing
behavior of chocolate product. While Trustworthiness, expertise and likability were
also not signifying construct (β = -0.044, -0.018 p >0.05). The negative beta indicated
that these construct had a negative relationship with the chocolate product.
H08i: There is no significant relationship between purchasing behaviors and
celebrity‟s characteristics in case of tea.
For purchasing behavior of tea, as shown in the table 4.85 under the title of analysis of
variance, this percentage was statistically significant (F=2.587, p<0.05).
Table: 4.85 Analysis of variance for tea product.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 8.475 4 2.119
2.587 0.036 Residual 788.760 963 0.819
Total 797.236 967
Source: Primary data collected by the researcher
As seen in the table 4.86, the intercept (identified as the constant) was significant at
the level of 0.05 (B = 2.192; t = 12.708). From all four factors, two of regression
coefficients were statistically significant (p<0.05). These coefficients implied that
there is significant relationship exists between celebrity‟s physical attractiveness and
expertise characteristics. So, two hypotheses were not rejected.
Table: 4.86 Regression analyses for tea product.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.192 12.708 0.000
Celebrity's Physical
Attractiveness 0.064 0.076 2.110 0.035 0.784 1.275
Celebrity's
Trustworthiness -0.035 -0.050 -0.991 0.322 0.398 2.510
Celebrity's Expertise -0.074 -0.103 -2.090 0.037 0.427 2.343
Celebrity's Likability 0.060 0.080 1.715 0.087 0.475 2.106
Source: Primary data collected by the researcher
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
Among the four characteristics, physical attractiveness had the highest beta score (β =
136
0.064, p<0.05) which was signifying and influenced on purchasing behavior of tea.
While Trustworthiness, and likability were not signifying characteristics (β = -0.035,
0.060 p >0.05).
H08j: There is no significant relationship between purchasing behaviors and
celebrity‟s characteristics in case of soft drink.
For purchasing behavior of soft drink, as shown in the table 4.87 under the title of
analysis of variance, this percentage was statistically significant (F=2.547, p<0.05).
There was no threat of multicollinearity among the predictor variables as the value of
tolerance for each of the independent variables was more than 0.2 and the VIF
(Variance Inflation Factor) was less than 5.
Table: 4.87 Analysis of variance for soft drink product.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 9.315 4 2.329
2.547 0.038 Residual 880.452 963 0.914
Total 889.768 967
Source: Primary data collected by the researcher
As seen in the table 4.88, the intercept (identified as the constant) was significant at
the level of 0.05 (B=2.611; t=14.329). From all four characteristics, none of
regression coefficients were statistically significant (p >0.05). These coefficients
implied that there were no significant relationship exists between celebrity‟s all
characteristics. So all hypothesis were not rejected.
Table: 4.88 Regression analysis of soft drink product.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.611 14.329 0.000
Celebrity's Physical
Attractiveness -0.002 -0.002 -0.054 0.957 0.784 1.275
Celebrity's
Trustworthiness -0.042 -0.057 -1.120 0.263 0.398 2.510
Celebrity's Expertise -0.065 -0.085 -1.729 0.084 0.427 2.343
Celebrity's Likability 0.040 0.051 1.090 0.276 0.475 2.106
Source: Primary data collected by the researcher
137
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
Among the four characteristics, likability characteristic had the highest beta score (β =
0.044, p>0.05), which was not signifying and not influenced on purchasing behavior
of soft drink.
H08k: There is no significant relationship between purchasing behaviors and
celebrity‟s characteristics in case of life insurance.
For purchasing behavior of life insurance, as shown in the table 4.89 under the title of
analysis of variance, this percentage was statistically significant (F=1.554, p<0.05).
There was a threat of multicollinearity among the predictor variables as the value of
tolerance for each of the independent variables was more than 0.2 and the VIF
(Variance Inflation Factor) was less than 5 (Churchill et al.,2010).
Table: 4.89 Analysis of variance for insurance.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 19.125 4 4.781
4.878 0.001 Residual 943.817 963 0.980
Total 962.942 967
Source: Primary data collected by the researcher
As seen in the table 4.90, the intercept (identified as the constant) was significant at
the level of 0.05 (B=2.858; t=15.147). From all four characteristics, none of
regression coefficients were statistically significant (p > 0.05). These coefficients
implied that there is no significant relationship exists between celebrity‟s
characteristics. So all the four hypotheses were not rejected
Table: 4.90 Regression analysis for life insurance.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.858 15.147 0.000
Celebrity's Physical
Attractiveness 0.063 0.068 1.895 0.058 0.784 1.275
Celebrity's
Trustworthiness -0.005 -0.006 -0.124 0.902 0.398 2.510
Celebrity's Expertise -0.060 -0.076 -1.561 0.119 0.427 2.343
Celebrity's Likability -0.074 -0.090 -1.947 0.052 0.475 2.106
Source: Primary data collected by the researcher
138
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
Among the four factors, physical attractiveness had the highest beta score (β = 0.063,
p >0.05) which was not signifying and not influenced on purchasing behavior of life
insurance. While other characteristics were also not signifying (β = -0.005, -0.060, -
0.074 p >0.05). The negative beta indicated that these construct had a negative
relationship with the insurance.
4.16.2: Hypothesis test for purchasing behavior and non-celebrity’s
characteristics:
H09a: There is no significant relationship between purchasing behaviors and non-
celebrity‟s characteristics in case of detergent product.
For purchasing behavior of detergent product for non-celebrity, as shown in the table
4.91 under the title of analysis of variance, this percentage was statistically not
significant (F = 0.802, p>0.05).
Table: 4.91 Analysis of variance for detergent product.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 4 0.582
0.802 0.524 Residual 699.287 963 0.726
Total 701.616 967
Source: Primary data collected by the researcher
As seen in the table 4.92, the intercept (identified as the constant) was significant at
the level of 0.05 (B = 2.220; t = 20.226). From all four characteristics, none of the
regression coefficients were statistically significant (p > 0.05). These coefficients
implied that there is no significant relationship exists between non-celebrity‟s
characteristics. So the hypothesis that all the non-celebrity‟s characteristics were not
rejected.
139
Table: 4.92 Regression analyses for detergent product.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.220 20.226 0.000
Non-celebrity's
Physical Attractiveness 0.021 0.034 0.787 0.431 0.552 1.813
Non-celebrity's
Trustworthiness 0.001 0.002 0.025 0.980 0.256 3.908
Non-celebrity's
Expertise -0.034 -0.061 -0.935 0.350 0.240 4.160
Non-celebrity's
Likability -0.008 -0.014 -0.252 0.801 0.317 3.153
Source: Primary data collected by the researcher
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
Among the four characteristics, physical attractiveness had the highest beta score (β =
0.021, p >0.05) which was not signifying and not influenced on purchasing behavior
of detergent. While Trustworthiness, expertise and likability were also not signifying
characteristics.
H09b: There is no significant relationship between purchasing behaviors and non-
celebrity‟s characteristics in case of home cleaning product.
For purchasing behavior of home cleaning product for non-celebrity, as shown in the
table 4.93, under the title of analysis of variance, this percentage was not statistically
significant (F = 1.29, p>0.05). There was no threat of multicollinearity among the
predictor variables as the value of tolerance for each of the independent variables was
more than 0.2 and the VIF (Variance Inflation Factor) was less than 5 (Churchill et
al.,2010).
Table: 4.93 Analysis of variance for home cleaning product.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 4.459 4 1.115
1.729 0.141 Residual 620.975 963 0.645
Total 625.434 967
Source: Primary data collected by the researcher
As seen in the table 4.94, the intercept (identified as the constant) was significant at
the level of 0.05 (B=2.285; t=22.088). From all four characteristics, only one
140
regression coefficients was statistically significant (p < 0.05). This coefficient implied
that there is a significant relationship exists among non-celebrity‟s expertise
characteristic.
Table: 4.94 Regression analyses for home cleaning product.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.285 22.088 0.000
Non-celebrity's
Physical Attractiveness 0.030 0.053 1.222 0.222 0.552 1.813
Non-celebrity's
Trustworthiness 0.025 0.047 0.736 0.462 0.256 3.908
Non-celebrity's
Expertise -0.082 -0.155 -2.368 0.018 0.240 4.160
Non-celebrity's
Likability 0.022 0.040 0.707 0.480 0.317 3.153
Source: Primary data collected by the researcher
H09c: There is no significant relationship between purchasing behaviors and non-
celebrity‟s characteristics in case of oral care product.
For purchasing behavior of oral care product for non-celebrity, as shown in the table
4.95 under the title of analysis of variance, this percentage was statistically not
significant (F = 0.539, p>0.05). There was no threat of multicollinearity among the
predictor variables as the value of tolerance for each of the independent variables was
more than 0.2 and the VIF (Variance Inflation Factor) was less than 5 (Churchill et
al.,2010).
Table: 4.95 Analysis of variance for oral care product.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 1.728 4 0.432
0.539 0.707 Residual 771.396 963 0.801
Total 773.124 967
Source: Primary data collected by the researcher
As seen in the table 4.96, the intercept (identified as the constant) was significant at
the level of 0.05 (B=2.558; t=19.589). From all four characteristics, none of the
regression coefficients were statistically significant (p > 0.05). These coefficients
implied that there is no significant relationship exists between non-celebrity‟s
characteristics. So, the entire hypotheses were not rejected.
141
Table: 4.96 Regression analysis for oral care product.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.258 19.589 0.000
Non-celebrity's Physical
Attractiveness 0.001 0.001 0.025 0.980 0.552 1.813
Non-celebrity's
Trustworthiness -0.025 -0.043 -0.677 0.498 0.256 3.908
Non-celebrity's
Expertise -0.007 -0.011 -0.172 0.863 0.240 4.160
Non-celebrity's
Likability 0.004 0.006 0.110 0.912 0.317 3.153
Source: Primary data collected by the researcher
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
Among the four characteristics, likability had the highest beta score (β = 0.004, p
>0.05), which was not signifying and not influenced on purchasing behavior of an
oral care product.
H09d: There is no significant relationship between purchasing behaviors and non-
celebrity‟s characteristics in case of skin care product.
For purchasing behavior of skin care product for non-celebrity, as shown in the table
4.97 under the title of analysis of variance, this percentage was not statistically
significant (F = 1.251, p>0.05). There was a threat of multicollinearity among the
predictor variables as the value of tolerance for each of the independent variables was
more than 0.2 and the VIF (Variance Inflation Factor) was less than 5 (Churchill et
al.,2010).
Table: 4.97 Analysis of variance for skin care product.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 4.350 4 1.088
1.251 0.288 Residual 837.365 963 0.870
Total 841.715 967
Source: Primary data collected by the researcher
As seen in the table 4.98, the intercept (identified as the constant) was significant at
the level of 0.05 (B=2.215; t=18.441). From all four characteristics, none of the
regression coefficients were statistically significant (p > 0.05). These coefficients
142
implied that there is no significant relationship exists between non-celebrity‟s
characteristics. So, the entire hypotheses were not rejected.
Table: 4.98 Regression analyses for skin care product.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.215 18.441 0.000
Non-celebrity's
Physical Attractiveness 0.006 0.008 0.196 0.845 0.552 1.813
Non-celebrity's
Trustworthiness -0.050 -0.082 -1.295 0.196 0.256 3.908
Non-celebrity's
Expertise -0.028 -0.046 -0.702 0.483 0.240 4.160
Non-celebrity's
Likability 0.048 0.075 1.310 0.190 0.317 3.153
Source: Primary data collected by the researcher
H09e: There is no significant relationship between purchasing behaviors and non-
celebrity‟s characteristics in case of hair care product.
For purchasing behavior of hair care product for non-celebrity, as shown in the table
4.99 under the title of analysis of variance, this percentage was statistically significant
(F=2.509, p<0.05). There was no threat of multicollinearity among the predictor
variables as the value of tolerance for each of the independent variables was more
than 0.2 and the VIF (Variance Inflation Factor) was less than 5 (Churchill et
al.,2010).
Table: 4.99 Analysis of variance for hair care product.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 8.523 4 2.131
2.509 0.041 Residual 817.985 963 0.849
Total 826.508 967
Source: Primary data collected by the researcher
As seen in the table 4.100, the intercept (identified as the constant) was significant at
the level of 0.05 (B=2.218; t=18. 681). From all four, none of regression coefficients
were statistically significant (p > 0.05). These coefficients implied that there is no
significant relationship exists between non-celebrity‟s characteristics. So all the four
hypotheses were not rejected
143
Table: 4.100 Regression analysis for hair care product.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.218 18.681 0.000
Non-celebrity's
Physical Attractiveness 0.043 0.066 1.524 0.128 0.552 1.813
Non-celebrity's
Trustworthiness -0.051 -0.084 -1.323 0.186 0.256 3.908
Non-celebrity's
Expertise -0.044 -0.072 -1.101 0.271 0.240 4.160
Non-celebrity's
Likability 0.014 0.022 0.384 0.701 0.317 3.153
Source: Primary data collected by the researcher
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
Among the four characteristics, physical attractiveness had the highest beta score (β =
0.043, p >0.05), While Trustworthiness, expertise and likability were also not
signifying characteristics (β = -0.051, -0.044, -0.014 p >0.05). The negative beta of
likability indicated that these construct had a negative relationship with the hair care
product.
H09f: There is no significant relationship between purchasing behaviors and non-
celebrity‟s characteristics in case of bakery product.
For purchasing behavior of bakery product for non-celebrity, as shown in the table
4.101 under the title of analysis of variance, this percentage was statistically not
significant (F = 1.896, p>0.05). There was no threat of multicollinearity among the
predictor variables as the value of tolerance for each of the independent variables was
more than 0.2 and the VIF (Variance Inflation Factor) was less than 5 (Churchill et
al.,2010).
Table: 4.101 Analysis of variance for bakery product.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 5.762 4 1.441
1.896 0.109 Residual 731.543 963 0.760
Total 737.306 967
Source: Primary data collected by the researcher
144
As seen in the table 4.102, the intercept (identified as the constant) was significant at
the level of 0.05 (B = 2.328; t = 20.739). From all four, two regression coefficients
were statistically significant (p < 0.05). This coefficient implied that there is a
significant relationship exists between celebrity‟s expertise and likability
characteristics.
Table: 4.102 Regression analyses for bakery product.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.328 20.739 0.000
Non-celebrity's
Physical Attractiveness 0.001 0.002 0.035 0.972 0.552 1.813
Non-celebrity's
Trustworthiness -0.041 -0.071 -1.112 0.266 0.256 3.908
Non-celebrity's
Expertise 0.080 0.140 2.137 0.033 0.240 4.160
Non-celebrity's
Likability -0.068 -0.114 -2.006 0.045 0.317 3.153
Source: Primary data collected by the researcher
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
Among the four characteristics, expertise factor had the highest beta score (β = 0.080,
p<0.05) which was signifying and influenced on purchasing behavior of the bakery
products. While likability factor is also signifying constructs (β = -0.068 p<0.05). The
negative beta of likability indicated that this factor had a negative relationship with
the bakery product.
H09g: There is no significant relationship between purchasing behaviors and non-
celebrity‟s characteristics in case of snack food product.
For purchasing behavior of snack food products for non-celebrity, as shown in the
table 4.103 under the title of analysis of variance, this percentage was statistically
significant (F = 3.639, p<0.05). There was a threat of multicollinearity among the
predictor variables as the value of tolerance for each of the independent variables was
more than 0.2 and the VIF (Variance Inflation Factor) was less than 5 (Churchill et
al.,2010).
145
Table: 4.103 Analysis of variance for snack food.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 10.572 4 2.643
3.639 0.006 Residual 699.493 963 0.726
Total 710.065 967
Source: Primary data collected by the researcher
As seen in the table 4.104, the intercept (identified as the constant) was significant at
the level of 0.05 (B=2.442; t=22.245). From all four, only one regression coefficients
were statistically significant (p < 0.05). This coefficient implied that there is a
significant relationship exists between non-celebrity‟s likability characteristic.
Table: 4.104 Regression analyses for snack food.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.442 22.245 0.000
Non-celebrity's
Physical Attractiveness 0.012 0.019 0.443 0.658 0.552 1.813
Non-celebrity's
Trustworthiness -0.046 -0.081 -1.283 0.200 0.256 3.908
Non-celebrity's
Expertise 0.041 0.072 1.107 0.268 0.240 4.160
Non-celebrity's
Likability -0.070 -0.120 -2.114 0.035 0.317 3.153
Source: Primary data collected by the researcher
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
Among the four characteristics, expertise factor had the highest beta score (β = 0.041,
p <0.05), which was signifying and influenced on purchasing behavior of snack food.
While physical attractiveness, Trustworthiness, and likability were also not signifying
characteristics (β = 0.012, -0.046, -.070 p >0.05). The negative beta indicated that
these factors had a negative relationship.
H09h: There is no significant relationship between purchasing behaviors and non-
celebrity‟s characteristics in case of chocolate.
For purchasing behavior of chocolate products for non-celebrity, as shown in the table
4.105 under the title of analysis of variance, this percentage was statistically
significant (F = 2.769, p<0.05). There was a threat of multicollinearity among the
146
predictor variables as the value of tolerance for each of the independent variables was
more than 0.2 and the VIF (Variance Inflation Factor) was less than 5 (Churchill et
al.,2010).
Table: 4.105 Analysis of variance for chocolate products.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 7.609 4 1.902
2.769 0.026 Residual 661.581 963 0.687
Total 669.190 967
Source: Primary data collected by the researcher
As seen in the table 4.106, the intercept (identified as the constant) was significant at
the level of 0.05 (B=2.279; t=21.345). From all four, none of regression coefficients
were statistically significant (p > 0.05). These coefficients implied that there is no
significant relationship exists between non-celebrity‟s constructs. So all the four
hypotheses were not rejected
Table: 4.106 Regression analyses for chocolate product.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.279 21.345 0.000
Non-celebrity's
Physical Attractiveness 0.022 0.037 0.863 0.388 0.552 1.813
Non-celebrity's
Trustworthiness -0.029 -0.054 -0.850 0.396 0.256 3.908
Non-celebrity's
Expertise -0.013 -0.024 -0.368 0.713 0.240 4.160
Non-celebrity's
Likability -0.033 -0.059 -1.033 0.302 0.317 3.153
Source: Primary data collected by the researcher
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
Among the four characteristics, physical attractiveness had the highest beta score (β =
0.022, p >0.05), While Trustworthiness, expertise and likability have beta score β = -
0.029, -0.013, -0.033. The negative beta indicated that these construct had a negative
relationship with the chocolate product.
147
H09i: There is no significant relationship between purchasing behaviors and non-
celebrity‟s characteristics in case of tea.
For purchasing behavior of tea products for non-celebrity, as shown in the table 4.107
under the title of analysis of variance, this percentage was statistically not significant
(F = 2.017, p>0.05). There was no threat of multicollinearity among the predictor
variables as the value of tolerance for each of the independent variables was more
than 0.2 and the VIF (Variance Inflation Factor) was less than 5 (Churchill et
al.,2010).
Table: 4.107 Analysis of variance for tea product.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 6.622 4 1.656
2.017 0.090 Residual 790.613 963 0.821
Total 797.236 967
Source: Primary data collected by the researcher
As seen in the table 4.108, the intercept (identified as the constant) was significant at
the level of 0.05 (B=2.490; t=21.334). From all four factors, none of regression
coefficients were statistically significant (p <0.05). These coefficients implied that
there is a significant relationship exists between non-celebrity‟s characteristics. So all
hypothesis were not rejected.
Table: 4.108 Regression analyses for tea product.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.490 21.334 0.000
Non-celebrity's
Physical Attractiveness 0.018 0.028 0.641 0.522 0.552 1.813
Non-celebrity's
Trustworthiness 0.003 0.005 0.080 0.936 0.256 3.908
Non-celebrity's
Expertise -0.020 -0.034 -0.521 0.602 0.240 4.160
Non-celebrity's
Likability -0.050 -0.080 -1.402 0.161 0.317 3.153
Source: Primary data collected by the researcher
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
148
Among the four characteristics, physical attractiveness had the highest beta score (β =
0.018, p>0.05).
H09j: There is no significant relationship between purchasing behaviors and non-
celebrity‟s characteristics in case of soft drink.
For purchasing behavior of soft drink products for non-celebrity, as shown in the table
4.109 under the title of analysis of variance, this percentage was statistically not
significant (F=2.275, p>0.05). There was no threat of multicollinearity among the
predictor variables as the value of tolerance for each of the independent variables was
more than 0.2 and the VIF (Variance Inflation Factor) was less than 5 (Churchill et
al.,2010).
Table: 4.109 Analysis of variance for soft drink product.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 2.275 4 0.569
0.617 0.650 Residual 887.492 963 0.922
Total 889.768 967
Source: Primary data collected by the researcher
As seen in the table 4.110, the intercept (identified as the constant) was significant at
the level of 0.05 (B = 2.371; t = 19.174). From all four characteristics, none of
regression coefficients were statistically significant (p >0.05). These coefficients
implied that there no significant relationship exists between non-celebrity‟s all
characteristics. So all hypothesis were not rejected.
Table: 4.110 Regression analyses for soft drink product.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.371 19.174 0.000
Non-celebrity's Physical
Attractiveness 0.006 0.009 0.209 0.835 0.552 1.813
Non-celebrity's
Trustworthiness -0.007 -0.011 -0.167 0.867 0.256 3.908
Non-celebrity's
Expertise -0.036 -0.057 -0.870 0.385 0.240 4.160
Non-celebrity's
Likability 0.009 0.014 0.238 0.812 0.317 3.153
Source: Primary data collected by the researcher
149
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
Among the four characteristics, likability had the highest beta score (β = 0.009,
p>0.05).
H09k: There is no significant relationship between purchasing behaviors and non-
celebrity‟s characteristics in case of life insurance.
For purchasing behavior of Life insurance for non-celebrity, as shown in the table
4.111 under the title of analysis of variance, this percentage was statistically
significant (F=3.617, p<0.05). There was a threat of multicollinearity among the
predictor variables as the value of tolerance for each of the independent variables was
more than 0.2 and the VIF (Variance Inflation Factor) was less than 5 (Churchill et
al.,2010).
Table: 4.111 Analysis of variance for Life insurance.
Sum of
Squares Df
Mean
Square F
Sig.
(p value)
Regression 14.251 4 3.563
3.617 0.006 Residual 948.691 963 0.985
Total 962.942 967
Source: Primary data collected by the researcher
As seen in the table 4.112, the intercept (identified as the constant) was significant at
the level of 0.05 (B=2.859; t=22.365). From all four, none of regression coefficients
were statistically significant (p > 0.05). These coefficients implied that there is no
significant relationship exists between non-celebrity‟s characteristics. So all the four
hypotheses were not rejected
150
Table: 4.112 Regression analysis for Life insurance.
Regression
Coefficients
B Beta
Standardized
Coefficients β
Beta
t value p
value
Collinearity
Statistics
Tolerance
VIF
(Constant) 2.859 22.365 0.000
Non-celebrity's
Physical Attractiveness 0.011 0.016 0.365 0.715 0.552 1.813
Non-celebrity's
Trustworthiness -0.035 -0.053 -0.835 0.404 0.256 3.908
Non-celebrity's
Expertise -0.024 -0.036 -0.559 0.576 0.240 4.160
Non-celebrity's
Likability -0.036 -0.052 -0.915 0.360 0.317 3.153
Source: Primary data collected by the researcher
The standardized coefficient β (beta) can be used to compare directly the relative
effect of each independent variable on the dependent variable (Hair et al., 2010).
Among the four characteristics, physical attractiveness had the highest beta score (β =
0.011, p >0.05).
Table 4.113: Summary of Regression analysis for hypothesis between product
categories among characteristics of celebrity / non-celebrity.
Hypothesis statement Result
H08a
There is no significant relationship between purchasing
behaviors and celebrity‟s characteristics in case of detergent
product.
Partially
supported
H08b
There is no significant relationship between purchasing
behaviors and celebrity‟s characteristics in case of home
cleaning product.
Not rejected
H08c
There is no significant relationship between purchasing
behaviors and celebrity‟s characteristics in case of oral care
product.
Partially
supported
H08d
There is no significant relationship between purchasing
behaviors and celebrity‟s characteristics in case of skin care
product.
Partially
supported
151
H08e
There is no significant relationship between purchasing
behaviors and celebrity‟s characteristics in case of hair care
product.
Not rejected
H08f
There is no significant relationship between purchasing
behaviors and celebrity‟s characteristics in case of bakery
product.
Partially
supported
H08g
There is no significant relationship between purchasing
behaviors and celebrity‟s characteristics in case of snack food
product.
Not rejected
H08h There is no significant relationship between purchasing
behaviors and celebrity‟s characteristics in case of chocolate. Not rejected
H08i There is no significant relationship between purchasing
behaviors and celebrity‟s characteristics in case of tea.
Partially
supported
H08j There is no significant relationship between purchasing
behaviors and celebrity‟s characteristics in case of soft drink. Not rejected
H08k
There is no significant relationship between purchasing
behaviors and celebrity‟s characteristics in case of life
insurance.
Not rejected
H09a
There is no significant relationship between purchasing
behaviors and non-celebrity‟s characteristics in case of
detergent product.
Not rejected
H09b
There is no significant relationship between purchasing
behaviors and non-celebrity‟s characteristics in case of home
cleaning product.
Partially
supported
H09c
There is no significant relationship between purchasing
behaviors and non-celebrity‟s characteristics in case of oral
care product.
Not rejected
H09d
There is no significant relationship between purchasing
behaviors and non-celebrity‟s characteristics in case of skin
care product.
Not rejected
152
Source: Primary data collected by the researcher
H09e
There is no significant relationship between purchasing
behaviors and non-celebrity‟s characteristics in case of hair
care product.
Not rejected
H09f
There is no significant relationship between purchasing
behaviors and non-celebrity‟s characteristics in case of bakery
product.
Partially
supported
H09g
There is no significant relationship between purchasing
behaviors and non-celebrity‟s characteristics in case of snack
food product.
Not rejected
H09h
There is no significant relationship between purchasing
behaviors and non-celebrity‟s characteristics in case of
chocolate.
Not rejected
H09i There is no significant relationship between purchasing
behaviors and non-celebrity‟s characteristics in case of tea. Not rejected
H09j
There is no significant relationship between purchasing
behaviors and non-celebrity‟s characteristics in case of soft
drink.
Not rejected
H09k
There is no significant relationship between purchasing
behaviors and non-celebrity‟s characteristics in case of life
insurance.
Not rejected
153
4.17 Cross tabulation of Gender and Likening of advertisement by endorsers’
type.
Table: 4.114 Cross tabulation of Gender and Likening of advertisement.
Product category Likening of Endorser in advertisement Male Female
Fabric care Celebrity endorsement 35.9 43.4
Non-celebrity endorsement 64.1 56.6
Home cleaning Celebrity endorsement 62.8 58.7
Non-celebrity endorsement 37.2 41.3
Oral care Celebrity endorsement 48.8 48.5
Non-celebrity endorsement 51.2 51.5
Skin care Celebrity endorsement 70.5 66.7
Non-celebrity endorsement 29.5 33.3
Hair care Celebrity endorsement 71.2 73.0
Non-celebrity endorsement 28.8 27.0
Bakery product Celebrity endorsement 37.0 33.3
Non-celebrity endorsement 63.0 66.7
Snack food Celebrity endorsement 66.5 78.1
Non-celebrity endorsement 33.5 29.1
Chocolate Celebrity endorsement 44.5 50.1
Non-celebrity endorsement 55.5 49.9
Tea Celebrity endorsement 51.8 56.6
Non-celebrity endorsement 48.2 43.4
Soft drink Celebrity endorsement 71.0 70.2
Non-celebrity endorsement 29.0 29.8
Life insurance Celebrity endorsement 36.3 34.2
Non-celebrity endorsement 63.7 65.8
Mobile handset Celebrity endorsement 72.2 66.8
Non-celebrity endorsement 28.0 33.3
Source: Primary data collected by the researcher
Above table 119, depicts the result of cross tabulation of likening of advertisement for
different product categories. The pair of advertisements was given to respondents
including celebrity and non-celebrity endorser. In the above table gender wise
likening were given. For celebrity endorser advertisement for various product
154
categories, the difference were observed on fabric care, home cleaning, snack food
and chocolate, where as for non-celebrity endorsement the difference on gender wise
observed on fabric care, home cleaning, bakery product, chocolate and mobile
handset. The soft drink is one product in which there was less difference observed.
4.18 Conclusion:
This chapter started with a presentation of the descriptive statistic relevant to each
hypotheses of this study. The inferential statistics were then shown and the results of
independent sample t- test ANOVA and multiple regression analysis conducted to test
the study‟s hypotheses were presented. The next chapter discusses the result
discussion and findings of the study based data analysis.
155
References:
Churchill G. A., Lacobucci D. & Israel D. (2010). Marketing Research A South
Asian Perspective, (4th ed.). Delhi: Cengage Learning.
Sherman, S. P. (1985, August 19). When you wish upon a star. Fortune, 66-71
Zikmund W., Barry Babin, Jon Carr, Mitch Griffi “Business Research Methods”, Cengage
Learning, 2013.