The McDonald's Omega coefficient provides an accurate measure of reliability by excluding
some scale items and considering actual population estimates. The McDonald's Omega
coefficient is computed by dividing the portion of variation attributable to the common
characteristic by the overall variance in the data. There were several independent variables
such as perceived cost, perceived usefulness, perceived ease of use, emotional appeal and
attitude has been tested for the reliability analyze. While for the intention of adoption
smartwatches is the dependent variable used to test the reliability of data.
Table below shows that the Mcdonald’s Omega value for all the variable is greater than 0.8,
which mean that the internal reliability is good. Hence, there are no item deleted according to
all the variable, which indicate that the data collected are reliable to be used in the hypothesis
and inter-items consistency.
Therefore, the perceived usefulness variable is the highest in the Mcdonald’s Omega Value,
which is 0.951 and followed up by the intention of adoption smartwatches which got 0.923.
This show that the variable of perceived usefulness and intention to adoption smartwatches
have the higher reliable data in analyzing in the research than other variables. While, for the
lowest in Mcdonald’s Omega Value is the perceived cost which is 0.880.
4.5 Reliability of Measurement
Mcdonald Omega coefficient (ω) is calculated as the ratio of the variation attributed to the
common characteristic to the overall variance. The omega values provided the most precise
assessment of reliability (Edwards et al., 2021). The reliability analysis included testing the
independent variables, such as perceived cost, perceived usefulness, perceived ease of use,
emotional appeal, and the mediator, which is attitude. Intention to adopt smartwatch is the
dependent variable used to test the reliability of data.
The table presented in 4.2 shows that Mcdonald’s Omega value for all the variables is greater
than 0.8, indicating the internal reliability is consistently. There are no items deleted in all of
the variables, which means the provided data is valid for testing hypotheses and assessing
inter-items consistency.
The perceived cost and perceived ease of use have McDonald'Omega values of 0.880 and
0.899, respectively, indicating reasonably good internal consistency and reliability. Perceived
usefulness, emotional appeal, attitude and intention to adopt a smartwatch have a McDonald’s
Omega value above 0.9, indicating that these variables have high internal consistency and
reliability. Thus, a higher McDonald’s Omega value indicates greater internal consistency and
reliability.
Table 4.2
Reliability Analysis
Variable Items Item Deleted Mcdonald’s Omega
Perceived Cost 3 - 0.880
Perceived Usefulness 4 - 0.951
Perceived Ease of Use 3 - 0.899
Emotional Appeal 4 - 0.901
Attitude 4 - 0.908
Intention to adopt smartwatch 3 - 0.923
4.6 Descriptive of Main Variables
Table 4.3 shows a descriptive analysis of the main variables. All variables were measured
using a 7-point Likert scale.The mean for all the independent variables, the mediator and the
dependent variable ranged between a value of 5 to interpret the statements in this survey as
neutral and agreed upon by the respondents. The standard deviation indicates the dispersion
or variability of the data points relative to the mean. In this study, the standard deviation
derived from the dataset is considered a small standard deviation, with each variable having
an approximate value of 1.
Skewness measures the degree of asymmetry in the distribution of data. A skewness value
between -1 and +1 is considered excellent (Hair et al., 2022, p. 66). Kurtosis measures the
“peakedness” or “tailedness” of a distribution relative to a normal distribution. Values within
the range of -2 to +2 for asymmetry and kurtosis are generally considered to indicate a
normal univariate distribution. (George & Mallery, 2010). When the values of both skewness
and kurtosis are close to zero, it indicates that the distribution of responses follows a normal
distribution. (George & Mallery, 2019; Hair et al., 2022, p. 66) The variables for all the main
variables in this study exhibit a negative skewness, which indicates a left-skewn distribution
has a long tail on the left side of its peak. Variable, which is PC and EA has a negative
kurtosis, which indicates a distribution with lighter tails, while variable, which is PU, PEU,
AT and IN has a positive kurtosis, which indicates a distribution with relatively heavy tails
compared to a normal distribution.
Table 4.3
Descriptive Analysis of Main Variables
Variable Mean Std. Dev Skewness Kurtosis
PC 4.6444 1.2583 -0.096 -0.697
PU 5.4700 1.3530 -1.099 0.548
PEU 5.6778 1.1723 -1.277 0.870
EA 5.1183 1.3004 -0.621 -0.358
AT 5.7167 1.2029 -1.036 0.323
IN 5.5956 1.3997 -1.355 1.157
4.7 Correlation Analysis
Pearson’s correlation analysis is widely regarded as the most effective method of test
statistics because it measures the statistical relationship, or association, between two
continuous variables. The table in 4.3 illustrates the correlation among the main variables. A
correlation coefficient (r) with a value greater than 0.5 indicates a strong positive relationship
between two variables. The dependent variable, IN has a quite strong positive relationship
with the independent variables PU and PEU as well as the mediator variable AT, where PU =
0.604, PEU = 0.516, and AT = 0.656, while it has a moderate positive relationship with EA,
where EA=0.439 However, the dependent variable, IN has a quite strong negative
relationship with the independent variable PC, where PC = -0.032.
Table 4.4
Correlation of the Main Variables
Variable PC PU PEU EA AT IN
PC 1.000
PU 0.077 1.000
PEU 0.139 0.626 1.000
EA -0.062 0.635 0.302 1.000
AT 0.126 0.738 0.565 0.509 1.000
IN -0.032 0.604 0.516 0.439 0.656 1.000
4.8 Hypothesis Testing
Linear multiple regression is used to test the hypotheses developed for this research.Before
we can interpret the results, there are seven assumtions must be tested. The seven
assumptions tested are normality, normaliy of the error terms, constant variance, linearity,
multicollinearity, autocorrelation and outliers.
4.8.1 Assumption 1 – Normality
The histogram of the data set studied is distributed as a bell-shaped, with a mean of 0 and
standard deviation of 0.983 which is in close proximity to 1. Thus, it has been determined
that the distribution of this dataset is normal.
4.8.2 Assumption 2 – Normality of the Error Term
The P-P plot chart investigates that the points in the chart are very close to the line. This
assumption allows for valid hypothesis testing and confidence interval estimation.
4.8.3 Assumption 3 – Constant Variance
The scatterplot chart examines that the points are scattered which mean the data is scattered.
The result has no clear pattern of increasing or decreasing data.
4.8.4 Assumption 4 – Linearity
The partial regression plot chart demonstrates a linear relationship between the independent
variables and the dependent variable. This indicates that changes in the dependent variable
are proportional to changes in the independent variables, with constant slope coeffiicients.
The partial regression plot chart illustrates a direct and proportional correlation between the
independent variables and the dependent variable. This means that variations in the dependent
variable are directly proportional to variations in the independent variables, while
maintaining constant slope coefficients.
4.8.5 Assumption 5 – Multicollinearity
The multicollinearity problem does not happen in the multiple linear regression where it can
be observed from the Variance inflation factor (VIF) values which were below 5. So, there is
no problematic multicollinearity.
4.8.6 Assumption 6 – Outliers
Case 1 was detected and removed from the multiple regression as outliers because the values
of this cases fell away from 3 standard deviations.
4.8.7 Assumption 7 – Autocorrelation
The Durbin-Watson value of 1.824, indicate that there is no significant autocorrelation.
Autocorrelation in the residuals would not be a problem.
4.8.8 Hypothesis Testing
Table 4.5 presents the results of multiple regression analysis to test the hypotheses generated.
Based on the ANOVA results, the adjusted R2 is 0.494 and R2is 0.551. The R2 value
investigates that approximately 55.1 % variants of the can be explained by the independent
variables, while the 44.9% of the variance cannot be explained because there are other
variables that are not included in this model.
Table 4.5
Hypotheses Testing
Hypothesis Construct Unstd. Std.Err Std. B t- p-value LL UL VIF
B or val
ue
H1 PC -0.142 0.539 -0.128 - 0.027 - -0.037 1.049
2.2 0.4
30 49
H2 PU 0.138 0.064 0.130 1.3 0.192 - 0.312 3.153
10 0.2
48
H3 PEU 0.250 0.105 0.205 2.7 0.006 - 0.398 1.708
94 0.0
36
H4 EA 0.082 0.089 0.075 1.0 0.309 .01 0.215 1.737
21 02
H5 AT 0.558 0.080 0.465 5.5 P<0.00 - 0.725 2.220
62 0 0.0
51
Perceived Cost, PC, (β = -0.142, p<0.027) has negative on the intention to adopt
smartwatches because the beta value of PC was shown as negative. The consumers' attitude
positively influenced their intention, however higher perceived cost had a negative impact on
their intention. On the other hand, perceived ease of use, PEU (β = 0.250, p<0.006) and
attitude, AT (β = 0.558, p<0.000) are positively related and significant to the dependent
variable of intention to adopt smartwatches as the values of p were less than 0.05.
Besides that, there is no relationship for the dependent variables which is Perceived
usefulness, PU (β = 0.138, p<0.192) and Perceived ease of use, PEU (β = 0.250, p<0.006).
Both the dependent variables are not significant to the dependent variable IN, because the
value of p>0.05, Therefore, it fails to reject the null hypothesis, H2 and H4.
In conclusion, H3 and H5 are supported in this research while H1, H2 and H4 are not
supported in the findings of the research.
4.9 Bootstrapping
Table 4.6 shows the boostrap for coefficients. In this context, bootstrapping is used to
calculate the corrected standard errors using a 10,000-bootstrap sampling. Based on the
findings, the values of standardised beta remained unchanged. The bias value for all variables
is almost 0, indicating a satisfactory value. The standard error for variables such as H1, H3
and H4 shows increasing trends. However, independent variable H2 and dependent variable
H5 show declining trends compared to hypothesis testing. The T-value-valuee calculated by
using the formula B/Std Error, the result of significant values has changed to H1, H3 and H5,
which are acceptable, while the values for H2 and H4 are rejected. The lower limit and upper
bound values experienced slight fluctuations but remained near their values prior to the
bootstrapping stage.
Table 4.6
Boostrap for Coefficients
Hypothesis Construct B Bias Std. t-value p- LL UL
Error value
H1 PC -0.142 0.004 0.069 -2.058 0.046 -0.250 -0.026
H2 PU 0.138 0.004 0.130 -1.062 0.299 -0.065 0.366
H3 PEU 0.250 -0.006 0.096 2.604 0.010 0.087 0.410
H4 EA 0.082 0.005 0.084 0.976 0.343 -0.042 0.236
H5 AT 0.558 0.005 0.121 4.612 <0.00 0.353 0.765
1
4.10 Summary of Findings /
The following table 4.7 summarizes the hypothesis testing result using multiple regression
analysis.
Table 4.7
Summary of Findings
Hypothesis Relationship Decision
H1 Perceived Cost Intention Supported
H2 Perceived Usefulness Intention Not Supported
H3 Perceived Ease Of Use Intention Supported
H4 Emotional Appeal Intention Not Supported
H5 Attitude Intention Supported
