49

Chapter 4

4. Data Analysis and Results

This chapter presents the analysis of data collected through an online and offline survey method.

Data was systematically compiled and tabulated for the conducting statistical analysis.

Descriptive and inferential analysis were performed using SPSS 20.0 and SmartPLS3 software.

The chapter is comprised of six sections: Section 4.1 and 4.2 presents the preliminary analysis

that includes Data Description and Data screening, which are the pre-requisites for the further

statistical analysis. Section 4.3 provides detailed evaluation of measurement model and

structural model respectively using Partial Least Square- Structural Equation Modelling. Section

4.4 offers moderator analysis while Section 4.4 includes analysis of mediating effect of financial

risk tolerance on the relationship between psychological constructs and financial risk-taking

behaviour. Section 4.6 signifies the impact-performance matrix analysis providing insights into

possible areas of improvements.

4.1. Data Preparation

Data Preparation involves coding and entering of data into a database followed by filtration and

identification of missing responses. Since both online and offline methods were used for data

collection, the responses from a web-based survey created over Google Docs were

automatically captured in a downloadable spreadsheet, while offline responses collected through

questionnaires were added to the same spreadsheet manually. Upon preliminary scrutiny, a total

of 303 usable responses as received were uploaded to IBM SPSS 20.0 to generate descriptive

statistical reports and to check for missing responses, normality of data and response biasness as

a precautionary step. Later the spreadsheet was converted into CSV format required for PLS-

SEM analysis.

4.1.1. Descriptive Statistics of Respondents

Table 4.1 displays the brief demographic profile of respondents including gender, age,

education, marital status, occupation, annual household disposable income and years of

experience in investment.
50

Table 4.1 Demographic Profile of Respondents

Characteristics Frequency (N Percent Mean STDEV
= 303)
Gender Male 247 81.50 1.18 0.39
Female 56 18.50
Age <=45 years 241 79.50 1.20 0.40
>45 years 62 20.50
Occupation Private Sector 192 63.40 1.37 0.48
Public Sector 111 36.60
Annual Household <=INR10,00,000 174 57.40 1.43 0.50
Disposable Income >INR10,00,000 129 42.60
Education Professional 185 61.10 1.61 0.49
Non-professional 118 38.90
Marital Status Married 187 61.70 1.38 0.49
Unmarried 116 38.30
Experience in <= 5 years 130 42.90 1.57 0.50
Investment > 5 years 173 57.10
Notes: STDEV, Standard Deviation
Source: Compiled from survey data

The sample comprised of 303 respondents with majority of male (81.50%) while female

constituted only 18.5% of the sample. Participants were further divided into two age groups

(Levinson, D.J., 1986), young adults (age = 18 to 45 years; n = 241) and mature adults (age > 45

years; n = 62). 38.90% respondents in the sample belonged to non-professional educational

background while 61.10% respondents belonged to professional education category. With

regard to marital status, 61.70% of respondents were married while 38.30% respondents

belonged to unmarried category. Since the study included only salaried individuals; 63.40%

respondents in the sample belonged to private sector while 36.60% were employed in public

sector. As per the reports of National Council for Applied Economic Research (NCAER), the

real annual household disposable income of 50 million people (classified as middle class) falls

between 200,000 to 10,00,000 hence two groups were created from the sample, the group 1 (n =

174) include the respondents having annual household disposable income less than equal to INR

10,00,000 while respondents in group 2 (n = 129) were having annual household disposable

income above INR 10,00,000. Out of 303 respondents, 42.90% respondents had investment

experience of 5 years or less while 57.10% were having investment experience of more than 5

years.
 51

4.2. Data Screening

It includes verifying characteristics of collected data. The objective is to ensure that the data

before being utilized for multivariate analysis is valid and complete. Therefore, it includes:

• Estimation of missing values,

• Checking for normality of data

• Common Method Bias

4.2.1. Estimation of Missing Values

The data was examined through visual inspection. No missing values were found in the data set.

4.2.2. Normality of Data

Though PLS-SEM being a non-parametric statistical approach does not require data to be

normally distributed, however, it becomes imperative to check normality of data otherwise it

could create problems in the assessing significance of parameters. Since, the traditional

Kolmogorov-Smirnov test and Shapiro-Wilks test provide only limited guidance when deciding

whether the data is too far from being normally distributed which means whether to reject the

null hypothesis of normally distributed data or not.

The normality of data was examined using two measures of distribution, i.e., skewness and

kurtosis which are built on underlying rule of thumb (Hair, Black, Babin, Anderson, & Tatham,

2006) for the data to be considered as normally distributed.

𝐒𝐤𝐞𝐰𝐧𝐞𝐬𝐬−𝟎
𝐙𝐒𝐤𝐞𝐰𝐧𝐞𝐬𝐬 = 𝐒𝐭𝐚𝐧𝐝𝐚𝐫𝐝 𝐄𝐫𝐫𝐨𝐫 = ± 2.58 (for sample size greater than 200)

Similarly,

𝐊𝐮𝐫𝐭𝐨𝐬𝐢𝐬−𝟎
𝐙𝐊𝐮𝐫𝐭𝐨𝐬𝐢𝐬 = 𝐒𝐭𝐚𝐧𝐝𝐚𝐫𝐝 𝐄𝐫𝐫𝐨𝐫 = ± 2.58 (for sample size greater than 200)

The results reveal that all the variables were following the rule (refer Table 4.2), signifying that

the departure from normality is not too extreme.

Table 4.2: Test of Normality

Mean STDEV SKEWNESS KURTOSIS
Variables Statistic Statistic Statistic Standard Z Statistic Standard Z
Error Error
EI 3.74 0.51 0.05 0.14 0.32 -0.64 0.28 -2.29
52

IMP 2.50 0.43 -0.33 0.14 -2.35 0.10 0.28 0.35

FRT 3.03 0.84 0.34 0.14 2.43 -0.39 0.28 -1.39
FRB 2.90 0.79 0.23 0.14 1.67 0.67 0.28 2.39
Notes: EI, Emotional Intelligence; IMP, Impulsiveness; FRT, Financial Risk Tolerance; FRB,
Financial Risk Taking Behaviour; STDEV, Standard Deviation
Source: Calculation by Author

4.2.3. Common Method Variance

Common method variance is a measurement error (Podsakoff, MacKenzie, Lee, & Podsakoff,

2003; Podsakoff, MacKenzie, & Podsakoff, 2012) that threatens the validity of a conclusion

drawn upon statistical results. The common method variance may occur due to the following

reasons (Podsakoff et al., 2003):

• respondents,

• complexity and ambiguity of items in the questionnaire, and

• measurement context (e.g. time and location of measurement, common medium to

obtain measurement)

Hence, there lies a strong need to identify the causes of the common method variance to better

control for their influence on the data. Harman’s single factor test is the simplest and most

widely used technique (Podsakoff et al., 2003).

Table 4.3: Harman’s Single Factor Test

Component Initial Eigenvalues Extraction Sums of Squared Loadings

Total % of Cumulative Total % of Cumulative

Variance % Variance %
1 1.46 48.80 48.80 1.46 48.8 48.80
2 1.01 33.57 82.38
3 0.53 17.62 100
Notes: Extraction Method: Principal Component Analysis
Source: Calculation by Author

The resulting 48.80% variance (refer Table 4.3) explained by a single factor signifies that the

common method variance is not a major concern since the percent of variance lies below the

threshold value of 50%.

4.3. Partial Least Square-Structural Equation Modeling Assessment

PLS-SEM technique, initially created by Wold (1974, 1980, 1982), is an iterative approach that

maximizes the explained variance of endogenous constructs (Fornell & Bookstein, 1982). As
 53

compared to CB-SEM, which expects to affirm hypotheses by deciding how well a model can

estimate a covariance matrix for the sample data, PLS-SEM works much like multiple

regression analysis (Hair et al., 2011) and aims to develop theories, hence making it especially

important for exploratory research purposes.

A great part of the expanded use of PLS-SEM can be attributed to the technique's capacity to

deal with issues like non-normality of data- a usual problem of social sciences researchers,

small sample sizes and model complexities.

The application of PLS-SEM follows the three most salient steps:

i. model specification;

ii. outer or measurement model evaluation; and

iii. inner or structural model evaluation.

4.3.1. Model Specification

The model specification involves setting up of the inner and outer models. The inner models,

also known as structural model describes the relationships between the constructs (also known

as latent variables) being assessed. The outer models, also referred as the measurement models,

describes the relationships between the indicator variables and their corresponding construct.

The initial step in utilizing PLS-SEM encompasses creating a path model that associates

variables and constructs in view of rationale and theory (Hair et al., 2014). Figure 4.1 represents

the path model of the study as shown below. The model represents the second-order also known

as higher-order or hierarchical component model as usually referred in context of PLS-SEM

(Lohmoller,1989). The hierarchical component model involves simultaneous mapping of first

order constructs also known as lower-order constructs (LOCs) and higher order constructs

(HOCs).

Figure 4.1 represents the constructs SEA, OEA, UOE, ROE represents the LOCs of more

general construct Emotional Intelligence (EI) which is measured with twenty indicators SEA1,

SEA2, SEA3, SEA4, SEA5, OEA1, OEA2, OEA3, OEA4, OEA5, UOE1, UOE2, UOE3,

UOE4, UOE5, ROE1, ROE2, ROE3, ROE4, ROE5. Similarly, the constructs NU, LOPe, LOPr,
54

PU represents the LOCs of construct Impulsiveness (IMP) measured using twenty indicators

NU1, NU2, NU3, NU4, NU5, LOPe1, LOPe2, LOPe3, LOPe4, LOPe5, LOPr1, LOPr2, LOPr3,

LOPr4, LOPr5, PU1, PU2, PU3, PU4, PU5. Therefore, using repeated-indicators approach, all

indicators of reflectively measured LOCs were simultaneously assigned to reflective

measurement model of HOC.

 55

Figure 4.1: Path Model

Notes: EI, Emotional Intelligence; SEA, Self-Emotion Appraisal; OEA, Others Emotion
Appraisal; UOE, Use of Emotions; ROE, Regulation of Emotions; IMP, Impulsiveness; NU,
Negative Urgency; LOPe, Lack of Perseverance; LOPr, Lack of Premeditation; PU, Positive
Urgency, FRT, Financial Risk Tolerance; FRB, Financial Risk-taking Behaviour
56

4.3.2. Outer Model/Measurement Model Evaluation

This stage involves running the PLS-SEM algorithm to assess the reliability and validity of the

constructs (refer figure 4.2) which form the basis for evaluating the relationship of inner model.

That is, evaluation of measurement model delivers confirmatory test of measurement theory.

The measurement theory states how measurement variables represent the constructs specified in

the model.

RELIABILITY VALIDITY

•Indicator Reliability •Convergent Validity

•Internal Consistency •Discriminant Validity
Reliability

Figure 4.2: Steps to follow while evaluating Measurement Model

Source: Adapted from Hair et al. (2013)

[Link]. Reliability Analysis

Reliability analysis alludes to the fact that a scale should consistently reflect the construct it is

ought to measure. It includes checking for indicator reliability and internal consistency

reliability.

[Link].1. Indicator Reliability

This represents how much of the variation in an item is explained by a variable (Hair et al.,

2013). Outer loadings determine the reliability of indicator. A higher outer loading on a variable

indicates that the associated measure has much in common, that is measured by the variable.

Items with loading greater than 0.40 were retained after analyzing the AVE and CR of the

variable (Hair et al., 2013). A total of eight indicators (One from each of the constructs SEA,

ROE, FRB, two items from the constructs LOPr while three items from LOPe) were omitted,

since they had loadings of less than 0.4 (refer appendix B).

Deleting these items resulted in an increase in CR and AVE above suggested threshold values of

0.70 and 0.50, respectively (Hair et al., 2013).

 57

Figure 4.3: Measurement Model

Notes: EI, Emotional Intelligence; SEA, Self-Emotion Appraisal; OEA, Others Emotion
Appraisal; UOE, Use of Emotions; ROE, Regulation of Emotions; IMP, Impulsiveness; NU,
Negative Urgency; LOPe, Lack of Perseverance; LOPr, Lack of Premeditation; PU, Positive
Urgency, FRT, Financial Risk Tolerance; FRB, Financial Risk-taking Behaviour
58

[Link].2. Internal Consistency Reliability

Internal Consistency Reliability is evaluated using Composite Reliability as it considers the

different outer loadings of the indicator variables (Hair et al., 2014). The values between 0.70

and 0.90 have been regarded as satisfactory (Nunnally & Bernstein, 1994) to confirm composite

reliability. Table 4.4 below depict the composite reliability values of each latent variable used in

this study and their values were found to be greater than 0.70 (Chin, 1998b, 2010; Götz et al.,

2010; Hair et al., 2012a; Latan & Ghozali, 2012a), deemed adequate, thus confirming the

internal consistency.

Table 4.4: Composite Reliability

Latent Variables Composite Reliability
EI 0.89
FRB 0.76
FRT 0.87
IMP 0.88
LOPe 0.84
LOPr 0.83
NU 0.86
OEA 0.92
PU 0.90
ROE 0.86
SEA 0.77
UOE 0.92
Notes: EI, Emotional Intelligence; SEA, Self-Emotion Appraisal; OEA, Others Emotion
Appraisal; UOE, Use of Emotions; ROE, Regulation of Emotions; IMP, Impulsiveness; NU,
Negative Urgency; LOPe, Lack of Perseverance; LOPr, Lack of Premeditation; PU, Positive
Urgency, FRT, Financial Risk Tolerance; FRB, Financial Risk-taking Behaviour
Source: Calculation by Author

[Link]. Construct Validity

Construct validity addresses the accuracy of measurement. In particular, it is the extent to which

a test measures what it claims to measure.

[Link].1. Convergent Validity

It refers the extent to which a measure correlates positively with alternative measures of the

same construct, i.e., the indicators or measures of a specific construct should converge or share

a high proportion of variance (Hair et. al, 2014). Average Variance Extracted (AVE) was used
 59

to establish convergent validity. The grand mean value obtained after the loadings on indicators

of construct are squared, determines AVE (Hair [Link]., 2014). In mathematical terms,
𝑆𝑢𝑚 𝑜𝑓 𝑆𝑞𝑢𝑎𝑟𝑒𝑑 𝐿𝑜𝑎𝑑𝑖𝑛𝑔𝑠
AVE =
𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑖𝑛𝑑𝑖𝑐𝑎𝑡𝑜𝑟𝑠

AVE values of 0.5 or higher indicates that, on average, the construct explains more than half of

the variance of its indicators. (Bagozzi & Yi, 1988). Table 4.5 displays the AVE values of all

the latent variables used in this study. The values met the threshold of 0.50 (Hair et al., 2006)

thus establishing convergent validity.

Table 4.5: Convergent Validity

Latent Variable Average Variance Extracted (AVE)
EI 0.61
FRB 0.52
FRT 0.57
IMP 0.64
LOPe 0.72
LOPr 0.61
NU 0.55
OEA 0.70
PU 0.65
ROE 0.61
SEA 0.45
UOE 0.69
Notes: EI, Emotional Intelligence; SEA, Self-Emotion Appraisal; OEA, Others Emotion
Appraisal; UOE, Use of Emotions; ROE, Regulation of Emotions; IMP, Impulsiveness; NU,
Negative Urgency; LOPe, Lack of Perseverance; LOPr, Lack of Premeditation; PU, Positive
Urgency, FRT, Financial Risk Tolerance; FRB, Financial Risk-taking Behaviour
Source: Calculation by Author

[Link].2. Discriminant Validity

It refers the extent to which the construct is empirically different from other constructs. For the

purpose, assessment of Heterotrait-Monotrait ratio (HTMT) of the correlations which is the ratio

of the between-trait correlations to the within-trait correlations is found to be most reliable

(Henseler et al., 2015). HTMT is the mean of all correlations of indicators across constructs

measuring different constructs (i.e., the Heterotrait-Heteromethod correlations) relative to the

(geometric) mean of the average correlations of indicators measuring the same construct

(Henseler et al., 2015). Precisely, the HTMT approach evaluates what the true correlation
60

among constructs would be, if they were perfectly reliable. HTMT value above 0.90 signifies a

lack of discriminant validity and thus threshold value of 0.85 is acceptable (Henseler et al.,

2015). Table 4.6 displays the values of all the constructs are within the threshold limits, thus

ensuring discriminant validity.

Table 4.6: Discriminant Validity (HTMT ratio)

EI FRB FRT IMP LOPe LOPr NU OEA PU ROE SEA UOE
EI
FRB 0.42
FRT 0.34 0.44
IMP 0.59 0.54 0.44
LOPe 0.63 0.29 0.24 0.74
LOPr 0.66 0.39 0.26 0.31 0.69
NU 0.32 0.47 0.44 0.27 0.29 0.34
OEA 0.83 0.26 0.23 0.36 0.18 0.37 0.32
PU 0.37 0.39 0.30 0.31 0.35 0.35 0.56 0.21
ROE 0.79 0.21 0.29 0.52 0.53 0.54 0.19 0.33 0.31
SEA 0.44 0.38 0.28 0.58 0.67 0.63 0.27 0.56 0.42 0.67
UOE 0.43 0.38 0.20 0.36 0.33 0.48 0.16 0.32 0.24 0.28 0.53
Notes: EI, Emotional Intelligence; SEA, Self-Emotion Appraisal; OEA, Others Emotion
Appraisal; UOE, Use of Emotions; ROE, Regulation of Emotions; IMP, Impulsiveness; NU,
Negative Urgency; LOPe, Lack of Perseverance; LOPr, Lack of Premeditation; PU, Positive
Urgency, FRT, Financial Risk Tolerance; FRB, Financial Risk-taking Behaviour
Source: Calculation by Author

As further suggested, while examining HTMT ratios, there lies a strong need to determine if

HTMT values are significantly different from 1. Therefore, bootstrap confidence intervals

results are computed to ensure discriminant validity of the constructs. The columns

characterized 2.5% and 97.5% (refer, table 4.7) illustrates the lower and upper bounds of the

95% (bias-corrected and accelerated) confidence interval.

Table 4.7: Confidence Intervals for HTMT

Original Sample Bias 2.50% 97.50%
Sample (O) Mean (M)

FRB -> EI 0.42 0.45 0.04 0.32 0.49

FRT -> EI 0.34 0.37 0.03 0.27 0.38
FRT -> FRB 0.44 0.47 0.03 0.33 0.51
IMP -> EI 0.59 0.60 0.02 0.50 0.66
IMP -> FRB 0.54 0.56 0.02 0.41 0.64
IMP -> FRT 0.44 0.45 0.02 0.33 0.53
LOPe -> EI 0.63 0.63 0.00 0.51 0.75
 61

Original Sample Bias 2.50% 97.50%

Sample (O) Mean (M)

LOPe -> FRB 0.29 0.31 0.02 0.17 0.43

LOPe -> FRT 0.24 0.26 0.02 0.16 0.29
LOPe -> IMP 0.74 0.76 0.02 0.65 0.83
LOPr -> EI 0.66 0.66 0.00 0.54 0.77
LOPr -> FRB 0.39 0.41 0.02 0.25 0.54
LOPr -> FRT 0.26 0.28 0.03 0.16 0.36
LOPr -> IMP 0.31 0.34 0.03 0.20 0.43
LOPr -> LOPe 0.69 0.70 0.00 0.55 0.83
NU -> EI 0.32 0.34 0.02 0.23 0.39
NU -> FRB 0.47 0.49 0.02 0.32 0.59
NU -> FRT 0.44 0.44 0.01 0.29 0.58
NU -> IMP 0.27 0.31 0.04 0.18 0.31
NU -> LOPe 0.29 0.32 0.04 0.21 0.35
NU -> LOPr 0.34 0.34 0.01 0.20 0.48
OEA -> EI 0.83 0.83 0.00 0.79 0.88
OEA -> FRB 0.26 0.28 0.02 0.16 0.35
OEA -> FRT 0.23 0.25 0.01 0.16 0.30
OEA -> IMP 0.36 0.38 0.01 0.26 0.46
OEA -> LOPe 0.18 0.19 0.01 0.08 0.33
OEA -> LOPr 0.37 0.38 0.00 0.23 0.50
OEA -> NU 0.32 0.33 0.00 0.21 0.45
PU -> EI 0.37 0.40 0.02 0.27 0.47
PU -> FRB 0.39 0.41 0.02 0.27 0.51
PU -> FRT 0.30 0.31 0.01 0.19 0.41
PU -> IMP 0.31 0.33 0.01 0.19 0.44
PU -> LOPe 0.35 0.36 0.01 0.22 0.49
PU -> LOPr 0.35 0.36 0.01 0.19 0.53
PU -> NU 0.56 0.56 0.00 0.42 0.67
PU -> OEA 0.21 0.23 0.02 0.13 0.30
ROE -> EI 0.79 0.79 0.01 0.72 0.85
ROE -> FRB 0.21 0.25 0.04 0.12 0.31
ROE -> FRT 0.29 0.32 0.02 0.23 0.34
ROE -> IMP 0.52 0.54 0.01 0.45 0.58
ROE -> LOPe 0.53 0.53 0.00 0.41 0.64
ROE -> LOPr 0.54 0.54 0.00 0.42 0.67
ROE -> NU 0.19 0.21 0.02 0.14 0.22
ROE -> OEA 0.33 0.33 0.00 0.22 0.44
ROE -> PU 0.31 0.32 0.02 0.22 0.38
SEA -> EI 0.44 0.44 0.01 0.29 0.58
SEA -> FRB 0.38 0.44 0.06 0.25 0.47
SEA -> FRT 0.28 0.30 0.03 0.19 0.34
SEA -> IMP 0.58 0.61 0.02 0.47 0.68
62

Original Sample Bias 2.50% 97.50%

Sample (O) Mean (M)

SEA -> LOPe 0.67 0.67 0.00 0.54 0.80

SEA -> LOPr 0.63 0.63 0.01 0.47 0.76
SEA -> NU 0.27 0.29 0.03 0.17 0.35
SEA -> OEA 0.56 0.57 0.00 0.45 0.68
SEA -> PU 0.42 0.44 0.03 0.26 0.57
SEA -> ROE 0.67 0.69 0.02 0.55 0.77
UOE -> EI 0.43 0.44 0.01 0.31 0.55
UOE -> FRB 0.38 0.39 0.01 0.28 0.48
UOE -> FRT 0.20 0.23 0.02 0.14 0.28
UOE -> IMP 0.36 0.37 0.01 0.28 0.44
UOE -> LOPe 0.33 0.33 0.00 0.18 0.47
UOE -> LOPr 0.48 0.49 0.00 0.36 0.60
UOE -> NU 0.16 0.18 0.02 0.10 0.23
UOE -> OEA 0.32 0.33 0.01 0.23 0.43
UOE -> PU 0.24 0.26 0.01 0.17 0.32
UOE -> ROE 0.28 0.29 0.01 0.19 0.37
UOE -> SEA 0.53 0.54 0.01 0.41 0.64
Notes: EI, Emotional Intelligence; SEA, Self-Emotion Appraisal; OEA, Others Emotion
Appraisal; UOE, Use of Emotions; ROE, Regulation of Emotions; IMP, Impulsiveness; NU,
Negative Urgency; LOPe, Lack of Perseverance; LOPr, Lack of Premeditation; PU, Positive
Urgency, FRT, Financial Risk Tolerance; FRB, Financial Risk-taking Behaviour
Source: Calculation by Author

As seen in the table above, neither of the confidence intervals includes the value 1, discriminant

validity is further supported.

Table 4.8 presents the summarized result of the reflective measurement model estimation. The

values represent that evaluation criteria have been met thereby providing support for the

measures’ reliability and validity

Table: 4.8 Results Summary of Measurement Model Estimation

INTERNAL
CONVERGENT
Latent Variables

CONSISTENCY
VALIDITY
Indicators

RELIABILITY
Loadings

Cronbach's

Composite
Reliability

Extracted
Variance
Average

(AVE)
Alpha

SEA1 0.75
Emotional Self-Emotion SEA2 0.62
0.60 0.86 0.77 0.89 0.45 0.61
Intelligence Appraisal SEA3 0.67
SEA4 0.64
 63

INTERNAL
CONVERGENT

Latent Variables
CONSISTENCY
VALIDITY

Indicators
RELIABILITY

Loadings

Cronbach's

Composite
Reliability

Extracted
Variance
Average

(AVE)
Alpha
OEA1 0.78
OEA2 0.83
Others’-Emotion
OEA3 0.88 0.89 0.92 0.70
Appraisal
OEA4 0.92
OEA5 0.76
UOE1 0.76
UOE2 0.89
Use of Emotions UOE3 0.71 0.88 0.92 0.69
UOE4 0.90
UOE5 0.89
ROE1 0.53
Regulation of ROE2 0.78
0.77 0.86 0.61
Emotions ROE3 0.87
ROE4 0.91
NU1 0.73
NU2 0.80
Negative Urgency NU3 0.77 0.80 0.86 0.55
NU4 0.69
NU5 0.72
Lack of LOPe2 0.83
0.62 0.84 0.72
Perseverance LOPe4 0.88
Impulsiveness LOPr3 0.80 0.85 0.88 0.64
Lack of
LOPr4 0.79 0.68 0.83 0.61
Premeditation
LOPr5 0.76
PU1 0.74
PU2 0.81
Positive Urgency PU3 0.82 0.87 0.90 0.65
PU4 0.89
PU5 0.77
FRB1 0.72
Financial Risk-taking Behaviour FRB2 0.6 0.55 0.76 0.52
FRB3 0.83
FRT1 0.81
FRT2 0.80
Financial Risk Tolerance FRT3 0.81 0.81 0.87 0.57
FRT4 0.73
FRT5 0.63
Notes: EI, Emotional Intelligence; SEA, Self-Emotion Appraisal; OEA, Others Emotion
Appraisal; UOE, Use of Emotions; ROE, Regulation of Emotions; IMP, Impulsiveness; NU,
Negative Urgency; LOPe, Lack of Perseverance; LOPr, Lack of Premeditation; PU, Positive
Urgency, FRT, Financial Risk Tolerance; FRB, Financial Risk-taking Behaviour
Source: Compiled by Author
64

4.3.3. Inner Model/Structural Model Evaluation

After confirming the reliability and validity of the construct, the next step is to assess the results

of structural model. This involves exploring the model’s predictive capabilities and the

relationships between the constructs. A systematic approach is followed while assessing the

structural model;

Assess Collinearity
Issues

Examine significance
of Path Coefficients

Assess Coefficient of
Determination

Assess effect size

Assess cross-validated
redundancy (Predictive
Relevance)

Figure 4.4: Procedure for Evaluation of Structural Model

Source: Adapted from Hair et. al., (2013)

[Link]. Collinearity Assessment

The structural model must be examined for collinearity issues, since the estimation of path

coefficients in the structural models is based on OLS regressions of each endogenous latent

variable on its corresponding predecessor constructs and their values and significances can be

subject to biases if constructs are highly correlated. The tolerance values below 0.20 (VIF > 5)

in the predictor constructs are considered as critical levels of collinearity (Hair, Black, & Babin,

2010). As can be seen in Table 4.9, the VIF values of all the constructs are below the threshold

value of 5 thus confirming that collinearity is not an issue.

Table 4.9: Collinearity Statistics

FRB FRT LOPe LOPr NU OEA PU ROE SEA UOE
EI 1.22 1.19 1.00 1.00 1.00 1.00
 65

FRB FRT LOPe LOPr NU OEA PU ROE SEA UOE

FRT 1.15
IMP 1.37 1.19 1.00 1.00 1.00 1.00
Notes: EI, Emotional Intelligence; SEA, Self-Emotion Appraisal; OEA, Others Emotion
Appraisal; UOE, Use of Emotions; ROE, Regulation of Emotions; IMP, Impulsiveness; NU,
Negative Urgency; LOPe, Lack of Perseverance; LOPr, Lack of Premeditation; PU, Positive
Urgency, FRT, Financial Risk Tolerance; FRB, Financial Risk-taking Behaviour
Source: Calculation by Author

[Link]. Path Coefficients

Figure 4.4 displays the path coefficient for the direct relationship between Emotional

Intelligence and Impulsiveness with Financial Risk Tolerance and Risk-taking Behaviour. The

bootstrapping procedure was employed to assess the significance of relationships.

“Bootstrapping is a resampling technique that draws a large number of subsamples from the

original data and estimates models for each subsample. It is used to determine standard errors of

coefficients in order to assess their statistical significance without relying on distributional

assumptions.” (Hair et al., 2013).

Table 4.10 below displays the significance testing results of Structural Model Path Coefficients.

Table 4.10 Results of Structural Model Path Coefficients

Path Path t Values p 𝐕𝐚𝐥𝐮𝐞𝐬 𝐛 95% Significance
Coefficients Confidence (p<0.05)
Intervals
EI -> FRB 0.14 3.06 0.00 (0.057,0.242) Yes
EI -> FRT 0.16 2.19 0.03 (0.010,0.295) Yes
EI -> OEA 0.75 19.61 0.00 (0.658,0.813) Yes
EI -> ROE 0.62 11.86 0.00 (0.506,0.714) Yes
EI -> SEA 0.73 22.48 0.00 (0.657,0.788) Yes
EI -> UOE 0.71 15.98 0.00 (0.586,0.773) Yes
FRT -> FRB 0.66 14.56 0.00 (0.560,0.743) Yes
IMP -> FRB 0.13 2.30 0.02 (0.016,0.238) Yes
IMP -> FRT 0.40 6.39 0.00 (0.268,0.509) Yes
IMP -> LOPe 0.49 6.94 0.00 (0.328,0.607) Yes
IMP -> LOPr 0.55 6.44 0.00 (0.332,0.678) Yes
IMP -> NU 0.78 22.59 0.00 (0.692,0.831) Yes
IMP -> PU 0.85 33.91 0.00 (0.784,0.888) Yes
Notes: EI, Emotional Intelligence; SEA, Self-Emotion Appraisal; OEA, Others Emotion
Appraisal; UOE, Use of Emotions; ROE, Regulation of Emotions; IMP, Impulsiveness; NU,
Negative Urgency; LOPe, Lack of Perseverance; LOPr, Lack of Premeditation; PU, Positive
Urgency, FRT, Financial Risk Tolerance; FRB, Financial Risk-taking Behaviour
b = p values were calculated using bootstrapping method on a subsample of 5000 (* significant
at p<0.05)
Source: Calculation by Author
66

Figure 4.5: Structural Model

Notes: EI, Emotional Intelligence; SEA, Self-Emotion Appraisal; OEA, Others Emotion
Appraisal; UOE, Use of Emotions; ROE, Regulation of Emotions; IMP, Impulsiveness; NU,
Negative Urgency; LOPe, Lack of Perseverance; LOPr, Lack of Premeditation; PU, Positive
Urgency, FRT, Financial Risk Tolerance; FRB, Financial Risk-taking Behaviour
 67

[Link]. Assessing Coefficient of Determination (𝑹𝟐 value)

The model’s predictive accuracy was obtained using 𝑅 2 value which is squared correlation of

actual and predicted values of specified endogenous construct. The coefficient also explains the

amount of variance on the endogenous constructs by combined effects of exogenous latent

variables. The 𝑅 2 values ranging from 0 to 1 with higher level indicates higher levels of

predictive accuracy, however, as a rule of thumb, the values of 0.75, 0.50, or 0.25 can be

considered as substantial, moderate, or weak respectively (Hair, Ringle, & Sarstedt, 2011;

Henseler et al., 2009).

The 𝑅 2 value of FRT is 0.13, which means that combined effect of the exogenous variables

(psychological constructs) cause 13% variation in financial risk tolerance. It was observed that

prior studies on FRT have reported the in the range of 7% to 36% (Grable, 2000; Grable & Joo,

2004; Grable & Lytton, 1998; Kannadhasan, 2015; Ryack, 2011). The value of the present study

thus obtained is found within the range. For FRB 𝑅 2 value is 0.52, meaning the combined effect

of all the exogenous variables (psychological constructs and FRT) can cause 52% variation in

financial risk taking behaviour.

Merely relying on 𝑅 2 value to predictive model’s accuracy leads to biased results, hence

analogous to multiple regression, the adjusted 𝑅 2 value was used as the criterion to eliminate

bias toward complex model.

Table 4.11, shows the variance exhibited by exogenous constructs on endogenous constructs.

Table 4.11: Coefficient of Determination

𝑹𝟐 Adjusted 𝑹𝟐 Predictive Accuracy

FRB 0.52 0.51 Substantial

FRT 0.13 0.13 Weak
Notes: FRT, Financial Risk Tolerance; FRB, Financial Risk-taking Behaviour
Source: Calculation by Author

[Link]. Assessing effect size (𝒇𝟐 values)

𝑓 2 is a measure to determine the effect size for each path model. The 𝑓 2 is calculated by taking

into account the change in 𝑅 2 when a specific construct is removed from the model, that is, two

path models are estimated. The first path model should include all the constructs as specified in
68

2
the hypotheses, yielding the 𝑅 2 of the full model (𝑅𝑖𝑛𝑐𝑙𝑢𝑑𝑒𝑑 ), while the second model should

exclude a specific exogenous construct, thereby yielding the 𝑅 2 of the reduced model
2
(𝑅𝑒𝑥𝑐𝑙𝑢𝑑𝑒𝑑 ). The 𝑓 2 values of 0.02, 0.15, and 0.35 represent small, medium, and large effects,

respectively, (Cohen, 1988). The 𝑓 2 values for each of the constructs is depicted below (refer,

table 4.12).

Table 4.12: Effect Size (𝒇𝟐 )

EI FRB FRT IMP LOPe LOPr NU OEA PU ROE SEA UOE
EI 0.04 0.02 1.30 0.63 1.14 0.99
FRB
FRT 0.79
IMP 0.03 0.15 0.32 0.43 1.51 2.57
Notes: EI, Emotional Intelligence; SEA, Self-Emotion Appraisal; OEA, Others Emotion
Appraisal; UOE, Use of Emotions; ROE, Regulation of Emotions; IMP, Impulsiveness; NU,
Negative Urgency; LOPe, Lack of Perseverance; LOPr, Lack of Premeditation; PU, Positive
Urgency, FRT, Financial Risk Tolerance; FRB, Financial Risk-taking Behaviour
Source: Calculation by Author

[Link]. Assessing Cross-validated redundancy (𝑸𝟐 value)

While 𝑅 2 was measured to determine model’s predictive accuracy, Stone-Geisser's, 𝑸𝟐

(Geisser, 1974; Stone, 1974) is a measure to assess the model’s predictive relevance. The

measure is evaluated using blindfolding procedure, which is a resampling technique, that omits

a part of the data point and estimates the parameter with remaining data point (Chin, 1998;

Henseler et al., 2009; Tenenhaus et al., 2005). The difference between the original and predicted

values are indicators of 𝑄 2 value. In structural model, 𝑄 2 > 0 for a certain reflective

endogenous construct indicates the construct’s predictive relevance of path model.

As can be seen, Table 4.13, all 𝑄 2 values are considerably above zero, thus providing support

for the model's predictive relevance regarding the endogenous latent variables.

Table 4.13: Cross-validated redundancy (𝑸𝟐 value)

SSO SSE Q² (=1-SSE/SSO)
FRB 909.00 684.45 0.25
FRT 1515.00 1414.43 0.07
Notes: FRT, Financial Risk Tolerance; FRB, Financial Risk-taking Behaviour
Source: Calculation by Author

Table 4.14: Summarized Results of Structural Model Estimation

Path VIF β 𝑹𝟐 Adj. t-value p 𝐯𝐚𝐥𝐮𝐞𝐛 𝒇𝟐 𝑸𝟐 Result
𝑹𝟐
 69

Path VIF β 𝑹𝟐Adj. t-value p 𝐯𝐚𝐥𝐮𝐞𝐛 𝒇𝟐 𝑸𝟐 Result

𝟐
𝑹
EI -> SEA 1.00 0.73 0.53 0.53 22.48 0.00 1.14 0.22 Significant
EI -> OEA 1.00 0.75 0.57 0.56 19.62 0.00 1.30 0.37 Significant
EI -> UOE 1.00 0.71 0.50 0.50 15.98 0.00 0.99 0.32 Significant
EI -> ROE 1.00 0.62 0.39 0.39 11.86 0.00 0.63 0.21 Significant
EI -> FRT 1.19 0.16 0.13 0.13 2.19 0.03 0.02 0.07 Significant
IMP -> FRT 1.19 0.40 6.39 0.00 0.15 Significant
IMP -> NU 1.00 0.78 0.60 0.60 22.60 0.00 1.51 0.31 Significant
IMP -> LOPe 1.00 0.50 0.25 0.24 6.94 0.00 0.32 0.17 Significant
IMP -> LOPr 1.00 0.55 0.30 0.30 6.44 0.00 0.43 0.17 Significant
IMP -> PU 1.00 0.85 0.72 0.72 33.91 0.00 2.57 0.44 Significant
FRT -> FRB 1.15 0.66 0.52 0.51 14.56 0.00 0.79 0.25 Significant
EI -> FRB 1.22 0.15 3.80 0.00 0.04 Significant
IMP -> FRB 1.37 0.13 2.30 0.02 0.03 Significant
Notes: EI, Emotional Intelligence; SEA, Self-Emotion Appraisal; OEA, Others Emotion
Appraisal; UOE, Use of Emotions; ROE, Regulation of Emotions; IMP, Impulsiveness; NU,
Negative Urgency; LOPe, Lack of Perseverance; LOPr, Lack of Premeditation; PU, Positive
Urgency, FRT, Financial Risk Tolerance; FRB, Financial Risk-taking Behaviour; b = p values
were calculated using bootstrapping method on a subsample of 5000 (* significant at p<0.05)
Source: Calculation by Author

𝐇𝟏 : There is a significant relationship between Emotional Intelligence and individual

investor’s financial risk tolerance.

Result and Interpretation: Table 4.14 depicts that Emotional Intelligence (EI) explained 13%

variance in Financial Risk Tolerance (FRT). The relationship was found significant (β = 0.16, t=

2.19, p = 0.03) providing the support for Hypothesis 1.

𝐇𝟐 : There is a significant relationship between impulsiveness and individual investor’s

financial risk tolerance.

Result and Interpretation: Table 4.14 depicts that Impulsiveness (IMP) explained 13%

variance in Financial Risk Tolerance (FRT). The relationship was found significant (β = 0.40, t=

6.39, p = 0.00) providing the support for Hypothesis 2.

𝐇𝟑 : Financial Risk Tolerance is significantly related to Financial Risk-taking Behaviour

Result and Interpretation: Table 4.14 depicts that Financial Risk Tolerance (FRT) explained

52% variance in Financial Risk-taking Behaviour (FRB). The relationship was found significant

(β = 0.66, t= 14.56, p = 0.00) providing the support for Hypothesis 3.

70

𝐇𝟒 : There is a significant relationship between Emotional Intelligence and Financial Risk-

taking Behaviour of individual investor.

Result and Interpretation: Table 4.14 depicts that Emotional Intelligence (EI) explained 52%

variance in Financial Risk-taking Behaviour (FRB). The relationship was found significant (β =

0.15, t= 3.80, p = 0.00) providing the support for Hypothesis 4.

𝐇𝟓 : There is a significant relationship between impulsiveness and Financial Risk-taking

Behaviour of individual investor.

Result and Interpretation: Table 4.14 depicts that Impulsiveness (IMP) explained 52%

variance in Financial Risk-taking Behaviour (FRB). The relationship was found significant (β =

0.13, t= 2.30, p = 0.02) providing the support for Hypothesis 5.

4.4. Moderator Analysis

To examine the moderating effects of demographics (gender, age, education, marital status,

occupation, annual household disposable income, experience in investment), a multigroup

analysis (PLS-MGA) was performed to tests the differences in path coefficients between the

two selected groups (Henseler et al., 2009) using bootstrapping method on a subsample of 5000.

PLS-MGA is performed across all possible paths to test the hypothesized relationships refer

Figure 4.6;

Gender, Age, Annual Household Disposable

Income, Marital Status, Education, Occupation,
Experience in Investment

Emotional
Intelligence

Financial
Financial Risk-taking
Risk tolerance behaviour

Impulsiveness

Figure 4.6: Moderating Effect of Demographics

 71

4.4.1. Gender

In financial domain, males tend to exhibit riskier behaviour as compared to females (Riley &

Chow, 1992; Grable, 2000; Barber & Odean, 2001; Hallahan, Faff, & McKenzie, 2003; Frijns,

Koellen, & Lehnert, 2008). The status being controversial, the present study proposed following

hypotheses;

𝑯𝟔𝒂: There is a significant moderating effect of gender on the relationship between Emotional

Intelligence and Financial Risk Tolerance.

𝑯𝟔𝒃: There is a significant moderating effect of gender on the relationship between

Impulsiveness and Financial Risk Tolerance.

𝑯𝟔𝒄 : There is a significant moderating effect of gender on the relationship between Emotional

Intelligence and Financial Risk-taking Behaviour.

𝑯𝟔𝒅 : There is a significant moderating effect of gender on the relationship between

Impulsiveness and Financial Risk-taking Behaviour.

Table 4.15: PLS-MGA Results for Gender as a Moderator

Path Gender (Male vs Female)
β-diff p-Value Result

EI -> FRT 0.06 0.63 Not Significant

IMP -> FRT 0.26 0.97 Not Significant
EI -> FRB 0.29 0.03 Significant
IMP -> FRB 0.09 0.69 Not Significant
𝐅𝐑𝐓 −> 𝐅𝐑𝐁 ∗ 0.02 0.62 Not Significant
Notes: EI, Emotional Intelligence; IMP, Impulsiveness; FRT, Financial Risk Tolerance; FRB,
Financial Risk-taking Behaviour; * Hypothesis was not formulated in conceptual model but as
an exploratory step was evaluated during model testing
Source: Calculation by Author

The results in Table 4.15 reveal gender to have moderating effect only on single relationship,

i.e., Emotional Intelligence and Financial Risk-taking Behaviour, thereby providing support for

𝐇𝟔𝒄 .

4.4.2. Age

The influence of age was investigated by dividing the sample into Young Adult and Mature

Adult age-groups. Some researchers reported no relationship with age (Grable, 1997; Grable &
72

Joo, 1997; Wang & Hanna, 1998) while others reported negative relationships (Wallach &

Kogan, 1961; McInish, 1982; Morin & Suarez, 1983; Riley & Chow,1992; Hallahan, Faff, &

McKenzie, 2003, 2004) and positive relationships too (Palsson, 1996, Wang & Hanna, 1997;

Grable 2000; Frijns, Koellen, & Lehnert, 2008). Since the status lacks consensus, the present

study hypothesized;

𝑯𝟕𝒂: There is a significant moderating effect of age on the relationship between Emotional

Intelligence and Financial Risk Tolerance.

𝑯𝟕𝒃: There is a significant moderating effect of age on the relationship between Impulsiveness

and Financial Risk Tolerance.

𝑯𝟕𝒄 : There is a significant moderating effect of age on the relationship between Emotional

Intelligence and Financial Risk-taking Behaviour.

𝑯𝟕𝒅 : There is a significant moderating effect of age on the relationship between Impulsiveness

and Financial Risk-taking Behaviour.

Table 4.16: PLS-MGA Results for Age as a Moderator

Path Age (Young Adult vs Mature Adult)
β-diff p-Value Result
EI -> FRT 0.31 0.05** Significant
IMP -> FRT 0.13 0.21 Not Significant
EI -> FRB 0.34 0.00 Significant
IMP -> FRB 0.09 0.77 Not Significant
𝐅𝐑𝐓 −> 𝐅𝐑𝐁 ∗ 0.19 0.02 Significant
Notes: EI, Emotional Intelligence; IMP, Impulsiveness; FRT, Financial Risk Tolerance; FRB,
Financial Risk-taking Behaviour; * Hypothesis was not formulated in conceptual model but as
an exploratory step was evaluated during model testing; ** Significant at p <0.10
Source: Calculation by Author

The results as exhibited in Table 4.16 indicated that age moderated three relationships:

(a) Emotional Intelligence and Financial Risk Tolerance 𝐇𝟕𝒂 ; (b) Emotional Intelligence and

Financial Risk-taking Behaviour 𝐇𝟕𝒄 . (c) Financial Risk Tolerance and Financial Risk-taking

Behaviour 𝐇𝟕𝒆 ;

4.4.3. Occupation

Occupation has rarely been focused on studies relating to Financial Risk Tolerance and Risk-

taking Behaviour (Quattlebaum, 1988), therefore the present study hypothesized;

 73

𝑯𝟖𝒂: There is a significant moderating effect of occupation on the relationship between

Emotional Intelligence and Financial Risk Tolerance.

𝑯𝟖𝒃: There is a significant moderating effect of occupation on the relationship between

Impulsiveness and Financial Risk Tolerance.

𝑯𝟖𝒄 : There is a significant moderating effect of occupation on the relationship between

Emotional Intelligence and Financial Risk-taking Behaviour.

𝑯𝟖𝒅 : There is a significant moderating effect of occupation on the relationship between

Impulsiveness and Financial Risk-taking Behaviour.

Table 4.17: PLS-MGA Results for Occupation as a Moderator

Path Occupation (Private Sector vs Public Sector)
β-diff p-Value Result
EI -> FRT 0.21 0.06** Significant
IMP -> FRT 0.01 0.48 Not Significant
EI -> FRB 0.25 0.04 Significant
IMP -> FRB 0.16 0.87 Not Significant
∗
𝐅𝐑𝐓 −> 𝐅𝐑𝐁 0.26 0.07** Significant
Notes: EI, Emotional Intelligence; IMP, Impulsiveness; FRT, Financial Risk Tolerance; FRB,
Financial Risk-taking Behaviour; * Hypothesis was not formulated in conceptual model but as
an exploratory step was evaluated during model testing; ** Significant at p <0.10
Source: Calculation by Author

As can be seen in Table 4.17, the moderating effect of occupation is found on three

relationships: (a) Emotional Intelligence and Financial Risk Tolerance 𝑯𝟖𝒂 ; (b) Emotional

Intelligence and Financial Risk-taking Behaviour 𝑯𝟖𝒄 ; and (c) Financial Risk Tolerance and

Financial Risk-taking Behaviour 𝑯𝟖𝒆 .

4.4.4. Annual Household Disposable Income

Higher level of income induces individuals to be more risk tolerant (Morin & Suarez, 1983;

Riley & Chow, 1992; Grable, 2000; Hallahan, Faff, & McKenzie, 2003, 2004). The present

study hypothesized;

𝑯𝟗𝒂: There is a significant moderating effect of income on the relationship between Emotional

Intelligence and Financial Risk Tolerance.

𝑯𝟗𝒃: There is a significant moderating effect of income on the relationship between

Impulsiveness and Financial Risk Tolerance.

74

𝑯𝟗𝒄 : There is a significant moderating effect of income on the relationship between Emotional

Intelligence and Financial Risk-taking Behaviour.

𝑯𝟗𝒅 : There is a significant moderating effect of income on the relationship between

Impulsiveness and Financial Risk-taking Behaviour.

Table 4.18: PLS-MGA Results for Income as a Moderator

Path Income (High Income vs Low Income)
β-diff p-Value Result
EI -> FRT 0.41 0.03 Significant
IMP -> FRT 0.15 0.83 Not Significant
EI -> FRB 0.69 0.00 Significant
IMP -> FRB 0.03 0.61 Not Significant
𝐅𝐑𝐓 −> 𝐅𝐑𝐁 ∗ 0.21 0.07** Significant
Notes: EI, Emotional Intelligence; IMP, Impulsiveness; FRT, Financial Risk Tolerance; FRB,
Financial Risk-taking Behaviour; * Hypothesis was not formulated in conceptual model but as
an exploratory step was evaluated during model testing; ** Significant at p <0.10
Source: Calculation by Author

The results as displayed in Table 4.18 provide evidence that income moderated three

relationships: (a) Emotional Intelligence and Financial Risk Tolerance 𝑯𝟗𝒂; (b) Emotional

Intelligence and Financial Risk-taking Behaviour 𝑯𝟗𝒄 ; (c) Financial Risk Tolerance and

Financial Risk-taking Behaviour 𝑯𝟗𝒆 .

4.4.5. Education

The higher the education, the higher is Financial Risk Tolerance (Maccrimmon & Wehrung,

1986; Lee & Hanna, 1991; Sung & Hanna, 1996; Grable & Lytton, 1999a, 1999b; Grable, 2000;

Hallahan et al., 2004; Al-Ajmi, 2008). However, the actual status remains controversial since

𝑯𝟏𝟎𝒂: There is a significant moderating effect of education on the relationship between

Emotional Intelligence and Financial Risk Tolerance.

𝑯𝟏𝟎𝒃: There is a significant moderating effect of education on the relationship between

Impulsiveness and Financial Risk Tolerance.

𝑯𝟏𝟎𝒄: There is a significant moderating effect of education on the relationship between

Emotional Intelligence and Financial Risk-taking Behaviour.

𝑯𝟏𝟎𝒅 : There is a significant moderating effect of education on the relationship between

Impulsiveness and Financial Risk-taking Behaviour.

 75

Table 4.19: PLS-MGA Results for Education as a Moderator

Path Education (Professional vs Non-professional)
β-diff p-Value Result
EI -> FRT 0.07 0.32 Not Significant
IMP -> FRT 0.02 0.25 Not Significant
EI -> FRB 0.18 0.05** Significant
IMP -> FRB 0.08 0.25 Not Significant
∗
𝐅𝐑𝐓 −> 𝐅𝐑𝐁 0.08 0.80 Not Significant
Notes: EI, Emotional Intelligence; IMP, Impulsiveness; FRT, Financial Risk Tolerance; FRB,
Financial Risk-taking Behaviour; * Hypothesis was not formulated in conceptual model but as
an exploratory step was evaluated during model testing; ** Significant at p <0.10
Source: Calculation by Author

The results in Table 4.19 depicts the significant moderating effect of education on the

relationship between Emotional Intelligence and Financial Risk-taking Behaviour 𝑯𝟏𝟎𝒄.

4.4.6. Marital Status

While few researches did not find any significant relationship (Riley & Chow, 1992; Grable &

Joo, 1997), the others debated over single versus married individuals (Hallahan, Faff, &

McKenzie, 2003, 2004; Sung & Hanna 1996; Grable, 2000; Yao & Hanna, 2005). The present

study hypothesized;

𝑯𝟏𝟏𝒂: There is a significant moderating effect of marital status on the relationship between

Emotional Intelligence and Financial Risk Tolerance.

𝑯𝟏𝟏𝒃: There is a significant moderating effect of marital status on the relationship between

Impulsiveness and Financial Risk Tolerance.

𝑯𝟏𝟏𝒄: There is a significant moderating effect of marital status on the relationship between

Emotional Intelligence and Financial Risk-taking Behaviour.

𝑯𝟏𝟏𝒅 : There is a significant moderating effect of marital status on the relationship between

Impulsiveness and Financial Risk-taking Behaviour.

The results in Table 4.20 reported significance of two relationships (a) Emotional Intelligence

and Financial Risk-taking Behaviour 𝑯𝟏𝟏𝒄 and (b) Financial Risk Tolerance and Financial Risk-

taking Behaviour 𝑯𝟏𝟏𝒆.

Table 4.20: PLS-MGA Results for Marital Status as a Moderator

Path Marital Status (Married vs Unmarried)
β-diff p-Value Result
76

EI -> FRT 0.16 0.86 Not Significant

IMP -> FRT 0.01 0.53 Not Significant
EI -> FRB 0.21 0.08** Significant
IMP -> FRB 0.11 0.19 Not Significant
∗
𝐅𝐑𝐓 −> 𝐅𝐑𝐁 0.19 0.04 Significant
Notes: EI, Emotional Intelligence; IMP, Impulsiveness; FRT, Financial Risk Tolerance; FRB,
Financial Risk-taking Behaviour; * Hypothesis was not formulated in conceptual model but as
an exploratory step was evaluated during model testing; ** Significant at p <0.10Source:
Calculation by Author

4.4.7. Experience in Investment

The increased level of experience in investment has been associated with increased level of

financial risk tolerance (Grable, 2000; Frijns, Koellen, & Lehnert, 2008). Therefore, the present

study hypothesized;

𝑯𝟏𝟐𝒂: There is a significant moderating effect of experience in investment on the relationship

between Emotional Intelligence and Financial Risk Tolerance.

𝑯𝟏𝟐𝒃: There is a significant moderating effect of experience in investment on the relationship

between Impulsiveness and Financial Risk Tolerance.

𝑯𝟏𝟐𝒄: There is a significant moderating effect of experience in investment on the relationship

between Emotional Intelligence and Financial Risk-taking Behaviour.

𝑯𝟏𝟐𝒅 : There is a significant moderating effect of experience in investment on the relationship

between Impulsiveness and Financial Risk-taking Behaviour.

Table 4.21: PLS-MGA Results for Experience in Investment as a Moderator

Path Experience in Investment
(Greater than 5 years vs Less than or equal to 5 years)
β-diff p-Value Result
EI -> FRT 0.24 0.08** Significant
IMP -> FRT 0.07 0.30 Not Significant
EI -> FRB 0.19 0.87 Not Significant
IMP -> FRB 0.18 0.08** Significant
∗
𝐅𝐑𝐓 −> 𝐅𝐑𝐁 0.25 0.07** Significant
Notes: EI, Emotional Intelligence; IMP, Impulsiveness; FRT, Financial Risk Tolerance; FRB,
Financial Risk-taking Behaviour; ** Significant at p <0.10
* Hypothesis was not formulated in conceptual model but as an exploratory step was evaluated
during model testing.
Source: Calculation by Author

As displayed in Table 4.21, the results concluded that experience in investment moderated three

relationships: (a) Emotional Intelligence and Financial Risk Tolerance 𝑯𝟏𝟐𝒂; (b) Impulsiveness
 77

and Financial Risk-taking Behaviour 𝑯𝟏𝟐𝒅 ; and (c) Financial Risk Tolerance and Financial

Risk-taking Behaviour 𝑯𝟏𝟐𝒆.

4.5. Mediation Analysis

As an exploratory step and to gain a better understanding of the role of Financial Risk Tolerance

(FRT) in the model, its potential mediating effect on the relationship between Emotional

Intelligence and Financial Risk-taking Behaviour, and that between Impulsiveness and Financial

Risk-taking Behaviour were examined by adopting Preacher and Hayes (2008) approach that

involves use of bootstrapping in two steps:

i. The mediating variable (Financial Risk Tolerance) is excluded from the model at first,

and the direct effect between independent (Emotional Intelligence and Impulsiveness)

variables and dependent variable (Financial Risk-taking Behaviour) is checked for

significance (refer Figure 4.7).

Emotional
Intelligence
Financial
Risk-taking
behaviour
Impulsiveness

Figure 4.7: Direct Effect (Without Mediator)

Therefore, in absence of Financial Risk Tolerance, Emotional Intelligence (t=3.08,

p=0.00), Impulsiveness (t=2.39, p=0.02) were found to be significantly related to

Financial Risk-taking Behaviour.

Emotional
Intelligence

Financial Financial
Risk Tolerance Risk-taking
Behaviour
Impulsiveness

ii. Figure 4.8: Indirect Effect (With Mediator)

78

Now, the mediating variable is included in the model, and the relationship between

independent variables and dependent variables in presence of intervening variable is

checked for its significance (refer Figure 4.8).

The indirect paths from Emotional Intelligence to Financial Risk-taking Behaviour via

Financial Risk Tolerance were found significant (β=0.11, p=0.04; refer, Table 4.22).

Similarly, the indirect paths from Impulsiveness to Financial Risk-taking Behaviour via

Financial Risk Tolerance were also significant (β=0.26, p=0.00; refer, Table 4.22).

The significance of indirect effect forms the basis of determining the strength of mediating

variable which is obtained through the use of Variance Accounted Factor (VAF). The value of

VAF below 0.20 represents ‘no mediation’; value between 0.20 – 0.80 represents ‘partial

mediation’; and value above 0.80 represents ‘full mediation’ (Hair et al., 2016).

Table 4.22: Mediation Analysis

Procedure Path β t Values p 𝐕𝐚𝐥𝐮𝐞𝐬 𝐛 Total VAF Result
Effects

Direct Effect EI -> FRB 0.15 3.08 0.00 Not Applicable

(Without IMP -> FRB 0.13 2.39 0.02 Not Applicable
Mediator)
Indirect EI -> FRT -> FRB 0.11 2.08 0.04 0.25 0.42 Partial
Effect Mediation
(With EI -> FRT -> FRB 0.26 5.76 0.00 0.39 0.67 Partial
Mediator) Mediation
Notes: EI, Emotional Intelligence; IMP, Impulsiveness; FRT, Financial Risk Tolerance; FRB,
Financial Risk-taking Behaviour; b = p values were calculated using bootstrapping method on a
subsample of 5000 (significant at p<0.05); β = Standardized Regression Coefficient; Total
Effects = Direct Effect + Indirect Effect; Variance Accounted Factor (VAF) = Indirect
Effect/Total Effect
Source: Calculation by Author

The results in Table 4.22 depict that Financial Risk Tolerance partially mediated the

relationships between Emotional Intelligence and Financial Risk-taking Behaviour (t=2.08, p =

0.04) and between Impulsiveness and Financial Risk-taking Behaviour (t=5.76, p = 0.00).

4.6. Importance-Performance Map Analysis (IPMA)

The importance-performance map analysis also known as importance-performance matrix or

priority map analysis is an extension of the standard structural model results reporting path
 79

Figure 4.9: Importance-Performance Map Analysis

Notes: EI, Emotional Intelligence; SEA, Self-Emotion Appraisal; OEA, Others Emotion
Appraisal; UOE, Use of Emotions; ROE, Regulation of Emotions; IMP, Impulsiveness; NU,
Negative Urgency; LOPe, Lack of Perseverance; LOPr, Lack of Premeditation; PU, Positive
Urgency, FRT, Financial Risk Tolerance; FRB, Financial Risk-taking Behaviour
80

coefficient estimates by adding a dimension that considers the average values of the latent

variable scores (refer, Figure 4.9).

In simple terms, it identifies the importance of the predecessor construct in predicting a specific

target construct with their average latent variable scores indicating their performance (Hair,

2017). The importance-performance map helps to prioritize managerial activities having high

importance for the aspect underlying the designated target requiring performance

improvements.

Table 4.23: Direct effects, Indirect effects, and Total Effects in the IPMA
Direct Indirect Total
Effect Effect Effect
EI 0.24 0.18 0.42
IMP 0.22 0.45 0.67
FRT 0.68 - 0.68
Notes: EI, Emotional Intelligence; IMP, Impulsiveness; FRT, Financial Risk Tolerance; FRB,
Financial Risk-taking Behaviour; *All effects denote unstandardized effects

Graph 4.1 and table 4.24 shows that two of the three direct predecessor constructs i.e., IMP and

FRT has relatively below average performance of 45.96 and 48.16, however the importance of

these construct is particularly high with a total effect of 0.67 and 0.68 respectively. However,

aspects related to EI follow third priority with a relatively low importance of 0.42 but high

performance of 65.77.

Table 4.24: Importance-Performance at Construct Level

Importance Performance
EI 0.42 65.77
FRT 0.68 48.16
IMP 0.67 45.96
Mean Value 0.59 53.30
Notes: EI, Emotional Intelligence; IMP, Impulsiveness; FRT, Financial Risk Tolerance; FRB,
Financial Risk-taking Behaviour
Source: Calculation by Author
Hence investment agents or individuals when working with portfolio should therefore prioritize

improving the IMP which can be achieved by focussing on FRT.

Interestingly, IPMA is not constrained to the construct level, rather IPMA can be conducted at

indicator level to identify the relevant and more specific areas of improvement. Hence, to obtain
 81

more detailed information on how to increase the performance of the constructs, particularly

that of FRT, the following analyses was conducted on the indicator level.

Graph 4.1: Importance-Performance Map at Construct Level

Notes: EI, Emotional Intelligence; IMP, Impulsiveness; FRT, Financial Risk Tolerance; FRB,
Financial Risk-taking Behaviour

Table 4.25 displays the output for FRT’s indicators, which leads to producing the importance

performance map as shown in Graph 4.2.

Table 4.25: Importance-Performance at Indicator Level

Importance Performance
FRT1 0.15 43.32
FRT2 0.14 51.65
FRT3 0.17 50.08
FRT4 0.12 47.52
FRT5 0.10 48.18
Mean Value 0.14 48.15
Notes: FRT, Financial Risk Tolerance
Source: Calculation by Author

The IPMA illustrates that indicator FRT3 (i.e., When I think of the word “risk” the term “loss”

comes to mind immediately), has particularly high importance for improving financial risk-

taking behaviour of individuals. Similarly, FRT1 (i.e., Investing is too difficult to understand)
82

has a relatively high importance but also shows a particularly below average performance.

These results reveal that investment agencies or firms should aim at providing awareness or

training programmes about investment management while aiming to improve financial risk-

taking of individuals.

Importance-Performance Map

53.00

52.00
0.14, 51.65
FRT2
51.00

50.00 0.17, 50.08

FRT3

49.00
PERFORMANCE

48.00 0.10, 48.18

FRT5
0.12, 47.52
47.00 FRT4

46.00

45.00

44.00

0.15, 43.32
43.00 FRT1

42.00
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18
IMPORTANCE

Graph 4.2: Importance-Performance Map at Indicator Level

Notes: FRT, Financial Risk Tolerance