CHAPTER 5 DATA ANALYSIS AND HYPOTHESIS TESTING Para No.
Particulars
Page No.
5.1
Data Tabulation
125
5.2
Data Collection Methodology
125
5.3
Data Analysis Software
125
5.4
Profile of Sample Units
126
5.4.1 Industry-wise Composition of Sample Units
126
5.4.2 Size-wise Composition of Sample Units
127
5.4.3 Year of Establishment of Sample Units
128
5.4.4 Workforce: Employee Composition
128
5.4.5 Education Profile of Workforce
129
5.5
Basic Statistical Measures
130
5.5.1 Statistical Tables
130
5.5.2 Bar Charts
136
5.5.3 Descriptive Data Analysis
150
5.6
Cronbach Alpha Reliability Test
150
5.7
Kolmogorov-Smirnov One-Sample NP Test for
153
Normality (H1 to H5) 5.8
Hypothesis Testing (H1 to H5)
156
5.8.1 Spearman’s Correlations (H1 to H5)
157
5.8.2 One Sample Chi-Square NP Test (H1 to H5)
159
5.8.3 H1 Test: Mann-Whitney Rank Sum U Test
162
5.8.4 H2 Test: Friedman One-Way ANOVA & Logistic
163
Regression test for Core Competencies 123
Para No.
Particulars
5.8.5 H3 Test: Friedman One-Way ANOVA & Logistic
Page No. 170
Regression test for ‘Quality Consciousness’ 5.8.6 H4 Test: Friedman One-Way ANOVA & Logistic
179
Regression test for ‘Planning & Organising’ 5.8.7 H5 Test: Mann-Whitney Rank Sum U Test for ‘Training’
185
5.8.8 Need for Additional Logistic Regression Tests
186
5.8.9 Logistic Regression test for Functional competencies
186
5.8.10 Logistic Regression test for Overall Competency level
192
with Extraneous Variables: Size, Type, Attrition level, and Hiring of Qualified Managers 5.8.11 Hypothesis Tests’ Confirmation : Causal Path Analysis
194
5.9
198
Factor Analysis Tests (Core, Leadership and Functional Competencies)
5.9.1 Factor Analysis of Core Competencies (Question 16)
199
5.9.2 Factor Analysis of Leadership Competencies (Question
205
17) 5.9.3 Factor Analysis of Functional Competencies (Question 15)
124
207
CHAPTER 5 DATA ANALYSIS AND HYPOTHESIS TESTING “The machinery of methodology occupies a very important position in all kinds of research. The research cannot perform its function without it, since it is methodology which lays out the way that formal research is to be carried out and outlines the detailed description of the research variable and procedure”. (Barr, 1960, p.1160). 5.1
Data Tabulation The data was tabulated using MS Excel and the variables were suitably named
so as to facilitate interpretation of results. The summary of data has been displayed in Appendix ‘D’ at the end of the report. 5.2
Data Collection Methodology Assuming a response rate of 90% due to co-location of all the units within a
small area and convenience of collecting the data through personally visiting the respondents, it was decided to attempt data collection from 108 respondents. Questionnaires were sent out initially by mail to the official addresses as they appeared in these directories to all 108 sample units. None responded for various possible reasons explained in Chapter 4. After approaching a number of agencies specified in Chapter 4, the researcher personally visited 108 MSMEs, out of which 100 agreed to fill up the questionnaires. In a number of cases, especially in micro and smaller enterprises, the owners had a number of doubts about the terminology used in the questionnaire for various types of employee competencies, but once explained with the help of an associate who could fluently speak in the local language, they did not take much time in answering the questions. The names of enterprises have been given in an alphabetical order (Appendix ‘C’) separately so as to maintain confidentiality of data as promised to the respondents during the survey. 5.3
Data Analysis Software IBM‟s Statistical Package for Social Studies (SPSS) version 17.0 was used for
data analysis. Questions 1 to 6, 9 and 10 were not analysed as they covered general data about the enterprises whose demographic details have been made available at Appendix ‘D’. However, data in Questions 7, 8 and 11 were used to find whether MSMEs belonging to different categories based on their size (Question 7), type of 125
industry (Question 8) and attrition level (Question 11) differ in their relationships. Most of the tests dealt with Questions 12 to 18 which involve the dependent and independent variables. 5.4
Profile of Sample Units (Questions 6 – 10)
5.4.1
Industry-wise Composition of Sample Units PCMC Bhosari MIDC had
2266 units whose industry-wise composition was obtained from PCMC MIDC R&D, and from these a sample size of 108 MSMEs had been selected proportionately, as explained in Chapter 4 above. Out of these, 100 responded (92.6% response rate) whose composition is given in Table 5.1 and Figure 5.1 below. Table 5.1: Industry-wise Composition of PCMC Sample Units (Question 8) S. No.
Type of Industry
No of Units
Sample Size
1.
Metallic products and fabrication
886
39
2.
Auto & Auto ancillaries
564
25
3,
Machine and Machine tools
259
11
4.
Processing
190
8
5.
Electrical and electronics
106
5
6.
Others (service, packaging, transport etc.)
261
12
Total
2266
100
Figure 5.1: Type-wise Distribution of Enterprises (Question 8)
1. Metallic products and fabrication - 39 2. Auto and Auto-ancillaries – 25 3. Machine and machine parts – 11 4. Processing industry - 8 5. Electricals and Electronics - 5 6. Others - 12
126
5.4.2
Size-wise Composition of Sample Units MSMEs are classified according to
their size of investments as per MSMED Act 2006, as explained in Chapter 1 above. Profile of the 100 responding sample units based on investment in assets disclosed by the owners in their questionnaires is shown in Table 5.2 and Figure 5.2 below. The majority 75% of them were Small enterprises which is approximately representative of their overall representation in the PCMC MIDC areas as well. Table 5.2 Size-wise Composition of Sample Units (Question 7) S. No
Size of MSME
No of Units
1.
Micro (Rs. 10 to 25 lakh)
10
2.
Small (Rs. 25 to 500 lakh)
75
3.
Medium (Rs. 500 to 1000 lakh)
15
Total
100
Figure 5.2: Size-wise Distribution of Enterprises (Question 7)
1. Micro – 10 2. Small – 75 3. Medium - 15
127
5.4.3
Year of Establishment of Sample Units
The age of sample units as
obtained from their questionnaires is given in Table 5.3 below. 41% of these were established prior to the year 2000 AD. Again the majority of 59% were established after 2000 AD when the MSME sector grew rapidly as explained in Chapter 3 above. It was ensured that none of the sample units were less than three years old. Accordingly, the youngest sample units belong to the year 2010 AD. Table 5.3: Classification of Main Survey Units by Year of Establishment (Question 6) S. No.
Year
No of Units
% of Total
1.
1960s
02
02%
2.
1970s
04
04%
3.
1980s
10
10%
4.
1990s
25
25%
5.
2000
04
04%
6.
2001
02
02%
7.
2002
05
05%
8.
2003
05
05%
9.
2004
05
05%
10.
2005
06
06%
11.
2006
06
06%
12.
2007
08
08%
13.
2008
07
07%
14.
2009
06
06%
15.
2010
05
05%
Total
100
100%
5.4.4
Workforce: Employee Composition In the 100 units surveyed, total work
force was 4567 persons. Out of these 4327 (95%) were working on regular basis, while 240 (5%) were casual, who were employed on ad-hoc basis. Their details are displayed in Table 5.4 below:
128
Table 5.4: Category-wise Distribution of Employees (Question 9) S. No.
i.
Category
No of Persons
% of Total
1.
Managerial (M)
256
5.6 %
2.
Supervisory (S)
423
9.3 %
3.
Office Staff (OS)
418
9.1 %
4.
Workers (W)
3470
76.0 %
The overall average employee profile in the 100 sample units was 45.67 employees (2.56 Managers, 4.23 Supervisory staff, 4.18 Office Staff and 34.70 Workers).
ii.
The 10 Micro enterprises had 60 employees, giving an average of 06 employees each (1 Managerial, 0.5 Supervisory, Nil Office Staff and 4.6 Workers);
iii.
15 Medium enterprises had 1730 employees, with an average of 115 employees each (7 Managerial, 13 Supervisory, 10 Office Staff, and 86 Workers).
iv.
75 Small enterprises had 2777 employees, with an average of 37 employees each (1.9 Managers, 3.0 Supervisory, 3.5 Office Staff, 28.4 Workers).
5.4.5 Education Profile of Workforce Work-force education profile as compiled from the survey data collected from 100 sample units is given in Table 5.5 below.
129
Table 5.5: Education Profile of Employees (Question 10) S. No.
5.5
Education Level
No. of Persons
% of Total
1.
Primary school
1260
27.6
2.
Matriculation
2136
46.8
3.
Diploma / BA / B Sc / B Com
847
18.5
4.
BE / M Tech
239
5.2
5.
PG / MBA
85
1.9
Basic Statistical Measures (Questions 12 – 20) Basic statistical measures were analysed primarily to examine for normality of
data. Mean values were also examined to indicate which specific employee competencies were the most preferred by the MSME employers, thus giving a preliminary idea of the applicability of
the hypotheses statements.
Descriptive
statistics of survey questionnaire in tabular form, and in the form of bar charts are shown in sub-paragraphs 5.5.1 and 5.5.2 below. Analysis of these descriptive statistics is given in sub-paragraph 5.5.3 below. 5.5.1 Statistical tables Statistical measures of survey questionnaire data consisting of number of valid and missing cases; Measures of average: Arithmetic Mean, Median and Mode; Measures of Variability: Standard deviation and Variance; Measures of shape: Skewness and Kurtosis; Standard errors of all measures of mean, are given in Tables 5.6 A, B and C below, as obtained from researcher‟s own calculations using SPSS 17.0.
130
Table 5.6 A: Descriptive Statistics of Survey Questionnaire (Questions 12 to 20) N S. No.
Item in Questionnaire
Range
Minimum
Maximum Median
StatisStatisStatistic tic Statistic tic Statistic
1.
Actual Profitability (ActProf)
100
1
1
2
2.00
2.
Overall Competency level (Comp)
100
2
3
5
4.00
3.
Functional competencies (Func)
100
3
2
5
4.00
4.
Leadership competencies (Ldr)
100
2
3
5
4.00
5.
Core Competencies (Core)
100
2
3
5
4.00
6.
Technical skill level (Tech-Func1)
100
3
2
5
4.00
7.
Approach towards Learning & Self Development (LD-Func2)
100
3
2
5
4.00
Adaptability to new technology and change (Techno-Func3)
100
4
1
5
4.00
Availability of Specialized skills (Spec-Func4)
100
4
1
5
4.00
10. Quality Consciousness (Qual-Core1)
100
3
2
5
5.00
11. Customer focus (Cust-Core2)
100
3
2
5
4.00
12. Cost consciousness (Cost-Core3)
100
3
2
5
4.00
13. Healthy work environment / Safety norms (Safe-Core4)
100
5
0
5
3.00
14. Team spirit (Team-Core5)
100
3
2
5
4.00
15. Creativity / Innovativeness (Innov-Core6)
100
4
1
5
4.00
16. Strategic thinking (Strat-Ldr1)
100
3
2
5
4.00
17, Interpersonal skills (Interper-Ldr2)
100
3
2
5
4.00
18. Planning & Organising skills (Plg-Ldr3)
100
3
2
5
4.00
19. Decision-making (DM-Ldr4)
100
3
2
5
4.00
8. 9.
131
N S. No.
Item in Questionnaire
Statistic
Range
Minimum
MaxiMedian mum
StatisStatisStatistic Statistic tic tic
20. Problem solving ability (ProbSol-Ldr5)
100
2
3
5
4.00
21. Training by enterprise in employees‟ competency development (Trg)
100
3
2
5
4.00
22. Impact of HRM on Profitability (HRProf)
100
4
1
5
4.00
23. Impact of Financial aspects on Profitability (FinProf)
100
2
3
5
4.00
24. Impact of Technological aspects on Profitability (TechProf)
100
4
1
5
4.00
25. Impact of Marketing on Profitability (MktgProf)
100
3
2
5
4.00
26. Impact of Operational aspects on Profitability (OperProf)
100
3
2
5
4.00
27. Impact of hiring professionally qualified managers on Profitability (QualMgrProf)
100
4
1
5
3.00
Valid N (listwise)
100
132
Table 5.6 B: Descriptive Statistics of Survey Questionnaire (Questions 12 to 20) S. No.
Mean Item in Questionnaire
Std. Deviation
Variance
Statistic
Statistic
.219
.048
1.
Actual Profitability (ActProf)
1.95
Std. Error .022
2.
Overall Competency level (Comp)
4.11
.057
.567
.321
3.
Functional competencies (Func)
3.95
.069
.687
.472
4.
Leadership competencies (Ldr)
4.14
.067
.667
.445
5.
Core Competencies (Core)
4.27
.066
.664
.442
6.
Technical skill level (Tech-Func1)
3.91
.090
.900
.810
7.
Approach towards Learning & Self
3.59
.073
.726
.527
(Techno-Func3)
3.64
.081
.811
.657
Availability of Specialized skills (Spec-Func4)
3.83
.107
1.074
1.153
10. Quality Consciousness (Qual-Core1)
4.49
.072
.718
.515
11. Customer focus (Cust-Core2)
4.26
.071
.705
.497
12. Cost consciousness (Cost-Core3)
3.75
.077
.770
.593
13. Healthy work environment / Safety norms
3.40
.083
.829
.687
14. Team spirit (Team-Core5)
4.09
.068
.683
.467
15. Creativity / Innovativeness (Innov-Core6)
3.55
.094
.936
.876
16. Strategic thinking (Strat-Ldr1)
4.02
.072
.724
.525
17, Interpersonal skills (Interper-Ldr2)
4.17
.074
.739
.547
18. Planning & Organising skills (Plg-Ldr3)
4.30
.067
.674
.455
19. Decision-making (DM-Ldr4)
4.19
.071
.706
.499
Statistic
Development (LD-Func2) 8.
9.
Adaptability to new technology and change
(Safe-Core4)
133
Mean
S. No.
Item in Questionnaire
Std. Deviation
Variance
Statistic
Statistic
.652
.425
20. Problem solving ability (ProbSol-Ldr5)
4.17
Std. Error .065
21. Training by enterprise in employees‟
4.08
.080
.800
.640
22. Impact of HRM on Profitability (HRProf)
3.51
.078
.785
.616
23. Impact of Financial aspects on Profitability
4.11
.063
.634
.402
4.24
.081
.806
.649
4.26
.079
.787
.619
26. Impact of Operational aspects on Profitability (OperProf)
4.08
.063
.631
.398
27. Impact of hiring professionally qualified managers on Profitability (QualMgrProf)
3.03
.110
1.096
1.201
Statistic
competency development (Trg)
(FinProf) 24. Impact of Technological aspects on Profitability (TechProf) 25. Impact of Marketing on Profitability (MktgProf)
134
Table 5.6 C: Descriptive Statistics of Survey Questionnaire (Questions 12 to 20) Kurtosis Skewness S. Item in Questionnaire Std. Std. No. Statistic Error Statistic Error 1. Actual Profitability (ActProf) 15.896 .478 .021 .241 2.
Overall Competency level (Comp)
.083
.478
-.507
.241
3.
Functional competencies (Func)
.714
.478
-.166
.241
4.
Leadership competencies (Ldr)
-.736
.478
-.364
.241
5.
Core Competencies (Core)
-.750
.478
-.414
.241
6.
Technical skill level (Tech-Func1)
-.630
.478
-.644
.241
7.
Approach towards Learning & Self Development .086
.478
-.641
.241
(Techno-Func3)
.540
.478
-.902
.241
Availability of Specialized skills (Spec-Func4)
.295
.478
-1.386
.241
10. Quality Consciousness (Qual-Core1)
1.650
.478
-.593
.241
11. Customer focus (Cust-Core2)
-.119
.478
-.211
.241
12. Cost consciousness (Cost-Core3)
-.253
.478
-.652
.241
1.860
.478
-.503
.241
14. Team spirit (Team-Core5)
.572
.478
-.336
.241
15. Creativity / Innovativeness (Innov-Core6)
-.112
.478
-.518
.241
16. Strategic thinking (Strat-Ldr1)
.390
.478
-.589
.241
17, Interpersonal skills (Interper-Ldr2)
.041
.478
-.646
.241
18. Planning & Organising skills (Plg-Ldr3)
.228
.478
-.639
.241
19. Decision-making (DM-Ldr4)
.476
.478
-.185
.241
20. Problem solving ability (ProbSol-Ldr5)
-.661
.478
-.508
.241
(LD-Func2) 8.
9.
Adaptability to new technology and change
13. Healthy work environment / Safety norms (SafeCore4)
135
Kurtosis S.
Item in Questionnaire
No.
Statistic
Std. error
Skewness Statistic
Std. Error
21. Training by enterprise in employees‟ competency development (Trg)
-.325
.478
-.417
.241
.328
.478
-.092
.241
-.496
.478
-1.176
.241
2.023
.478
-.625
.241
26. Impact of Marketing on Profitability (MktgProf)
-.698
.478
-.308
.241
28. Impact of hiring professionally qualified managers
-.924
.478
.021
.241
22. Impact of HRM on Profitability (HRProf) 24. Impact of Financial aspects on Profitability (FinProf) 25. Impact of Technological aspects on Profitability (TechProf)
on Profitability (QualMgrProf) 5.5.2 Bar Charts Responses to Questions 12 to 20 represented in the form of Bar Charts derived from SPSS 17.0 are shown in Figures 5.3 to 5.29 below. Question 12: Profitability of Enterprise in last three years – Profitable (2) / Not profitable (1): Figure 5.3: Profitability response (Question 12 - 95% profitable)
136
Question 13: Impact of Overall Competency level on Profitability of Enterprise: Figure 5.4: Impact of Overall Competency level on Profitability (Question 13 - Mean 4.11)
Question 14: Impact of Type of Employee Competency Factor on Profitability of enterprise: Figure 5.5: Impact of Functional competencies on Profitability (Question 14(a) - Mean 3.95)
137
Figure 5.6: Impact of Leadership competencies on profitability (Question 14(b) - Mean 4.14)
Figure 5.7: Impact of Core competencies on Profitability (Question 14(c) - Mean 4.27)
138
Question 15: Impact of available Functional Competencies on Profitability of enterprise: Figure 5.8: Impact of Technical skills on Profitability (Question 15(a) - Mean 3.91)
Figure 5.9: Impact of Approach of Learning and Self Development on Profitability (Question 15(b) – Mean 3.59)
139
Figure 5.10: Impact of Adaptability to New Technology and Change on Profitability (Question 15(c) - Mean 3.64)
Figure 5.11: Impact of Availability of Specialised Skills on Profitability (Question 15(d) - Mean 3.83)
140
Question 16: Impact of available Core competencies on Profitability of enterprise: Figure 5.12: Impact of Quality Consciousness on Profitability (Question 16(a) - Mean 4.49)
Figure 5.13: Impact of Customer Focus on Profitability (Question 16(b) - Mean 4.26)
141
Figure 5.14: Impact of Cost Consciousness on Profitability (Question 16(c) - Mean 3.75)
Figure 5.15: Impact of Healthy Work Environment and Safety Norms on Profitability (Question 16(d) - Mean 3.40)
142
Figure 5.16: Impact of Team Spirit on Profitability (Question 16(e) - Mean 4.09)
Figure 5.17: Impact of Creativity / Innovativeness on Profitability (Question 16(f) - Mean 3.55)
143
Question 17: Impact of available Leadership competencies on Profitability of enterprise: Figure 5.18: Impact of Strategic Thinking on Profitability (Question 17(a) - Mean 4.02)
Figure 5.19: Impact of Interpersonal Skills on Profitability (Question 17(b) - Mean 4.17)
144
Figure 5.20: Impact of Planning and Organising ability on Profitability (Question 17(c) - Mean 4.30)
Figure 5.21: Impact of Decision-making on Profitability (Question 17(d) - Mean 4.19)
145
Figure 5.22: Impact of Problem solving ability on Profitability (Question 17(e) - Mean 4.17)
Question 18: Impact of Training provided by enterprise in employees‟ competency development on Profitability of enterprise: Figure 5.23: Impact of Training provided by enterprise for employee competency development on Profitability (Question 18 - Mean 4.08)
146
Question 19: Impact of other general management aspects on Profitability of enterprise: Figure 5.24: Impact of HRM aspects on Profitability (Question 19(a) - Mean 3.51)
Figure 5.25: Impact of Financial management aspects on Profitability (Question 19(b) - Mean 4.11)
147
Figure 5.26: Impact of Technological aspects on Profitability (Question 19(c) - Mean 4.24)
Figure 5.27: Impact of Marketing aspects on Profitability (Question 19(d) - Mean 4.26)
148
Figure 5.28: Impact of Operational aspects on Profitability (Question 19(e) - Mean 4.08)
Question 20: Impact of hiring of professionally qualified managers (QualMgr) on profitability of enterprise: Figure 5.29: Impact of Hiring of Qualified Managers on Profitability (Question 20 - Mean 3.03)
149
5.5.3
Descriptive Data Analysis Examination of the data in Tables 5.6 A, B, and
C above as well as the bar charts indicated that some of the data was not displaying normality. Gregory and Mallery (2011) have stated that if the mean is above median value, and the skewness and kurtosis values vary between +1.0 and -1.0, this data is likely to display normality. While most of the variables displayed such characteristics, quite a few did show notable departures. Examination of the bar charts also displayed that most of the variables displayed a marked negative skewness. Besides, likert scale data being ordinal in nature does not satisfy the criteria for parametric tests. Hence it was decided to adopt non-parametric tests, and logistic regression which is not very sensitive to non-normality of data, was considered for data analysis. Core competencies had the highest mean value amongst the three competency factors, supporting H2. Quality Consciousness had the highest mean value amongst all 15 competencies supporting H3 as well as indicating its predominance amongst all the variables. Planning and organising competency had the highest mean value amongst all the leadership competencies, supporting H4. 5.6
Cronbach Alpha Reliability Test Test values greater than 0.8 indicate good reliability. Details of this test for
responses to Questions 12 to 20 are displayed in Tables 5.7 A to D below: Table 5.7 A: Case Processing Summary - Questions 12 - 20
Cases
Valid Excludeda Total
N
%
100
100.0
0
.0
100
100.0
a. Listwise deletion based on all variables in the procedure. Table 5.7 B: Reliability Statistics - Questions 12 - 20 Cronbach's Alpha
Cronbach's Alpha Based on Standardized Items
N of Items
.885
.893
27
150
Table 5.7 C: Item-Total Statistics - Questions 12 - 20 Question No.
Variable
Scale Mean if Item Deleted
Scale Corrected Variance if Item-Total Item Correlation Deleted 102.869 .421
12
Profitability
103.14
13
Competency level
100.98
97.232
.648
14 (a)
Functional competencies
101.14
97.778
.481
14 (b)
Leadership competencies
100.95
99.098
.395
14 (c)
Core competencies
100.82
97.099
.554
15 (a)
Technical skills
101.18
96.048
.450
15 (b)
Approach to Learning
101.50
97.566
.467
15 (c)
Adaptability to New
101.45
94.452
.614
Technology & Change 15 (d)
Specialised skills
101.26
94.720
.427
16 (a)
Quality Consciousness
100.60
96.444
.556
16 (b)
Customer focus
100.83
98.506
.413
16 (c)
Cost Consciousness
101.34
101.015
.205
16 (d)
Healthy work environment/
101.69
97.428
.408
Safety norms 16 (e)
Team spirit
101.00
96.141
.611
16 (f)
Creativity & Innovativeness
101.54
96.049
.429
151
Question No.
Variable
Scale Mean Scale Corrected if Item Variance if Item-Total Deleted Item Deleted Correlation 101.07 96.652 .534
17 (a)
Strategic thinking
17 (b)
Interpersonal skills
100.92
97.509
.461
17 (c)
Planning & Organising
100.79
97.966
.477
17 (d)
Decision-making
100.90
97.242
.506
17 (e)
Problem solving
100.92
98.155
.480
19 (a)
HRD aspects
101.58
97.438
.435
19 (b)
Financial aspects
100.98
101.171
.252
19 (c)
Technological aspects
100.85
95.462
.551
19 (d)
Marketing aspects
100.83
100.203
.252
19 (e)
Operational aspects
101.01
100.050
.344
Hiring of Qualified Managers
102.06
97.309
.290
20
Table 5.7 D: Cronbach’s Alpha if Item deleted - Questions 12 - 20 Question No.
Item in Questionnaire
Cronbach's Alpha if Item Deleted
12
Profitability
.884
13
Competency level
.878
14 (a)
Functional competencies
.880
14 (b)
Leadership competencies
.882
14 (c)
Core competencies
.879
15 (a)
Technical skills
.881
15 (b)
Approach to Learning
.881
15 (c)
Adaptability to New Technology & Change
.877
15 (d)
Specialist skills
.883
16 (a)
Quality Consciousness
.879 152
Question No.
Item in Questionnaire
Cronbach's Alpha if Item Deleted
16 (b)
Customer focus
.882
16 (c)
Cost Consciousness
.887
16 (d)
Healthy work environment/ Safety norms
.882
16 (e)
Team spirit
.878
16 (f)
Creativity & Innovativeness
.882
17 (a)
Strategic thinking
.879
17 (b)
Interpersonal skills
.881
17 (c) Planning & Organising
.880
17 (d)
Decision-making
.880
17 (e)
Problem solving
.880
Training
.875
19 (a)
HRD aspects
.881
19 (b)
Financial aspects
.885
19 (c)
Technological aspects
.878
19 (d)
Marketing aspects
.886
19 (e)
Operational aspects
.883
Hiring of Qualified Managers
.887
18
20
Table 5.7 B showed a Cronbach alpha value of 0.885 which was highly reliable. It was very close to that based on standardised items 0.893 which was based on Z value. This could be further improved marginally as Table 5.7 D showed that deletion of the extra Question 20 on „Hiring of Qualified Managers‟ would raise the Alpha value to 0.887. Next step was to check the normality of the pilot sample data using nonparametric tests as they followed the likert scale which is partly ordinal. 5.7
Kolmogorov-Smirnov One-Sample NP Test for Normality This tests whether the distribution of the members of a single group differ
significantly from a Normal or Uniform or Poisson or Exponential distribution. Significance values close to zero would indicate that the sample distribution actually 153
displays a non-normal distribution (George and Mallory p.216). Table 5.8 A below indicates the test values for Questions 12, 13 and 14 variables, Table 5.8 B for Question 15 variables, Table 5.8 C for Question 16 variables, Table 5.8 D for Questions 17 and 18 variables, Table 5.8 E for Questions 19 and 20. Table 5.8 A : One-Sample Kolmogorov-Smirnov Test – Questions 12, 13 & 14 Profitability
Competency Functional Leadership level
comp
comp
Core Comp
N
100
100
100
100
100
Mean
1.95
4.11
3.95
4.14
4.27
Std. Deviation
.219
.567
.687
.667
.664
Most Extreme Absolute
.540
.357
.329
.283
.268
Differences
Positive
.410
.357
.291
.283
.268
Negative
-.540
-.313
-.329
-.257
-.254
Kolmogorov-
5.403
3.570
3.290
2.831
2.678
.000
.000
.000
.000
.000
Normal a,,b
Parameters
Smirnov Z Asymp. Sig. (2tailed) a. Test distribution is Normal.
b. Calculated from data.
Table 5.8 B: One-Sample Kolmogorov-Smirnov Test - Question 15 Technical
LD
Techno
Specialist
Comp
Comp
Comp
Comp
N
100
100
100
100
Normal
Mean
3.91
3.59
3.64
3.83
Parameters
Std. Deviation
.900
.726
.811
1.074
Most Extreme
Absolute
.230
.344
.312
.273
Differences
Positive
.170
.236
.228
.147
Negative
-.230
-.344
-.312
-.273
Kolmogorov-Smirnov Z
2.298
3.439
3.115
2.729
Asymp. Sig. (2-tailed)
.000
.000
.000
.000
154
Table 5.8 C One-Sample Kolmogorov-Smirnov Test - Question 16 Quality Consciou sness 100
N
Cust Comp
Cost Comp
Safe Comp
Team Comp
Innov Comp
100
100
100
100
100
Mean
4.49
4.26
3.75
3.40
4.09
3.55
Std. Deviation
.718
.705
.770
.829
.683
.936
Absolute
.361
.253
.277
.235
.298
.225
Positive
.239
.244
.223
.215
.292
.182
Negative
-.361
-.253
-.277
-.235
-.298
-.225
Kolmogorov-Smirnov Z
3.613
2.530
2.772
2.355
2.976
2.246
Asymp. Sig. (2-tailed)
.000
.000
.000
.000
.000
.000
a. Test distribution is Normal.
b. Calculated from data.
Table 5.8 D: One-Sample Kolmogorov-Smirnov Test - Questions 17 & 18
Strat Comp
Interper- Planning sonal & Orgacomp nising 100 100
N
100
Mean
4.02
4.17
Std. Deviation
.724
Absolute
DM Comp
ProbSol Training Comp
100
100
100
4.30
4.19
4.17
4.08
.739
.674
.706
.652
.800
.299
.249
.262
.266
.293
.240
Positive
.271
.241
.262
.266
.293
.210
Negative
-.299
-.249
-.260
-.264
-.257
-.240
Kolmogorov-Smirnov Z
2.990
2.491
2.618
2.660
2.928
2.402
Asymp. Sig. (2-tailed)
.000
.000
.000
.000
.000
.000
155
Table 5.8 E One-Sample Kolmogorov-Smirnov Test- Questions 19 & 20
Normal
Oper
Qual
Prof
Prof
Mgr
100
100
100
100
4.11
4.24
4.26
4.08
3.03
.785
.634
.806
.787
.631
1.096
Absolute
.274
.309
.247
.287
.320
.242
Positive
.202
.309
.197
.173
.320
.166
Negative
-.274
-.281
-.247
-.287
-.310
-.242
Kolmogorov-
2.738
3.089
2.473
2.866
3.205
2.419
.000
.000
.000
.000
.000
.000
HR
Fin
Tech
Mktg
Prof
Prof
Prof
N
100
100
Mean
3.51
Std. Deviation
Parametersa,,b
Most Extreme Differences
Smirnov Z Asymp. Sig. (2tailed) a. Test distribution is Normal. b. Calculated from data. The Kolmogorov-Smirnov significance values for all variables were 0.000 indicating non-normality, confirming the requirement of non-parametric tests. 5.8
Hypothesis Testing As brought out in Chapter 4 earlier, two or more non-parametric tests were
adopted for testing of each hypothesis, as these are inherently weaker than their parametric equivalents, and more than one positive result would help in confirming the hypothesis better. Firstly, Spearman‟s Correlation Coefficients (rho), (NP equivalent of Pearson‟s „r‟ Coefficients applicable for parametric data) were calculated to bring out relevant correlations for data collected from Questions 12 to 20, applicable to all Hypotheses, H1 to H5. Next, One-Sample Chi-Square NP Test 156
(NP equivalent of cross-tabulated t-test for two independent samples applicable for parametric data) was conducted for testing the independence between the dependent and independent variables for data collected from Questions 12 to 20, applicable to all Hypotheses, H1 to H5. Thereafter, following tests were used for testing of each hypothesis: H1 Test: Mann - Whitney Rank Sum U-Test. H2 Test: Friedman One-Way ANOVA Test, and Logistic Regression Test H3 Test: Friedman One-Way ANOVA Test, and Logistic Regression Test H4 Test: Friedman One-Way ANOVA Test, and Logistic Regression Test H5 Test: Mann - Whitney Rank Sum U-Test. Logistic Regression tests were applied to Functional Competencies (Question 15), and to Overall Competency level (Question 13) with external variables to further support the study hypotheses, as well as find out recommended measures for improving the current MSME scenario. Finally, Causal Path Analysis with Partial Ordinal Coefficients was conducted for confirming all the Hypotheses test results. 5.8.1
Spearman’s Correlations for H1 to H5 This is a bivariate correlation based on the rank order of values which had
been adopted instead of the more popular Pearson‟s Correlations because all the variables were not normally distributed. Higher correlation coefficients indicate more positive relationship between the variables. The Spearman‟s correlations are shown in Tables 5.9 A, B and C below:
Table 5.9 A: Correlations – Questions 12, 13, 14 and part 15 Compe Func- Leader Techni LD TechProfi- -tency tional -ship Core -cal Com no tability level comp comp Comp Comp p Comp Spear- Profi- Correlation man's tability Coefficient rho Sig. (2-tailed) N
1.000
.427**
.153
.173
.
.000
.128
.085
.001
.009
.020
.001
100
100
100
100
100
100
100
100
157
.332** .258** .232* .341**
Table 5.9 B: Correlations – Questions 15, 16, and part 17 Quality Spec- Conialised scious Cust Cost Safe Team Innov Strat Comp -ness Comp Comp Comp Comp Comp Comp Spear- Profita Correlation man's -bility Coefficient rho Sig. (2tailed) N
Interpersonal comp
.231* .263** .093
.013
.101
.180
.226*
.162
.060
.021
.008
.356
.898
.316
.074
.024
.107
.554
100
100
100
100
100
100
100
100
100
Table 5.9 C: Correlations – Questions 17, 18, 19 and 20 PlanProbning & Sol Trai- HR Fin Tech Mktg Oper Qual Orga- DM nising Comp Comp ning Prof Prof Prof Prof Prof Mgr Spear- Profita- Correlation man's bility Coefficient rho Sig. (2tailed)
.211*
.140
.183 .339** .233* .183 .154 -.047 .253* .131
.035
.164
.069
.001
.019 .068 .127 .640 .011 .193
100
100
100
100
100
N
100
100
100 100 100
** Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2-tailed). The variables of interest and their corresponding Spearman‟s correlation coefficients and significance values have been highlighted in bold letters and italics. Spearman‟s correlation between the dependent variable Profitability (Question 12) and the independent variable Overall Competency level (Question 13) is 0.427 and is significant at the 0.01 level indicating a strong positive relationship. This supports H1. Spearman‟s correlation between the dependent variable Profitability (Question 12) and the independent variable Core Competency factor (Question 14) is 0.332 and is significant at the 0.01 level indicating a strong positive relationship. None of the other two i.e. Functional and Leadership competency factors have significant correlations. This supports H2. 158
Spearman‟s correlation between the dependent variable Profitability (Question 12) and the independent variable under Functional competency factor i.e. Adaptability to new technology and change (Question 15), is 0.341 and is significant at the 0.01 level indicating a strong positive relationship. Spearman‟s correlation between the dependent variable Profitability (Question 12) and the independent variable under Core competency factor i.e. Quality Consciousness (Question 16), is 0.263 and is significant at the 0.01 level indicating a strong positive relationship. Another core competency Creativity/Innovativeness also is correlated with a correlation of 0.226 but is significant at the 0.05 level indicating a lower level of relationship. This supports H3. Spearman‟s correlation between the dependent variable Profitability (Question 12) and the independent variable under Leadership competency factor i.e. Planning and Organising ability (Question 17) is 0.427 and is significant at the 0.01 level indicating a strong positive relationship. None of the other Leadership traits show significant correlation. This supports H4. Spearman‟s correlation between the dependent variable Profitability (Question 12) and the independent variable Training in employees skills / competency development provided by enterprise (Question 18), is 0.339 and is significant at the 0.01 level indicating a strong positive relationship. This supports H5. 5.8.2 One-Sample Chi-Square NP Test for H1 to H5 This procedure conducts a one-sample chi square test rather than the more traditional chi-square test of cross-tabulated data. The chi-square test is a test of independence between the observed and expected values. Small significance values would demonstrate that the sample values deviate from the expected values thus discrediting the null hypotheses (George and Mallery p.217). This test was necessary for all the variables involved in H1 to H5, to confirm the results derived from other tests. Tables 5.10 A to F show the observed and expected values in respect of all variables associated with hypotheses H1 to H5, and their chi-square statistics in Table 5.11 below.
159
Table 5.10 A: Profitability - Question 12 (Dependent variable H1 to H5) Observed N
Expected N
Residual
1
5
50.0
-45.0
2
95
50.0
45.0
Total
100 Table 5.10 B: Overall Competency level - Question 13 (Independent variable H1) Observed N
Expected N
Residual
3
11
33.3
-22.3
4
67
33.3
33.7
5
22
33.3
-11.3
Total
100
Table 5.10 C: Core Competencies - Question 14(c) (Independent variable H2) Observed N Expected N
Residual
3
12
33.3
-21.3
4
49
33.3
15.7
5
39
33.3
5.7
Total
100
Table 5.10 D: Quality Consciousness - Question 16(a) (Independent variable H3) Observed N
Expected N
Residual
2
2
25.0
-23.0
3
7
25.0
-18.0
4
31
25.0
6.0
5
60
25.0
35.0
Total
100
160
Table 5.10 E: Planning & Organising - Question 17(c) (Independent variable H4) Observed N Expected N
Residual
2
1
25.0
-24.0
3
9
25.0
-16.0
4
49
25.0
24.0
5
41
25.0
16.0
Total
100
Table 5.10 F: Training – Question 18 (Independent variable H5) Observed N Expected N
Residual
2
3
25.0
-22.0
3
19
25.0
-6.0
4
45
25.0
20.0
5
33
25.0
8.0
Total
100
Table 5.11: One-Sample Chi-Square Test Statistics – Questions 12 – 18 (H1 to H5) Profitability Chi-Square df Asymp. Sig.
Core Competency Competenlevel cies
Quality Consciousness
Planning & Training Organising
81.000a
52.820b
21.980b
84.560c
66.560c
39.360c
1
2
2
3
3
3
.000
.000
.000
.000
.000
.000
a. 0 cells (.0%) have expected frequencies less than 5. The minimum expected cell frequency is 50.0. b. 0 cells (.0%) have expected frequencies less than 5. The minimum expected cell frequency is 33.3. c. 0 cells (.0%) have expected frequencies less than 5. The minimum expected cell frequency is 25.0. Visual inspection of the differences between observed and expected values in Tables 5.10 A to 5.10 F reveal wide discrepancies. The large chi-square (x2) values in Table 5.11 also substantiate these large discrepancies – the larger the discrepancy, the 161
larger the chi-square value e.g. „profitability (81.000)‟ and „quality consciousness (84.560)‟. The extremely small significance level associated with the one-sample chisquare analysis (p = 0.000) demonstrates that each ordinal level within each of the variables deviate substantially from the expected values (equal frequency of each ordinal level). This test confirms that the independent and dependent variables in each of the hypotheses H1 to H5 are not independent of each other, thus supporting all five hypotheses. 5.8.3
H1 Test
H1: MSMEs with higher competency levels have greater profitability Dependent Variable: Profitability (Question 12) Independent Variable: Overall competency level (Question 13) H1 has already been supported by positive results of Spearman‟s Correlations and One-Sample Chi-Square NP Test brought out above. Positive result in Mann-Whitney Rank-Sum U Test will lead to validation of the Hypothesis. 5.8.3.1 Mann-Whitney Rank-Sum U Test for H1 This non-parametric equivalent of the t-test for two independent samples has been used by a number of researchers McMahon (2001), Dyba et al. (2005)231 for hypothesis testing. Tables 5.12 A and 5.12 B display the test values for H1. Table 5.12 A: Ranks (H1 - Questions 12 and 13)
Competency level
Profitability
N
Mean Rank
Sum of Ranks
1
5
6.00
30.00
2
95
52.84
5020.00
Total
100
231
Dyba,T. Kampenes, V.B. and Sjoberg, D.I.K. (2005). A systematic review of statistical power in software engineering experiments. Information and Software Technology, 48(2006), 745-755. Retrieved on 15 Sep 2014 from .
162
Table 5.12 B: Test Statisticsa (H1 - Questions 12 and 13) Competency level Mann-Whitney U
15.000
Wilcoxon W
30.000
Z
-4.245
Sig. (2-tailed)
.000
a. Grouping Variable: Profitability Table 5.12 A shows that the mean value 52.84 for „profitability‟ was much greater than that for „non-profitable‟. The U statistic 15.000 in Table 5.12 B is the number of times members of the lower ranked group „not-profitable‟ precede members of the higher-ranked group „profitable‟ enterprises. The Z value of - 4.245 is the standardized score associated with the significance value (p = .000). The p value being so small, it can be concluded that there is a strong significant difference between competency levels of „profitable‟ and „not-profitable‟ enterprises (McMahon 2001). Hence H1 is validated. 5.8.4 H2 Test H2: Value based organisational core competencies have greater impact than functional or leadership competencies on the profitability of MSMEs. Dependent Variable: Profitability (Question 12) Independent Variable: Core competencies (Question 14(c) H2 has already been supported by positive results of Spearman‟s Correlations and One-Sample Chi-Square NP Test brought out above. Positive results in Friedman One-Way ANOVA and Logistic Regression Analysis tests will lead to validation of the Hypothesis. 5.8.4.1 Friedman One-Way ANOVA Test The Friedman one-way ANOVA is similar to traditional analysis of variance with two notable exceptions: firstly comparisons in the Friedman procedure are based on mean rank of variables rather than on means and standard deviations of raw scores, and secondly, rather than calculating an F ratio, Friedman compares ranked values 163
with expected values in a chi-square analysis (George and Mallory p.218). Although the power of the Friedman operation is less than that of normal analysis of variance, but since the sample data deviated far from normality, the Friedman one-way ANOVA had to be used (Anderson and Sohal, 1999)232. Higher ranked values indicate better relationship between variables. Table 5.13 shows the Friedman oneway ANOVA rank for Question 14, and Table 5.14 displays the Friedman one-way ANOVA chi-square value. Table 5.13: Friedman One-Way ANOVA Rank (H2 - Question 14) S. No. Question No.
Variable
Mean Rank
1.
14 (a)
Functional competencies
1.77
2.
14 (b)
Leadership competencies
2.02
3.
14 (c)
Core competencies
2.21
Table 5.14: Friedman One-Way ANOVA Test Statistics (H2 - Question 14) N
100
Chi-Square
15.101
df
2
Asymp. Sig.
.001
Table 5.13 displays the Friedman ANOVA mean ranks of all three independent variables obtained from Question 14. In Table 5.14, the significance value associated with the chi-square analysis (p = 0.001) indicates that there was a significant difference between the variables. Visual inspection of Table 5.13 indicated that Core competencies had the highest rank amongst Question 14 variables, supporting Hypothesis H2.
232
Anderson, M. and Sohal, A.S. (1999). A study of the relationship between quality management practices and performance in small businesses. International Journal of Quality & Reliability Management, 16(9), 859-877.
164
5.8.4.2 Logistic Regression Test of Hypothesis 2 (H2) Logistic regression is an extension of multiple regression in which the dependent variable is not a continuous variable, and could have only two values, as in the case of this study, where Profitability could be either „Yes‟ or „No‟. In logistic regression the value of dependent variable that is being predicted represents a probability, and it varies between 0 and 1. As in multiple regression, it is meant to show the influence of two or more variables as in H2, on a common dependent variable (profitability). This method has been used by a number of researchers (McMahon 2001; Hsueh & Tu 2009233). The predicted variable here is the dependent variable Profitability. The predictor variables in H2 are Functional, Core and Leadership competencies. Output is shown here in Tables 5.15 A and B and 5.16 A to J below. Block 0: Beginning Block Table 5.15 A: Variables in the Equation - No Predictors (H2 - Question 14)
Step 0 Constant
B
S.E.
Wald
df
Sig.
Exp(B)
2.944
.459
41.181
1
.000
19.000
The Wald test shown in Table 5.15 A suggests that there is significant difference between the frequencies of the two categories (Wald: 41.181, p 0.05) suggesting that the predicted probability match the observed probability. Thus it can be said that the model is predicting in the required manner. Table 5.16 G: Variables in the Equation Step 1 (H2 - Question 14) Step 1a Core Competencies Constant
B
S.E.
Wald
df
Sig.
Exp(B)
3.251
1.136
8.195
1
.004
25.823
-9.072
3.751
5.848
1
.016
.000
a. Variable(s) entered on step 1: Core Competencies. Table 5.16 G shows that Core competencies was the only predictor variable selected in the regression equation along with the constant. The weighting value of B = 3.251 for the predictor variable Core competencies indicates positive effect of the predictor variable on the predicted variable. If „core competencies‟ is increased by one unit, the odds of the profitability going up would increase by 3.251 units. The Wald value of 8.195 with significance 0.004 indicates that the difference between the two categories is significant. The Exp(B) value of 25.823 indicates that the odds of increase in profitability with core competencies is 25.823 times higher than that with no core competencies.
168
Table 5.16 H: Correlation Matrix (H2 - Question 14)
Step 1 Constant Core Competencies
Constant
Core Competencies
1.000
-.990
-.990
1.000
Table 5.16 H shows a high level of correlation between the constant and core competencies. Table 5.16 I: Model if Term Removed (H2 - Question 14)
Variable Step 1 Core Competencies
Model Log Likelihood
Change in -2 Log Likelihood
df
Sig. of the Change
-19.852
14.600
1
.000
Table 5.16 J: Variables not in the Equation (H2 - Question 14) Score
df
Sig.
Step 1 Variables Functional Competencies
.152
1
.697
Leadership Competencies
1.088
1
.297
Overall Statistics
1.324
2
.516
Table 5.16 J indicates the balance two variables that were not entered into the equation. Their significance (0.697 and 0.297) indicate that both Functional and Leadership competencies have no significant impact on profitability. Interpretation of Logistic Regression Test for Hypothesis 2 (H2) The logistic regression test indicated that „core competencies‟ was the only variable retained in the equation with the other two having been removed. The weighting values of the constant (-9.072) and that of core competencies (3.251) could be included in the regression equation. The positive weighting value for core competencies validated Hypothesis 2. With this input the logistic regression could be presented as follows: 169
Profitability increase (probability of the event occurring as X) = - 9.072 (constant) + 3.251 (core competencies). Hence H2 is validated. 5.8.5
H3 Test
H3: Quality Consciousness is the most important value-based organisational core competency for yielding higher profitability of MSMEs. Dependent Variable: Profitability (Question 12) Independent Variable: Quality Consciousness (Question 16(a)) H3 has already been supported by positive results of Spearman‟s Correlations and One-Sample Chi-Square NP Test brought out above. Positive results in Friedman One-Way ANOVA and Logistic Regression tests will lead to validation of the Hypothesis. 5.8.5.1 Friedman One-Way ANOVA Test (H3) The Friedman one-way ANOVA is similar to traditional analysis of variance with two notable exceptions: firstly comparisons in the Friedman procedure are based on mean rank of variables rather than on means and standard deviations of raw scores, and secondly, rather than calculating an F ratio, Friedman compares ranked values with expected values in a chi-square analysis (George and Mallory p.218). Although the power of the Friedman operation is less than that of normal analysis of variance, but since the sample data deviated far from normality, the Friedman one-way ANOVA had to be used (Anderson and Sohal, 1999)234. Higher ranked values indicate better relationship between variables. Table 5.17 shows the Friedman oneway ANOVA ranks for H3 variables, and Table 5.18 displays the Friedman one-way ANOVA chi-square values.
234
Anderson, M. and Sohal, A.S. (1999). A study of the relationship between quality management practices and performance in small businesses. International Journal of Quality & Reliability Management, 16(9), 859-877.
170
Table 5.17: Friedman One-Way ANOVA Ranks (H3 - Question 16) S. No. Question No.
Variable
Mean Rank
1.
16 (a)
Quality Consciousness
4.70
2.
16 (b)
Customer focus
4.15
3.
16 (c)
Cost Consciousness
3.10
4.
16 (d)
Healthy work environment and safety norms
2.45
5.
16 (e)
Team spirit
3.86
6.
16 (f)
Creativity / Innovativeness
2.75
Table 5.18: Friedman One-Way ANOVA Test Statistics (H3 - Question 16) N
100
Chi-Square
147.279
Df
5
Asymp. Sig.
.000
Table 5.17 displays the Friedman ANOVA mean ranks of all the independent variables obtained from Question 16. In Table 5.18, the significance value associated with the chi-square analysis (p = 0.000) indicated that there was a significant difference between the variables. Visual inspection of Table 5.17 indicated that Quality consciousness had the highest rank amongst Question 16 variables, supporting Hypothesis 3. 5.8.5.2 Logistic Regression Test of Hypothesis 3 (H3) The predicted variable here is the dependent variable Profitability. The predictor variables in H3 are the six Core competencies of Quality Consciousness, Customer focus, Cost consciousness, Healthy work environment and safety norms, Team spirit and Creativity / Innovativeness. Output is shown here in Tables 5.19 A, B and Tables 5.20 A to J below.
171
Block 0: Beginning Block Table 5.19 A: Variables in the Equation – No predictors (H3 - Question 16)
Step 0 Constant
B
S.E.
Wald
df
Sig.
Exp(B)
2.944
.459
41.181
1
.000
19.000
The Wald test shown in Table 5.19 A suggests that there is significant difference between the frequencies of the two categories (Wald: 41.181, p 0.05) suggesting that the predicted probability match the observed probability. Thus it can be said that the model is predicting in the required manner. Table 5.20 G: Variables in the Equation with Predictors (H3 - Question 16) B
S.E.
Wald
df
Sig.
Exp(B)
1.549
.548
7.983
1
.005
4.707
Constant
-3.335
2.055
2.632
1
.105
.036
Quality Consciousness
1.560
.617
6.394
1
.011
4.758
Innovativeness
1.069
.566
3.568
1
.059
2.911
Constant
-6.659
3.023
4.853
1
.028
.001
Step 1a Quality Consciousness b
Step 2
a. Variable(s) entered on step 1: Quality Consciousness (QualComp). b. Variable(s) entered on step 2: Innovativeness (InnovComp). 176
Table 5.20 G shows that Quality consciousness and Creativity / Innovativeness were the two predictor variables after Step 2 selected in the regression equation along with the constant. The weighting value of B = 1.560 for the predictor variable Quality consciousness indicates positive effect of the predictor variable on the predicted variable. If Quality consciousness is increased by one unit, the odds of the profitability going up would increase by 1.560 units. The Wald value of 6.394 with significance 0.011 (< 0.05) indicates that the difference between the two categories is significant. The Exp(B) value of 4.758 indicates that the odds of increase in profitability with Quality consciousness is 4.758 times higher than that with no Quality consciousness. The weighting value of B = 1.069 for the predictor variable Creativity / Innovativeness indicates positive effect of the predictor variable on the predicted variable. If Creativity / Innovativeness is increased by one unit, the odds of the profitability going up would increase by 1.069 units. The Wald value of 3.568 with significance 0.059 (> 0.05) indicates that the difference between the two categories is not very significant. The Exp(B) value of 2.911 indicates that the odds of increase in profitability with Creativity / Innovativeness is 2.911 times higher than that with no Creativity / Innovativeness. Table 5.20 H: Correlation Matrix with Predictors (H3 - Question 16) Quality Constant conscious- Constant ness Step 1 Constant Quality consciousness
1.000
-.970
-.970
1.000
Step 2 Constant
Quality Innovativeconsciousness ness
1.000
-.845
-.650
Quality consciousness
-.845
1.000
.181
Innovativeness
-.650
.181
1.000
177
Low values of correlation between Quality consciousness and Innovativeness in Table 5.20 H indicate that the regression does not display multi-collinearity and is therefore stable. Table 5.20 I: Model if Term Removed (H3 - Question 16) Model Log Likelihood
Change in -2 Log Likelihood
df
Sig. of the Change
Step 1 Quality consciousness
-19.852
8.875
1
.003
Step 2 Quality consciousness
-17.241
7.626
1
.006
-15.414
3.971
1
.046
Variable
Innovativeness
Table 5.20 J: Variables not in the Equation (H3 - Question 16) Score
df
Sig.
.865
1
.352
Cost consciousness
.093
1
.760
Healthy work environment and Safety norms
.068
1
.794
Team spirit
.056
1
.812
Innovativeness
3.988
1
.046
Overall Statistics
6.441
5
.266
2.035
1
.154
Cost consciousness
.574
1
.449
Healthy work environment and Safety norms
.014
1
.906
Team spirit
.068
1
.794
Overall Statistics
2.235
4
.693
Step 1 Variables Customer focus
Step 2 Variables Customer focus
Table 5.20 J indicates the balance four variables that were not entered into the equation. Their significance (0.154, 0.449, 0.906 and 0.693) indicate that they do not have significant impact on profitability.
178
Interpretation of Logistic Regression Test for Hypothesis 3 (H3) The logistic regression test indicated that Quality consciousness and Creativity / Innovativeness were two variables retained in the equation with the other four having been removed. The weighting values of the constant (-6.659), that of Quality consciousness (1.560) and Creativity / Innovativeness (1.069) could be included in the regression equation. The larger positive weighting value for Quality Consciousness validated Hypothesis 3. With this input the logistic regression could be presented as follows: Profitability increase (probability of the event occurring as X) = - 6.659 (constant) + 1.560 (Quality Consciousness) + 1.069 (Creativity / Innovativeness). Hence H3 is validated .
5.8.6
H4 Test
H4: Planning and Organising ability is the most important leadership competency for yielding higher profitability of MSMEs. Dependent Variable: Profitability (Question 12) Independent Variable: Planning and Organising ability (Question 17(c)) H4 has already been supported by positive results of Spearman‟s Correlations and One-Sample Chi-Square NP Test brought out above. Positive results in Friedman One-Way ANOVA and Logistic Regression tests will lead to validation of the Hypothesis. 5.8.6.1 Friedman One-Way ANOVA Test (H4 – Question 17) The Friedman one-way ANOVA is similar to traditional analysis of variance with two notable exceptions: firstly comparisons in the Friedman procedure are based on mean rank of variables rather than on means and standard deviations of raw scores, and secondly, rather than calculating an F ratio, Friedman compares ranked values with expected values in a chi-square analysis (George and Mallory p.218). Although the power of the Friedman operation is less than that of normal analysis of variance, but since the sample data deviated far from normality, the Friedman one-way ANOVA had to be used (Anderson and Sohal, 1999)235. Higher ranked values 235
Anderson, M. and Sohal, A.S. (1999). A study of the relationship between quality management practices and performance in small businesses. International Journal of Quality & Reliability Management, 16(9), 859-877.
179
indicate better relationship between variables. Table 5.21 shows the Friedman oneway ANOVA ranks for H4 variables, and Table 5.22 displays the Friedman one-way ANOVA chi-square values.
Table 5.21: Friedman One-Way ANOVA Ranks (H4 - Question 17) S. No. Question No.
Variable
Mean Rank
1.
17 (a)
Strategic thinking
2.70
2.
17 (b)
Interpersonal skills
2.96
3.
17 (c)
Planning & Organising ability
3.28
4.
17 (d)
Decision-making ability
3.05
5.
17 (e)
Problem-solving ability
3.02
Table 5.22: Friedman One-Way ANOVA Test Statistics (H4 - Question 17) N
100
Chi-Square
11.095
Df
4
Asymp. Sig.
.026
Table 5.21 displays the Friedman ANOVA mean ranks of all the independent variables obtained from Question 17. In Table 5.22, the significance value associated with the chi-square analysis (p = 0.026) indicated that there was a significant difference between the variables. Visual inspection of Table 5.21 indicated that Planning and Organising ability had the highest rank amongst Question 17 variables, supporting H4. 5.8.6.2 Logistic Regression Test of Hypothesis 4 (H4) The predicted variable here is the dependent variable Profitability. The predictor variables in H4 are the five Leadership competencies of Strategic thinking, Interpersonal skills, Planning and Organising ability, Decision-making and
180
Problem-solving ability. Output is shown here in Tables 5.23 and Tables 5.24 A to J below. Block 0: Beginning Block Table 5.23 : Variables in the Equation – No predictors (H4 - Question 17)
Step 0 Constant
B
S.E.
Wald
df
Sig.
Exp(B)
2.944
.459
41.181
1
.000
19.000
The Wald test shown in Table 5.23 suggests that there is significant difference between the frequencies to two categories (Wald: 41.181, p 0.05) suggesting that the predicted probability match the observed probability. Thus it can be said that the model is predicting in the required manner. Table 5.24 G: Variables in the Equation with Predictor (H4 - Question 17) Step 1a Planning & Organising Constant
B
S.E.
Wald
df
Sig.
Exp(B)
1.409
.642
4.820
1
.028
4.090
-2.682
2.424
1.224
1
.269
.068
a. Variable(s) entered on step 1: Planning & Organising. Table 5.24 G shows that Planning & Organising is the only predictor variable after Step 1 selected in the regression equation along with the constant. The weighting value of B = 1.409 for the predictor variable Planning & Organising indicates positive effect of the predictor variable on the predicted variable. If Planning & Organising is increased by one unit, the odds of the profitability going up would increase by 1.409 units. The Wald value of 4.820 with significance 0.028 183
(p < 0.05) indicates that the difference between the two categories is significant. The Exp(B) value of 4.090 indicates that the odds of increase in profitability with Planning & Organising is 4.090 times higher than that with no Planning & Organising. Table 5.24 H: Correlation Matrix with Predictor (H4 - Question 17)
Step 1 Constant
Constant
Planning & Organising
1.000
-.980
-.980
1.000
Planning & Organising
Table 5.24 I: Model if Term Removed (H4 - Question 17)
Variable Step 1 Planning & Organising
Model Log Likelihood
Change in -2 Log le Likelihood
df
Sig. of the Change
-19.852
5.046
1
.025
Table 5.24 J: Variables not in the Equation (H4 - Question 17) Score
df
Sig.
.938
1
.333
Interpersonal skills
.027
1
.868
Decision making
1.379
1
.240
Problem solving
1.454
1
.228
Overall Statistics
2.640
4
.620
Step 1 Variables Strategic thinking
Table 5.24 J indicates the balance four variables that were not entered into the equation. Their significance (.333, .868, .240 and .228) indicate that they have insignificant impact on profitability. Interpretation of Logistic Regression Test for Hypothesis 4 (H4) The logistic regression test indicated that Planning & Organising was the only variable retained in the equation with the other four having been removed. The weighting values of the constant (-2,682), and that of Planning & Organising (1.409) could be included in the regression equation. The positive weighting value for only Planning & Organising validated Hypothesis 4. With this input the logistic regression 184
could be presented as follows: Profitability increase (probability of the event occurring as X) = - 2.682 (constant) + 1.409 (Planning and Organising capability).
Hence H4 is validated. 5.8.7
H5 Test
H1: Training and development of employees in skills and competencies yield higher profitability of MSMEs. Dependent Variable: Profitability (Question 12) Independent Variable: Training and development of employees in skills and competencies. H5 has already been supported by positive results of Spearman‟s Correlations and One-Sample Chi-Square NP Test brought out above. Positive result in Mann-Whitney Rank-Sum U Test will lead to acceptance of H5. 5.8.7.1 Mann-Whitney Rank-Sum U Test for H5 This non-parametric equivalent of the t-test for two independent samples has been used by a number of researchers McMahon (2001), Dyba et al. (2005)236 for hypothesis testing. Tables 5.25 A and 5.25 B display the test values for H5. Table 5.25 A: Ranks for Questions 12 and 18 (H5 – Questions 12 & 18)
Training
Profitability
N
Mean Rank
Sum of Ranks
1
5
10.80
54.00
2
95
52.59
4996.00
Total
100
236
Dyba,T. Kampenes, V.B. and Sjoberg, D.I.K. (2005). A systematic review of statistical power in software engineering experiments. Information and Software Technology, 48(2006), 745-755. Retrieved on 15 Sep 2014 from .
185
Table 5.25 B: Test Statisticsa (H5 – Questions 12 & 18) Competency level Mann-Whitney U
15.000
Wilcoxon W
30.000
Z
-4.245
Sig. (2-tailed)
.000
a. Grouping Variable: Profitability Table 5.25 A shows that the mean value 52.84 for „profitability‟ was much greater than that for „non-profitable‟. The U statistic 15.000 in Table 5.25 B is the number of times members of the lower ranked group „not-profitable‟ precede members of the higher-ranked group „profitable‟ enterprises. The Z value of - 4.245 is the standardized score associated with the significance value (p = 0.000). The p value being so small, it can be concluded that there is a strong significant difference between competency levels of „profitable‟ and „not-profitable‟ enterprises (McMahon 2001). Hence H5 is validated. 5.8.8
Additional Logistic Regression tests It was decided to conduct two more
Logistic Regression Tests which are indirectly connected to the study Objective iv and Hypothesis tests: one for the four Functional Competencies to examine which one of them has the highest impact on Profitability, and the other to check whether there is unreasonably high adverse effect of external variables on Profitability when compared to that of Overall Competency Level in H1. 5.8.9
Logistic Regression test for Functional Competencies (Question 15)
The predicted variable here is the dependent variable Profitability. The predictor variables are the four Functional competencies of Technical skills (TechComp), Approach to learning and Self development (LDComp), Adaptability to Technology and change (TechnoComp), and Availability of Specialist Skills (SpecialistComp). Output is shown here in Tables 5.26 A and B, and Tables 5.27 A to J below.
186
Block 0: Beginning Block Table 5.26 A: No Predictor Classification Tablea,b (Question 15) Predicted Profitability 1
2
Percentage Correct
1
0
5
.0
2
0
95
100.0
Observed Step 0 Profitability
Overall Percentage
95.0
a. Constant is included in the model. b. The cut value is .500 Classification Table 5.26 A shows that with no predictor in the model, the probability that profitability will be there would be 95% correct. Table 5.26 B: Variables in the Equation with no Predictor (Functional Competencies) (Question 15)
Step 0 Constant
B
S.E.
Wald
df
Sig.
Exp(B)
2.944
.459
41.181
1
.000
19.000
The Wald test shown in Table 5.26 B suggests that there is significant difference between the frequencies of the two categories (Wald: 41.181, p < 0.01).
187
Block 1: Method = Forward Stepwise (Likelihood Ratio) Table 5.27 A: Step 1 Iteration Historya,b,c,d (Functional competencies) (Question 15) Coefficients -2 Log likelihood
Constant
Adaptability to Technology & Change (TechnoComp)
Step 1 1
43.867
.188
.443
2
30.267
-1.070
1.039
3
25.920
-2.468
1.646
4
24.894
-3.414
2.078
5
24.791
-3.805
2.262
6
24.790
-3.859
2.288
7
24.790
-3.860
2.289
Iteration
a. Method: Forward Stepwise (Likelihood Ratio) b. Constant is included in the model. c. Initial -2 Log Likelihood: 39.703 d. Estimation terminated at iteration number 7 because parameter estimates changed by less than .001. Table 5.27 B: Omnibus Tests of Model Coefficients (Functional competencies) (Question 15) Chi-square
df
Sig.
14.913
1
.000
Block
14.913
1
.000
Model
14.913
1
.000
Step 1 Step
Table 5.27 A shows that the initial -2 LL in Step 0 was 39.703. In Step 1 it reduced to 24.790, indicating that the model had improved its prediction capability. The improvement in -2LL by 14.913 is reflected by the final chi-square value with 0.000 significance in Table 5.27 B.
188
Table 5.27 C: Model Summary with Predictor (Functional competencies) (Question 15) Step
-2 Log likelihood
1
24.790
Cox & Snell R Square
Nagelkerke R Square
.139
.423
a
a. Estimation terminated at iteration number 7 because parameter estimates changed by less than .001. The value of Nagelkerke R Square is what is considered the most appropriate. Its value for Step 1 is 0.423, which indicates that about 42.3% of variance in the dependent variable is explained by the predictor. Table 5.27 D: Hosmer and Lemeshow Test (Functional competencies) (Question 15) Step 1
Chi-square
df
Sig.
.807
2
.668
Table 5.27 E: Contingency Table for Hosmer and Lemeshow Test (Functional competencies) (Question 15) Profitability = 1
Profitability = 2
Observed Expected Observed Expected
Total
Step 1 1
3
3.452
6
5.548
9
2
2
1.273
25
25.727
27
3
0
.270
54
53.730
54
4
0
.005
10
9.995
10
Hosmer and Lemeshow Test shown in Tables 5.27 D and E checks whether the predicted probabilities match the observed probabilities shown in Table 5.27 C above. The chi-square value in Table 5.27 D is significant (x2 (1) = 0.807, p > 0.05) suggesting that the predicted probability match the observed probability. Thus it can be said that the model is predicting in the required manner.
189
Table 5.27 F: Classification Table with Predictor (Functional competencies) (Question 15) Predicted Profitability 1
2
Percentage Correct
1
1
4
20.0
2
0
95
100.0
Observed Step 1 Profitability
Overall Percentage
96.0
Classification Table 5.27 F shows that with predictor in the model, the probability that profitability will be there has improved to 96% correct, from 95% correct without predictor. Table 5.27 G: Variables in the Equation with Predictor (Functional competencies) (Question 15) Step 1a TechnoComp Constant
B
S.E.
Wald
df
Sig.
Exp(B)
2.289
.767
8.897
1
.003
9.861
-3.860
1.978
3.810
1
.051
.021
a. Variable(s) entered on step 1: TechnoComp. Table 5.27 G shows that Adaptability to New Technology & Change (TechnoComp) is the only predictor variable after Step 1 selected from the functional competencies in the regression equation along with the constant. The weighting value of B = 2.289 for the predictor variable Adaptability to Technology & Change (TechnoComp) indicates positive effect of the predictor variable on the predicted variable. If Adaptability to New Technology & Change (TechnoComp) is increased by one unit, the odds of the profitability going up would increase by 2.289 units. The Wald value of 8.897 with significance 0.003 (p < 0.05) indicates that the difference between the two categories is significant. The Exp(B) value of 9.861 indicates that the odds of increase in profitability with Adaptability to New Technology & Change (TechnoComp) is 9.861 times higher than that with no Adaptability to New Technology & Change (TechnoComp). 190
Table 5.27 H: Correlation Matrix with Predictor (Functional competencies) (Question 15) Constant
TechnoComp
1.000
-.962
-.962
1.000
Step 1 Constant TechnoComp
Table 5.27 I: Model if Term Removed with Predictor (Functional competencies) (Question 15) Change in -2 Model Log Log Likelihood Likelihood
Variable Step 1 TechnoComp
-19.852
14.913
df
Sig. of the Change
1
.000
Table 5.27 J: Variables not in the Equation Step 1 (Functional competencies) (Question 15) Score
df
Sig.
.100
1
.751
LDComp
.488
1
.485
SpecialistComp
.357
1
.550
Overall Statistics
2.040
3
.564
Step 1 Variables TechnicalComp
Table 5.27 J indicates the balance four variables that were not entered into the equation. Their significance (0.751, 0.485, 0.550 and 0.564) indicate that they have in-significant impact on profitability. Interpretation of Logistic Regression Test for Functional Competencies The logistic regression test indicated that Adaptability to New Technology & Change (TechnoComp) was the only variable retained in the equation with the other four having been removed. The weighting values of the constant (-3.860), that of Adaptability to New Technology & Change (2.289) could be included in the regression equation. With this input the logistic regression could be presented as follows: 191
Profitability increase (probability of the event occurring as X) = - 3.860 (constant) + 2.289 (Adaptability to Technology & Change).
5.8.10
Logistic Regression test for Overall Competency Level with Extraneous Variables: Size, Type, Attrition and Qualified Managers (H1)
The predicted variable here is the dependent variable Profitability (Question 12). The predictor variables are the Overall Competency level (Question 13), Size of business (Question 7), Type of business (Question 8), Attrition level (Question 11) and Hiring of Qualified Managers (Question 20). Output is shown here in Tables 5.28 A to E below. Block 1: Method = Forward Stepwise (Likelihood Ratio) Table 5.28 A: Model Summary (H1-Questions 7, 8, 11, 12, 13 and 20)
Step 1
-2 Log
Cox & Snell R
likelihood
Square
Nagelkerke R Square
15.158a
.218
.664
a. Estimation terminated at iteration number 20 because maximum iterations has been reached. Final solution cannot be found. The model summary at Table 5.28 A presents three measures of how well the logistic model fits the data: -2 LL, Cox & Snell R Square and Nagelkerke R Square. Out of these three, the value of Nagelkerke R Square is what is considered the most appropriate. Its value for Step 1 is 0.664, which indicates that about 66.4% of variance in the dependent variable is explained by the predictor. Table 5.28 B: Variables in the Equation (H1-Questions 7, 8, 11, 12, 13 and 20)
Step 1a CompProf Constant
B
S.E.
Wald
df
Sig.
Exp(B)
20.308
3438.623
.000
1
.995
6.602E8
-60.742 10315.870
.000
1
.995
.000
a. Variable(s) entered on step 1: CompProf.
192
Table 5.28 C: Corrélation Matrix (H1-Questions 7, 8, 11, 12, 13 and 20) Constant
CompProf
1.000
-1.000
-1.000
1.000
Step 1 Constant CompProf
Table 5.28 D: Model if Term Removed (H1-Questions 7, 8, 11, 12, 13 and 20) Change in -2
Variable Step 1 CompProf
Model Log
Log
Likelihood
Likelihood
df
Sig. of the Change
-19.852
24.545
1
.000
Table 5.28 E: Variables not in the Equation (H1-Questions 7, 8, 11, 12, 13 and 20) Score
df
Sig.
1.320
1
.251
Type
.775
1
.379
Attrition
1.640
1
.200
QualMgr
.449
1
.503
Overall
6.194
4
.185
Step 1 Variables Size
Statistics
Table 5.28 B shows that Overall Competency level (CompProf) is the only predictor variable after Step 1 selected from the different variables in the regression equation along with the constant. The weighting value of B = 20.308 for the predictor variable Overall Competency level (CompProf) indicates positive effect of the predictor variable on the predicted variable. If Overall Competency level (CompProf) is increased by one 193
unit, the odds of the profitability going up would increase by 20.308 units. The high significance values of the balance extraneous variables in Table 5.28 E indicate that they have negligible impact on profitability of the enterprises. 5.8.11 Hypotheses Test’s Confirmation: Causal Path Analysis (H1 to H5) Bento & Bento (2004)237 have recommended use of causal path analysis techniques for testing causality of all research data. Being non-parametric, this research data was tested by assigning partial correlation coefficients and drawing a Causal Path Diagram with the results. Partial Correlation coefficients were obtained for Overall Competency Level (Question 13),
three competency factors (Core,
Leadership and Functional) (Question 14), all 15 individual competencies (Questions 15 to 17), Training provided by MSMEs for competency development (Question 18), and the dependent variable Profitability (Question 12) of the MSMEs. The extraneous variables: 'Organisation Size' (Question 7) and 'Type of Industry' (Question 8) were kept as control variables. The partial correlation coefficient values and causal path diagram are shown in Table 5.29 and Figure 5.30 below.
237
Bento, A. & Bento, R. (2004). The Use of Causal Analysis Techniques in Information Systems Research: A Methodological Note. Journal of Information Technology Management. (XV,3-4). ISSN#1042-1319.
194
Table 5.29: Partial Correlation for Causal Path Analysis with ‘Organisation Size’ and ‘Type of Business’ as Control Variables (Questions 7, 8, and 12 to 18) Variables
Profitability
Profitability (Dependent Variable for all 5 Hypotheses)
Correlation
1.000
Significance
.
Competency level (Independent Variable for H1)
Correlation
.444
Significance
.000
Functional competency factor
Correlation
.179
Significance
.078
Correlation
.179
Significance
.078
Core Competency Factor (Independent Variable for H2)
Correlation
.406
Significance
.000
Technical skill level
Correlation
.279
Significance
.005
Approach towards Learning & Self development
Correlation
.284
Significance
.005
Adaptability to new technology & change
Correlation
.425
Significance
.000
Correlation
.313
Significance
.002
Quality Consciousness (Independent Variable for H3)
Correlation
.359
Significance
.000
Customer focus
Correlation
.098
Significance
.336
Correlation
.059
Significance
.567
Correlation
.123
Significance
.227
Correlation
.190
Significance
.061
Leadership competency factor
Availability of Specialised skills
Cost consciousness
Healthy work environment / safety norms
Team spirit
195
Variable
Profitability
Creativity / Innovativeness
Correlation
.214
Significance
.034
Correlation
.183
Significance
.071
Correlation
.164
Significance
.108
Planning & Organising skills (Independent Variable for H4)
Correlation
.250
Significance
.013
Decision-making
Correlation
.207
Significance
.041
Correlation
.198
Significance
.050
Correlation
.353
Significance
.000
Strategic thinking
Interpersonal skills
Problem solving ability
Training given by MSMEs in employees' competency development (Independent Variable for H5)
It is seen from Table 5.29 above that Partial Correlation Coefficients after catering for the two external variables are larger in all cases than the Spearman‟s Coefficients derived earlier.
196
Figure 5.30: Causal Path Diagram with Ordinal Partial Correlations (Questions 7, 8, and 12 to 18) Organisational Size (Control variable) Quality Consciousness (H3)
0.359
Customer Focus Cost Consciousness
0.098 0.059
Safety Norms
0.123
Team Spirit
0.190
Creativity
0.214
Core Competencies (H2)
0.406
0.457
Strategic Thinking
0.183
Interpersonal Skills
0.164
Planning & Org Skills (H4) Decision Making
Problem Solving
Technical Skills
Learning &Devp Adaptability to Tech & Change Specialised Skills
Leadership Competencies
0.179
0.343
0.250 0.207 0.198 0.279
0.284
Profitability
Overall Competency Level (H1)
0.444
Variable)
0.324
Functional Competencies
0.179 0.566
0.425 0.313
Training (H5)
Type of business (Control variable)
197
(Dependent
0.353
All five hypothesis are validated by Causal Path Analysis: i.
H1 – Overall Competency Level
0.444
Profitability
ii.
H2 - Core Competency Factor
0.406
Profitability (More than 0.179
for Functional and Leadership Competency Factors) iii.
H3 - Quality Consciousness
0.359
Profitability (More than 5 other
Core competencies) iv.
H4 – Planning & Organising ability 0.250
Profitability (More than 4 other
Leadership competencies) v.
H5 – Training in Competency Development
vi.
Adaptability to new technology & change 0.425
0.353
Profitability
Profitability (more than all
other functional competencies) The results showed that Partial Correlations for all the Hypothecated variables had emerged with the highest values amongst their respective competency factor groups and hence supported their Hypotheses. Over and above the five hypothecated independent variables, an additional relationship between a non-hypothecated variable 'Adaptability to New Technology and Change' also came up here as well as in the Logistic Regression Analysis with the highest partial correlation value after 'overall competency level‟, as seen in Logistic Regression Test in sub-paragraph 5.8.9. Hence H1 to H5 are confirmed and strongly validated. 5.9
Factor Analysis Factor Analysis is most frequently used to represent relationships among sets
of interrelated variables (George and Mallery 2011, p.246). Four basic steps are required to conduct a factor analysis: i.
Calculate a correlation matrix
ii.
Extract factors
iii.
Rotate factors to create a more understandable factor structure
iv.
Interpret results. The net effect of interactions between the different variables is analysed with
the help of factor analysis. Factor analysis was used for dimension reduction and to determine the factor structures under each of the employee competency factor groupings viz. Core competencies, Leadership competencies and Functional competencies. These factor structures obtained would provide assistance in focusing 198
on the critical competencies for driving financial performance of MSMEs. 5.9.1 Factor Analysis of Core Competencies (Question 16) The factor analysis output for core competencies viz. between the six competencies of Quality consciousness, Customer focus (CustComp), Cost consciousness (CostComp), Healthy work environment and Safety norms (SafeComp), Team spirit (TeamComp) and Creativity / Innovativeness (InnovComp). is shown in Tables 5.30 A to F below.
Table 5.30 A: Correlation Matrix for Core Competencies (Question 16) Quality Cust Conscious ness Comp Correlation
Cost
Safe
Team
Innov
Comp
Comp
Comp
Comp
Quality Consciousness
1.000
.404
.151
.295
.445
.136
CustComp
.404
1.000
.251
.114
.224
.240
CostComp
.151
.251
1.000
.411
.158
.291
SafeComp
.295
.114
.411
1.000
.418
.273
TeamComp
.445
.224
.158
.418
1.000
.222
InnovComp
.136
.240
.291
.273
.222
1.000
Calculating a correlation matrix of all variables of interest is the starting point for factor analysis, and provides some initial clues to the whole process. Table 5.30 A shows that Quality consciousness enjoys high correlation with Customer focus and Team spirit.
199
Table 5.30 B: KMO and Bartlett's Test (Core competencies) (Question 16) Kaiser-Meyer-Olkin Measure of Sampling Adequacy. Bartlett's Test of Sphericity
Approx. Chi-Square
.665
98.193
df
15
Sig.
.000
KMO and Bartlett‟s Test of Sphericity are both tests of sampling adequacy (adequacy of the study variables for conducting factor analysis) and multivariate normality. Kaiser-Meyer-Olkin (KMO) Measure of Sampling Adequacy value of 0.665 shows a mediocre sampling adequacy. Bartlett‟s Test of Sphericity with a significance value < 0.05 indicates that these data do not produce an identity matrix (or differ significantly from identity), and are thus approximately multivariate normal and acceptable for factor analysis. Table 5.30 C: Communalities (Core competencies) (Question 16) Variable
Initial Communality
Quality Consciousness
1.000
CustComp
1.000
CostComp
1.000
SafeComp
1.000
TeamComp
1.000
InnovComp
1.000 Extraction Method: Principal Component Analysis.
Communalities are designed to show the proportion of variance that the factors contribute to explaining a particular variable. These values range from 0 to 1 and may be interpreted in a manner similar to a Multiple R in Multiple Regression Analysis, 200
with 0 indicating that common factors explain none of the variance in a particular variable, and 1 indicating that all the variance in that variable is explained by the common factors. Each of the six variables is initially assigned a Communality value of 1.0 by default. The method of factor extraction includes seven options. The Principals Components method used here is the one most frequently used. Table 5.30 D: Total Variance Explained (Core competencies) (Question 16) Initial Eigenvalues Component
Rotation Sums of Squared Loadings
Total
% of Variance
Total
% of Variance
Cumulative %
Cumulative %
1
2.357
39.276
39.276
1.759
29.320
29.320
2
1.058
17.627
56.904
1.655
27.584
56.904
3
.926
15.432
72.336
4
.741
12.352
84.688
5
.480
7.996
92.684
6
.439
7.316
100.000
Extraction Method: Principal Component Analysis. Component is the number of each factor extracted. Eigenvalues indicate the proportion of variance explained by each factor. In Table 5.30 D, the first two components display eigenvalues greater than 1, and account for almost 57 % of the total variance. Variance is derived by dividing the eigenvalue by the sum of the communalities (6 in this case).
201
Figure 5.31: Scree Plot (Factor Analysis: Core competencies - Question 16)
Figure 5.31 shows the scree plot for factor analysis of core competencies. It plots the eigenvalues on a bi-coordinate plane, and derives its name from the scree that is deposited at the base of a landslide. It is helpful in selecting how many factors to rotate to a final solution. The traditional construct for interpretation is that the scree slope should be ignored and that only factors on the steep portion of the graph should be selected and rotated. The scree plot in Figure 5.31 indicates 2 factors for rotation.
202
Table 5.30 E: Rotated Component Matrixa
(Core competencies - Question 16) Variable
Component 1
2
Quality Consciousness
.866
.042
Team spirit (TeamComp)
.698
.251
Customer focus (CustComp)
.640
.153
Cost consciousness (CostComp)
.058
.802
Healthy work environment & Safety norms (SafeComp)
.306
.687
Creativity / Innovativeness (InnovComp)
.121
.673
Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. a. Rotation converged in 3 iterations. Table 5.30 E shows the rotated factor structure with factor loadings for „core competencies‟. Rotation is needed because the original factor structure is mathematically correct but is difficult to interpret. The goal of rotation is to achieve a simple structure with high factor loadings on one factor and low loadings on all others. Factor loadings vary between +1.0 and -1.0, and indicate the strength of relationship between a particular variable and a particular factor, in a way similar to a correlation. Varimax is the default procedure used by SPSS for rotation of axes to the best simple structure. Varimax rotations are called orthogonal rotations because the axes that are rotated remain at right angles to each other. In Table 5.30 E Factor 1 consists of the variables Quality consciousness, Team spirit and Customer focus with factor loadings of 0.866, 0.698 and 0.640 respectively. The second Factor 2 shows high factor loadings for the balance three competencies. However, the researcher could not arrive at any underlying construct to differentiate these two Factors 1 and 2. 203
Table 5.30 F: Component Transformation Matrix (Core competencies) (Question 16) Component
1
2
1
.735
.678
2
-.678
.735
Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. Table 5.30 F shows the final component transformation matrix for core competencies showing a communality of 0.735 for Component 1 and 0.678 for Component 2.
Figure 5.32: Component Plot in Rotated Space (Core competencies - Question 16)
204
Figure 5.32 graphically shows the two components with their variables clustered together around two axes as estimated from the Tables above. 5.9.2 Factor Analysis of Functional Competencies (Question 15) The factor analysis output for functional competencies viz. between the four competencies of Technical skills (TechnicalComp), Approach to learning and self-development (LDComp), Adaptability to New Technology and Change (TechnoComp) and Availability of Specialised skills (SpecialistComp) is shown in Tables 5.31 A to E below.
Table 5.31 A : Correlation Matrix (Functional competencies) (Question 15) Technical Comp
LD Comp
Techno Comp
Specialist Comp
1.000
.778
.647
.245
LDComp
.778
1.000
.708
.260
TechnoComp
.647
.708
1.000
.416
SpecialistComp
.245
.260
.416
1.000
Correlation TechnicalComp
Table 5.31 A shows that all three besides Specialised skills enjoy high correlation with each other.
Table 5.31 B : KMO and Bartlett's Test (Functional competencies) (Question 15) Kaiser-Meyer-Olkin Measure of Sampling Adequacy. Bartlett's Test of Sphericity
Approx. Chi-Square df
.729 180.689 6
Sig.
.000
Kaiser-Meyer-Olkin (KMO) Measure of Sampling Adequacy value of 0.729 shows a middling sampling adequacy. Bartlett‟s Test of Sphericity with a significance value < 0.05 indicates that these data do not produce an identity matrix (or differ significantly from identity), and are thus approximately multivariate normal and acceptable for factor analysis. 205
Table 5.31 C : Communalities (Functional competencies) (Question 15) Initial TechnicalComp
1.000
LDComp
1.000
TechnoComp
1.000
SpecialistComp
1.000
Extraction Method: Principal Component Analysis. Table 5.31 D : Total Variance Explained (Functional competencies) (Question 15) Initial Eigenvalues Component
Total
% of Variance
Cumulative %
1
2.599
64.982
64.982
2
.863
21.574
86.556
3
.326
8.160
94.716
4
.211
5.284
100.000
Extraction Method: Principal Component Analysis. In Table 5.31 D, only one component displays eigenvalue of 2.599 greater than 1, and accounts for almost 65% of the total variance. Table 5.31 E : Component Matrixa (Functional competencies) (Question 15) Component 1 LDComp
.897
TechnoComp
.882
TechnicalComp
.871
SpecialistComp
.509
Extraction Method: Principal Component Analysis. a. 1 component extracted.
206
Factor analysis of Functional competencies indicates that the selected four competencies are correlated favourably with each other as seen in Table 5.31 E above, and further dimension reduction is not possible. 5.9.3 Factor Analysis of Leadership competencies (Question 17)
The factor
analysis output for leadership competencies viz. between the five competencies of Strategic thinking (StratComp), Interpersonal competencies (InterpersonalComp), Planning and Organising , Decision-making (DMComp),
Problem solving
(ProbSolComp) is shown in Tables 5.32 A to E below.
Table 5.32 A : Correlation Matrix (Leadership competencies) (Question 17) Strat Comp Correlation StratComp
Interpersonal Planning & comp Organising
DM Comp
ProbSol Comp
1.000
.390
.339
.427
.249
Interpersonal comp
.390
1.000
.383
.305
.442
Planning & Organising
.339
.383
1.000
.325
.388
DMComp
.427
.305
.325
1.000
.434
ProbSolComp
.249
.442
.388
.434
1.000
Table 5.32 A shows that all five leadership competencies are satisfactorily correlated with each other. Table 5.32 B : KMO and Bartlett's Test (Leadership competencies) (Question 17) Kaiser-Meyer-Olkin Measure of Sampling Adequacy. Bartlett's Test of Sphericity
Approx. Chi-Square df
.746 99.104 10
Sig.
.000
Kaiser-Meyer-Olkin (KMO) Measure of Sampling Adequacy value of 0.746 shows a middling sampling adequacy. Bartlett‟s Test of Sphericity with a significance value < 0.05 indicates that these data do not produce an identity matrix (or differ
207
significantly from identity), and are thus approximately multivariate normal and acceptable for factor analysis. Table 5.32 C : Communalities (Leadership competencies) (Question 17) Initial StratComp
1.000
Interpersonal comp
1.000
Planning & Organising
1.000
DMComp
1.000
ProbSolComp
1.000
Extraction Method: Principal Component Analysis. Table 5.32 D : Total Variance Explained (Leadership competencies) (Question 17) Initial Eigenvalues Component
Total
% of Variance
Cumulative %
1
2.474
49.474
49.474
2
.770
15.405
64.879
3
.701
14.026
78.905
4
.628
12.562
91.467
5
.427
8.533
100.000
Extraction Method: Principal Component Analysis. In Table 5.32 D, only one component displays eigenvalue greater than 1, and accounts for almost 50% of the total variance.
208
Table 5.32 E : Component Matrixa (Leadership competencies) (Question 17) Component 1 Interpersonal comp
.721
ProbSolComp
.721
DMComp
.709
Planning & Organising
.689
StratComp
.676
Extraction Method: Principal Component Analysis. a. 1 component extracted. Factor analysis of Leadership competencies as seen in Table 5.32 E indicates that the selected five competencies are loaded on one component only and further dimension reduction is not possible.
209
