The methodology adopted here is in line with the suggestions provided by Sarstedt, M et al., (2016) to select an appropriate version of structural equation modeling. We have adopted Common factor Structural Equation Modeling (CBSEM) as per the context and the definitions of the construct (Figure − 2).
4.1. Case Setting:
The Swachh Bharat Mission of the Government of India (GOI) has successfully improved the sanitization condition in India within a short period. The key approach adopted by the government was Community-Led Total Sanitation (CLTS) also termed as Community Approaches to Total Sanitation (CATS). This research study was undertaken in the Dantewada District of Indian State Chattisgarh and Cooch Behar and Malda Districts of Indian State West Bengal.
Dantewada, also known as South Bastar District, is a small district located on the southern tip of Chhattisgarh and shares its boundaries with other Indian states, Maharashtra, Telangana, and Odisha. The district is home to nearly 0.25 million people, out of which more than 75 percent belong to scheduled tribes, living in and off the forests that they inhabit. Gond is the major tribe with nearly 60 percent population and the remaining ones include Halba, Muriya, and Maria. The district comprises 4 development blocks which cover 124 Gram Panchayats (GPs), the primary units of elected local self-government in the rural areas.
At the time of the launch of Swachh Bharat Mission Gramin (SBM-G) in October 2014, the sanitation coverage of the district was less than 10 percent, with all 124 GPs remaining to be declared as ODF. To make the entire country Open Defecation Free (ODF) within the next five years was not an easy task. The Community-Led Total Sanitation (CLTS) approach was the key to this success. The impact and results of (CLTS) had been showing across the world and it was considered to be the best option to make this happen. The power of collective community action and the quick and effective results of CLTS had proven that things change at a rapid pace (KAR, 2008). Being a participatory and empowering approach, it engages communities in a manner that stimulates self-analysis and collective local action to put an end to open defecation (Kumar and Shukla, 2011). This CLTS intervention supported by UNICEF sought to combine CLTS tools and techniques with other social mobilization methods ensuring multi-stakeholder engagement at various levels to achieve credible Open Defecation Free (ODF) results with speed and on the scale. The key components of the strategy adopted included the following: capacity building; institutional strengthening; and ODF celebrations. A team of 30 trained CLTS facilitators was constituted to help implement SBM-G in the district. It worked well as within a year the coverage reached 40 percent and within 30 months the entire district of Dantewada was declared ODF.
Within two months Jhodiabadam, a GP with 212 households (HH) became the first GP to be declared ODF. This GP of Gidam Development Block showed extremely positive results of CLTS implementation. The effective community triggering sessions and continuous follow-ups added to this success. The key player behind this achievement was Balram Kashyap who came up as the natural leader and paved the way for this achievement. A grand event was organized by the community wherein the district officials, public representatives, and people from other GPs were invited. It became an inspirational event for other GPs to gear up and work together to obtain ODF Status for their districts Malda is located in the northern part of West Bengal and is one of the major districts of the state. It shares an international boundary with Bangladesh on the eastern part and has a population of about 4 million which includes 53% Hindus and 46% Muslims. 1% of people belong to other religions and tribal communities. There are 9 development blocks and 146-gram panchayats in the district. Before the implementation of CLTS, sanitation coverage in the district was a meager 40%. To commence with, a habitation in the Manikchak block was selected as a pilot initiative and after its success, the program was successfully scaled up in the entire district.
Bhutni Chandipur High school is a Sansad of Dakshin Chandipur GP of Manikchak development block in Malda district of West Bengal. It is an island surrounded by the river Ganges (Ganga). Commutation becomes a major challenge for the people residing in this area as they have to cross the river through boats, the bad condition of the road adds further to misery as it becomes nearly impossible for residents to use it during the rainy season. Given the difficult topography, the place is also prone to heavy floods. The main source of income in the region is agriculture. Men usually work in major industrial cities like Kolkata, Delhi NCR, Mumbai, etc. to earn a living leaving their family members behind to face drudgery.
Bhutni comprises six villages/ pada with a population of around 3000. The total number of households in the sansad is about 415 as described below in Table-3.
Table-3: Details of House-holds in Village/Pada
Sr.
|
Village/Pada
|
No. of House-holds
|
1.
|
Bhonusarkar Tola
|
54
|
2.
|
Chabimanjhi Tola
|
125
|
3.
|
Nabadeep Tola
|
34
|
4.
|
Biru Tola
|
52
|
5.
|
Sonatan Tola – 1
|
95
|
6.
|
Sonatan Tola – 2
|
55
|
|
Total
|
415
|
Initially, CLTS training for CFs, SHG members, ICDS workers, CSPs, and other volunteers was conducted by Knowledge Links (NGO). Thereafter, an action plan for carrying out the triggering exercise was made and implemented aggressively in the sansad. Within three weeks Bhonusarkar Tola became the first ODF village. However, the momentum slowed down a bit after this achievement. Many residents expressed their inability to construct toilets citing poverty reasons and looked forward to government support in this regard. A base-line survey was done to identify the house-holds that required financial aid to construct toilets. It was found after the survey that 7 house-holds required financial assistance, which was extended. The financial position of the remaining households was found to be satisfactory. They were repeatedly triggered and convinced to construct toilets on their own. The real challenge was to get toilets constructed in house-holds that really could not afford it. To address this issue, all the SHG members, ICDS workers, CSPs, and teachers came forward and contributed money to get toilets constructed in these house-holds.
All the residents enthusiastically participated in the CLTS implementation program and cooperated with the work teams. Despite commutation challenges owing to uneven roads and harsh weather conditions, they were successful in making the Sansad ODF.
Youth clubs also extended their support in an ample measure to achieve this objective. They dug pits, assembled the rings, arranged bamboos, and performed all other tasks that were required to support the cause. The exciting part was that all the political leaders and workers, irrespective of the political parties to which they belonged and political differences that they may have had, joined hands and came together for this purpose.
Nazardari teams regularly checked OD during the early morning hours and late evenings. Gandhigiri was practiced throughout to motivate and make the people understand the importance of ODF and curb it.
In Bhonusarkar Tola, residents also wanted to construct a community toilet for which each one of them contributed and collected around Rs. 22,000/- (twenty-two thousand rupees). They were unable to find space at a central location in the village where the community toilet was proposed. Bimla Mandol, a widow of around 70 years, came forward and donated a part of her residence for this purpose. She was recognized as a Natural Leader along with Sukhendu Mandal, Sameer Mandal, and Rakhi Deb.
The hard work and persistent efforts continued until the entire Sansad became ODF. Thereafter, a grand celebration was organized in which the District Magistrate, Administrative Officers, and Local Leaders participated and appreciated the efforts that were put-in by the residents to make the Sansad ODF. Sikha Mandol, the Pradhan of GP, congratulated the Sansad and requested other sansads to take inspiration and become ODF. This was the first-ever ODF celebration in the state of West Bengal. The District Magistrate assured the availability of safe drinking water in the Sansad for improving the quality of life of residents and decreasing the spread of water-borne diseases.
The impact that the CFs, SHG members, ICDS workers, CSPs, etc. had on the residents was tremendous. They played a vital role in making the residents understand the importance of ODF and counseled them to stop practicing open defecation and use safe toilets. Their familiarity with local culture and language, made it is easier for them to communicate and associate with the people. This played a key role in the proper implementation of CLTS and helped in making the region ODF.
4.5. Structure Equation Modelling
All the constructs that are measured were first-order latent constructs. The measured value against the respective items is shown in Table-3. The output of the model is shown in Figure-2. SEM was conducted using Lavaan package. The output model generated is presented in Figure-3.
4.5.1. Measurement Models
The latent constructs of the model (PU, PEU, FC, AT, SN, CI, HY, and RRD) were measured using SEM using the pre-existing scale (15 percent of questions were changed). The scale for the measurement of RRD was developed by the authors, prior to using the scales a pilot study was conducted to observe its reliability. The authors tested multiple models for a better fit. The measurement loadings of the few constructs (highlighted in yellow colour) in Model − 1 were below or above the required cut off and were hence deleted in Model – 2. It is evident that Model − 2 exhibits a good model fit and better reliability (Table − 5). It is a good fit model for further analysis.
Table-5: Measurement Models
Latent Variables:
|
Model1
|
|
|
|
Estimate
|
Std.Err
|
z-value
|
P(>|z|)
|
PU =~
|
|
|
|
|
PU1
|
1
|
|
|
|
PU2
|
0.688
|
0.072
|
9.505
|
0.000
|
PU3
|
-0.325
|
0.063
|
-5.16
|
0.000
|
PU4
|
-0.796
|
0.079
|
-10.038
|
0.000
|
PU5
|
-0.555
|
0.075
|
-7.429
|
0.000
|
PU6
|
-0.131
|
0.066
|
-1.986
|
0.047
|
PEU =~
|
|
|
|
|
PEU1
|
1
|
|
|
|
PEU2
|
1.146
|
0.29
|
3.958
|
0.000
|
FC =~
|
|
|
|
|
FC1
|
1
|
|
|
|
FC2
|
1.169
|
0.07
|
16.797
|
0.000
|
FC3
|
1.575
|
0.083
|
19.071
|
0.000
|
FC4
|
1.278
|
0.074
|
17.22
|
0.000
|
AT =~
|
|
|
|
|
AT1
|
1
|
|
|
|
AT2
|
1.235
|
0.086
|
14.431
|
0.000
|
AT3
|
1.495
|
0.095
|
15.759
|
0.000
|
AT4
|
1.559
|
0.1
|
15.607
|
0.000
|
SN =~
|
|
|
|
|
SN1
|
1
|
|
|
|
SN2
|
0.829
|
0.095
|
8.723
|
0.000
|
SN3
|
0.984
|
0.116
|
8.498
|
0.000
|
SN4
|
2.121
|
0.359
|
5.907
|
0.000
|
SN5
|
1.586
|
0.165
|
9.597
|
0.000
|
SN6
|
1.374
|
0.132
|
10.409
|
0.000
|
SN7
|
0.77
|
0.116
|
6.633
|
0.000
|
SN8
|
1.454
|
0.145
|
10.011
|
0.000
|
CI =~
|
|
|
|
|
CI1
|
1
|
|
|
|
CI2
|
1.989
|
0.393
|
5.059
|
0.000
|
CI3
|
1.788
|
0.202
|
8.845
|
0.000
|
CI4
|
1.867
|
0.211
|
8.857
|
0.000
|
CI5
|
2.474
|
0.264
|
9.358
|
0.000
|
HY =~
|
|
|
|
|
HY1
|
1
|
|
|
|
HY2
|
1.085
|
0.131
|
8.262
|
0.000
|
HY3
|
1.076
|
0.129
|
8.333
|
0.000
|
HY4
|
1.521
|
0.167
|
9.101
|
0.000
|
HY5
|
1.33
|
0.153
|
8.689
|
0.000
|
RRD =~
|
|
|
|
|
RRD1
|
1
|
|
|
|
RRD2
|
2.21
|
1.15
|
1.921
|
0.055
|
RRD3
|
5.204
|
1.709
|
3.044
|
0.002
|
RRD4
|
4.725
|
1.546
|
3.057
|
0.002
|
Latent Variables:
|
Model2
|
|
|
|
Estimate
|
Std.Err
|
z-value
|
P(>|z|)
|
PU =~
|
|
|
|
|
PU1
|
1
|
|
|
|
PU2
|
0.627
|
0.069
|
9.119
|
0.000
|
PU4
|
-0.738
|
0.076
|
-9.768
|
0.000
|
PU5
|
-0.606
|
0.074
|
-8.239
|
0.000
|
PEU =~
|
|
|
|
|
PEU1
|
1
|
|
|
|
PEU2
|
1
|
|
|
|
FC =~
|
|
|
|
|
FC1
|
1
|
|
|
|
FC2
|
1.0
|
0.04
|
26.705
|
0.000
|
FC3
|
1
|
|
|
|
FC4
|
0.83
|
0.048
|
17.244
|
0.000
|
AT =~
|
|
|
|
|
AT1
|
1
|
|
|
|
AT2
|
0.942
|
0.046
|
20.483
|
0.000
|
AT3
|
1.0
|
0.046
|
24.812
|
0.000
|
AT4
|
1
|
|
|
|
SN =~
|
|
|
|
|
SN1
|
1
|
|
|
|
SN2
|
0.798
|
0.08
|
9.926
|
0.000
|
SN3
|
0.941
|
0.098
|
9.605
|
0.000
|
SN4
|
1
|
|
|
|
SN5
|
1.0
|
0.132
|
9.609
|
0.000
|
SN6
|
1.0
|
0.104
|
11.034
|
0.000
|
SN7
|
0.737
|
0.1
|
7.357
|
0.000
|
SN8
|
1.0
|
0.117
|
10.663
|
0.000
|
CI =~
|
|
|
|
|
CI1
|
1
|
|
|
|
CI2
|
1.0
|
0.203
|
5.783
|
0.000
|
CI3
|
1.00
|
0.068
|
14.829
|
0.000
|
CI4
|
0.993
|
0.071
|
14.013
|
0.000
|
CI5
|
1
|
|
|
|
HY =~
|
|
|
|
|
HY1
|
1
|
|
|
|
HY2
|
0.862
|
0.079
|
10.918
|
0.000
|
HY3
|
0.878
|
0.077
|
11.336
|
0.000
|
HY4
|
1
|
|
|
|
HY5
|
1.0
|
0.088
|
12.039
|
0.000
|
RRD =~
|
|
|
|
|
RRD1
|
1
|
|
|
|
RRD2
|
1.0
|
0.307
|
4.444
|
0.000
|
RRD3
|
1
|
|
|
|
RRD4
|
1.0
|
0.133
|
8.649
|
0.000
|
Model Validity
The goodness of Model − 2 has rendered adequate model validity (Table − 6), the authors followed the steps to check the common method bias and discriminant validity of the CBSEM as suggested by Hair et al., 2006. Harman’s single factor test was run to examine the common method bias. It revealed that the maximum variance explained by the considered study variables was 29.11%. This proves that common method bias is not an issue since the retrieved variance explanation value is much less than the recommended threshold value of 50%.
Confirmatory Factor Analysis was carried out for examining the validity and reliability of the study constructs. To begin with, the composite reliability and Cronbach’s alpha values for study measures were greater than 0.70 (Fornell and Larcker, 1981).
Table – 6: Fit Indices
Fit indices’ analysis of the research model
|
Model fit
|
Reference index (Upadhyay & Kumar, 2020)
|
Source
|
χ2/df
|
2.16
|
< 3
|
(Barrett, 2007; Falke, Schröder, & Endres, 2020; Oberski, 2014; Rosseel et al., 2017)
|
Goodness-of-fit index (GFI)
|
0.8973
|
> 0.9
|
(Barrett, 2007; Falke et al., 2020; Oberski, 2014; Rosseel et al., 2017)
|
Adjusted goodness-of-fit index (AGFI)
|
0.8777
|
> 0.9
|
(Barrett, 2007; Falke et al., 2020; Oberski, 2014; Rosseel et al., 2017)
|
Normed fit index (NFI)
|
0.894
|
> 0.9
|
(Barrett, 2007; Falke et al., 2020; Oberski, 2014; Rosseel et al., 2017)
|
Bentler–Bonnet Non-Normed fit index (NNFI)
|
0.933
|
> 0.9
|
(Barrett, 2007; Falke et al., 2020; Oberski, 2014; Rosseel et al., 2017)
|
Tucker–Lewis Index (TLI)
|
0.933
|
> 0.9
|
(Barrett, 2007; Falke et al., 2020; Oberski, 2014; Rosseel et al., 2017)
|
Comparative fit index (CFI)
|
0.94
|
> 0.9
|
(Barrett, 2007; Falke et al., 2020; Oberski, 2014; Rosseel et al., 2017)
|
Standardized Root Mean Square Error of Approximation (SRMR)
|
0.049
|
< 0.08
|
(Barrett, 2007; Falke et al., 2020; Oberski, 2014; Rosseel et al., 2017)
|
4.5.2. Path Model
The path model is based on Model − 2 which has shown an adequate model fitness and validity. The results of the path model are shown in Figure − 4. The path model shows the results of the hypotheses and their significance (Refer to Table − 9 for hypothesis results). The regression results of the path model are provided in Table − 8.