LENG SOCHEA

THE CAMBODIAN EXPERIENCE OF DEMINING
AFTER THE CIVIL WAR – 3

BECOMING MINE-FREE
BY THE 2025 DEADLINE

 

In the third and concluding part of his pathbreaking study of demining in Cambodia, Dr. Leng Sochea presents his findings using both qualitative and quantitative methods. The author makes a set of significant recommendations that are essential reading for government policymakers to ensure that the country not only meets its 2025 deadline to free the nation of the scourge of mines, but also to become self-reliant afterwards to remove any remaining mines. Donors should ensure that there is no duplication of existing initiatives. Where necessary, the national standards and approaches should be amended to make full use of non-technical and technical surveys, based on international standards. The authorities should look for new donors such as China, Russia, India, South Korea, and the ASEAN Regional Mine Action Centre to get more funding before 2025. The author advises the government to restructure the mine action framework and funding to fit into the new context.

The objective of this study was to (1) determine the factors contributing to the effective management of demining operations (hereafter known as “Demining Factors”) across Cambodia by the Cambodian Mine Action Center (CMAC); (2) to assess how each factor influenced effective management of demining in Cambodia; and (3) to provide suggestions and recommendations to the Royal Government of Cambodia and the donors related to mine action in the country.

This research project faced many limitations relating to both the methodology and the primary research questions. First, with respect to the research question, the research was primarily interested in demining in Cambodia from 1992 to 2016. The sampling techniques, therefore, reflect this interest and the findings should be applied with caution outside of Cambodia or the mine action context. Second, the project focuses solely on measuring the productive performance of demining from 2006 to 2016. It uses regression models to analyze data. Finally, the project faced difficulty in gaining reliable facts from the existing literature regarding the current status of effective management of demining in Cambodia because there is insufficient research undertaken in this field. The survey conducted by local teams generated specific discussions on the methodological limitations of this research project, for instance, a thorough discussion of each unit of analysis and its limitations.

This study brought together theory and practice. As for theory, this project expands previous research on demining and contributes to the literature on effective management of the factors determining demining, which traditionally concentrates on CMAC rather than other demining operators. The use of CMAC demining data from 2006 to 2016 enabled this study to test theories of clearance performance and helped to confirm and expand the scope of theoretical applications. In practice, this study is significant for demining operators in Cambodia. The results will determine the factors influencing the performance productivity of demining operators and will assist demining operators to improve their clearance performance.

 The reason that this research focused only on the CMAC is because it is the sole National Demining Operator, and it is the biggest and best-known operator with rich resources, know-how and experience in the field of demining internally and externally.

To assess the relative importance of the effectiveness of the factors surrounding demining in Cambodia, the study used quantitative methods in collecting data. Questionnaires were distributed by using convenience and quota sampling methods among 414 informants of which 39 questionnaires were distributed to supervisors and staff at the CMAC headquarters/field deminers and the Cambodian Mine Action and Victim Assistance Authority (CMAA), 14 instructors and supervisors of a Training Center in Kampong Chhang, 65 supervisors/de-miners and Provincial Mine Action Committee (PMAC)/Mine Action Planning Units (MAPUs) in Banteay Meanchey, 185 supervisors, deminers and PMAC/MAPU in Battambang), 37 supervisors, de-miners and PMAC/MAPU in Kampong Cham, and 76 supervisors, deminers and PMAC/MAPU in Siem Reap province.

The data analysis was based on software application: Statistical Package for Social Sciences (SPSS) 18 and Analysis of Moment Structures (AMOS) 18. A conceptual model of the research was constructed with hypothesis testing to assess their implications on demining outcomes. 

The Research Construct
Relationship between Demining Factors (DFs)
and Effective Management of Demining (EMD)

The interconnected nature of the constraints is paramount to gaining an understanding of the criteria or factors that impact the future development and implementation of new technology. Cultural, social, economic, political and operational conditions are unique to each mine action operation, thereby creating extremely varied conditions that make designing new technology for the Humanitarian Mine Action (HMA) field complicated and difficult.[1]

Interviews with CMAC management reveal that three factors are commonly being used by CMAC to clear Mine/ERW (explosive remnants of war) in the field to improve productivities through technology development and innovation such as non-technical surveys, technical surveys and full clearance; standard operating procedures (SOPs) development; and revision and training to provide new skills and multi-skills to improve demining capacity. Based on historical data from 1992-2004, CMAC achieved mine clearance of 123,189,778 square metres over a 12-year period as the yearly average was only 24,637,956 sq. metres due to limitation of tools development and their integration/combination as the operations were more focused on full clearance techniques. From 2005 to 2009, CMAC achieved a clearance of 141,811,307 sq. metres over a five-year period as the yearly average increased to 28,362,261 sq. metres due to the introduction of new technology, methodologies and integration/combination of various tools to support field operations such as brush cutter, Mine Detection Dog (MDD), etc. with trained skills staff. From 2010 to 2014, CMAC achieved 403,818,582 sq. metres for a five-year period with a yearly average of 80,763,716 sq. metres due to the introduction of the abovementioned techniques. Hence, this study proposes Hypothesis 1 (H1): The Demining Factors Positively Lead to Effective Management of Demining.

Relationship between the Donor Factors (DNFs)
and Effective Management of Demining (EMD)

In May and June 2010, the Geneva International Centre for Humanitarian Demining (GICHD) commissioned a survey of 25 donors that had contributed to the mine-action programs. The study’s objective was to gain insight into the donors’ motivation in funding the programs, as well as the issues that played a role in driving their continued support and the factors that would influence future funding. The findings indicate that short-term commitment and financial support remain strong. However, the sustainability of the current level of support for future mine action was difficult to ascertain.

Donors responding to the survey indicated that in the near future they would be subject to program reviews, multi-year approvals for the renewal of funding for mine action, or broader-defined programs that include mine action. They anticipated budget cuts after 2015 (as mentioned in the GICHD survey) and planned reduction in expenditures in mine action step by step. Consequently, donors and officials responsible for mine action in affected countries needed to rethink their approach to funding. In fact, the budget for mine action has been gradually cut since 2015.

The Ottawa Convention has had remarkable success and provided substantial assistance to countries in need. However, monetary resources need to be redirected to continue achieving viable results in the coming years. In the future, many factors are expected to converge, posing challenges and offering opportunities to officials concerned with mine action. New approaches are needed to confront continuing challenges such as growing competition for financial resources in the broader peace and security field, a more pronounced desire to integrate mine action in the security-development nexus, reduced human resources in donor administrations dedicated to mine action, and greater affected-country ownership and capacity for dealing with residual mine and ERW contamination. Officials will need to work on strategies for integrating capacity-building into government priorities in affected countries, ensuring maximum protection of populations at risk, reducing the size of suspected geographies and concentrating on priority areas for socio-economic development.[2]Hence, this study proposes Hypothesis 2 (H2): Donor Factors Positively Lead to Effective Management of Demining

Relationship between Government Factors (GF)
and the Effective Management of Demining (EMD)

A national government’s interference or support will impact productivity and clearance rates of operations. Most respondents listed their national governments’ interference or support of Humanitarian Mine Action (HMA) as one of the primary factors that affect the productivity and clearance rate of operations. A national government can hamper productivity through corrupted use of donor funds or by prioritizing land that benefits one specific person rather than the community. However, the opposite also has proved to be true. A respondent from Tajikistan reported that the government’s strong support of HMA was a key factor that ensured smooth and productive operations.[3]Hence, this study proposes Hypothesis 3 (H3): Government Factors Positively Lead to Effective Management of Demining.

Conceptual Model framework

IV is Independent Variables

DV is dependent Variables

Selection of Subjects

A total of 414 respondents of CMAC were selected as subjects who would complete the questionnaire. The respondents were selected by quota sampling of 25 percent at every demining unit.

Location	Unit	Staff	Section	Remarks
Phnom Penh	CMAC HQ & CMAA	155	39
Kampong Chhnang	Training Center	54	14
Banteay Meanchey	Demining Unit 1	259	65
Battambang	Demining Unit 2	737	184
Kampong Cham	Demining Unit 5	147	37
Siem Reap	Demining Unit 6	305	76
Total		1,657	414

Instrumentation
The Demining Factors

A literature review of numerous research articles was conducted to explore questionnaire items used to measure the dimensions of the construct of the demining factors. Four questions were used to measure demining factors.[4]Each item was measured on a five-point Likert Scale (1 = strongly disagree; 5 = strongly agree). The questions covered the following areas of concern:

1. Are the deminers offered good health care?
2. Are salaries and incentives for deminers acceptable?
3. Are the tools of demining up to date?
4. Is the Standard Operation Procedure concise and clear?

Measuring the Donor Factors

Three questions were used to measure the donor factors.[5] Each item was measured on a five-point Likert Scale (1 = strongly disagree; 5 = strongly agree).

5. Is the pool funding important for effective demining in priority areas?
6. Is bilateral funding important for effective demining in targeted areas?
7. Is development project funding important to fulfill demining project development?

Measuring the Government Factors

Four questions were used to measure the government factors. Each item was measured on a five-point Likert Scale (1 = strongly disagree; 5 = strongly agree).

8. Is clear government policy and regulation on demining action important for effective demining?
9. Is sufficient government funding for overhead costs (utilities, food, etc) important for effective demining?
10. Is the commitment to implement the National Action Strategy Plan important for effective demining?
11. Is the 10-year extension request (2010-2019) important for effective demining?

Measuring The Effectiveness
Of Management Of Demining

Two-item questionnaires were used to measure the success of demining.[6]Each item was measured on a five-point Likert Scale (1 = strongly disagree; 5 = strongly agree). The questionnaire covered the following areas of concern:

12. The number of mine causalities in my province has decreased dramatically.
13. The area of land cleared from mines in my province has increased steadily.

The Research Sites

For the purpose of this study the following research sites were chosen: Phnom Penh, and the five provinces of Kampong Chhnang, Kampong Cham, Battambang, Banteay Meanchey and Siem Reap. They were chosen because of two reasons: first, these were the most affected provinces in Cambodia, and secondly, all the demining operators concentrated their demining work in these provinces.

Data Collection Procedure

The population sample for this study consisted of deminers aged eighteen years or older who were working in the capital city, Phnom Penh, and five selected provinces. A sampling plan was developed to ensure that the appropriate respondents were included in this study. As Structural Equations Modeling (SEM) matures and additional research is undertaken on key research design issues, previous guidelines of required sample sizes of 300 are no longer appropriate. It is true that larger samples generally produce more stable solutions that are more likely to be replicable. However, it has been shown that sample size decisions must be made based on a set of factors such as the multivariate distribution of the data, estimation technique, and model complexity. Based on the multivariate distribution of data, a generally accepted ratio to minimize problems with deviations from normality is fifteen respondents for each parameter estimated in the model. The estimation technique suggests sample sizes in the range of 150 to 400. It should be noted that as the sample size becomes larger (＞400), the method becomes more sensitive and when any difference is detected, it suggests that goodness-of-fit measures are a poor fit. In the simplest sense, the model complexity in SEM leads to the need for a larger sample.[7]

Based on the above discussion of sample size, 300 deminers were targeted for this study. Applying the convenience- and quota-sampling technique, deminers in Phnom Penh and five provinces were approached and asked to participate in the study, stating their participation was voluntary and that their identities would be kept confidential. In the first quarter of 2016, respondents were asked to either complete the fifteen-minute survey on the spot or fill a take-home questionnaire and return it to the researcher at their earliest convenient time.

The Data Analysis

In order to test the hypotheses proposed in this study, the Statistical Package for Social Sciences (SPSS) 18.0 and AMOS 18.0 were employed to analyze the collected data. The data analysis methods are explained in detail below.

 The Descriptive Statistics Analysis

We analyzed the descriptive statistics to explain the main characteristics of the collected data in quantitative terms. To gain a deeper understanding of the features of each variable, descriptive statistics analysis was used to illustrate the means, standard deviation, and rank of the characteristics of respondents such as gender, age, marital status, and the construct variables including the demining factors, the donors’ factors, the government factors, and the factors of effective management of demining.

The Exploratory Factor Analysis and Reliability Test

To verify the dimensionality and reliability of the research constructs of this study, we conducted several purification processes such as exploratory factor analysis, correlation analysis, and internal consistency analysis (Cronbach’s Alpha). First, exploratory factor analysis (EFA) was employed to identify the dimensionality of each research construct, to select questionnaire items with high factor loading, and to compare the selected items with items suggested theoretically. EFA is good for detecting “misfit” variables. Generally, an EFA prepares the variables to be used for cleaner structural equation modeling.[8] In order to identify the dimensionality and reliability of the research constructs, it was necessary to conduct the measurement items’ purification procedure using EFA analysis, which consists of the factor loading, Eigenvalue, cumulative explained variance, and communality of the factors being extracted from the measurement items.

To identify the internal consistency and reliability of the construct measurement, we conducted item-to-total correlation and internal consistency analysis (Cronbach’s α). Factor analysis was used to identify the dimensionality of the construct and to select questionnaire items with the principle component extraction method. Any of the factor loadings which were less than 0.5 were deleted until all of the existing factors with factor loading were equal or larger than 0.5. More than two factors may be extracted and the difference between the largest factor loading and second-largest factor loading in terms of absolute value should be equal or larger than 0.3. In addition, the communality should be equal or larger than 0.5. Any item which did not meet the above criteria was deleted. Latent roots (Eigenvalues) and other criteria were used to determine the number of dimensions to be extracted from the principal Component Factor Analysis. The selected criteria are Factor Loading ≧0.5, Eigenvalue ≧1, accumulatively explained variance ≧ 0.5, item-to-total correlation ≧ 5, and coefficient alpha (α) ≧ 0.7.[9]

The Confirmatory Factor Analysis (CFA)

Confirmatory Factor Analysis, which was also used within the Structural Equations Modelling (SEM), is the most common method used to evaluate construct validity. Construct validity is a more broadly applied form to measure validity than content validity and criterion validity. CFA is used to test the relationship between observed indicators and latent constructs and to assess the convergent validity of the measurement model. CFA procedures have two other factor models—first-order factor model and second-order factor model. Second-order factor model was adopted to examine all the research constructs and test the fit of the overall model. To satisfy the criteria of CFA, as suggested byJöreskog and Sörbom,[10]and Hair et al,[11] generally the ratios (χ²:d.f.) of chi-square goodness-of-fit to degree of freedom should be on the order of 3:1, root mean squared error of approximation (RMSEA) of less than 0.50, comparative fit index (CFI) and non-norm fit index (NNFI) in excess of 0.90, all standardized loading should exceed 0.50, and each indicator t-value exceed 1.96 (p<0.001).

CFI = the comparative fit index (CFI)[12]is given by:

        

Where C_, is the discrepancy, d is the degree of freedom, NCP is the noncentrality parameter estimate for the model being evaluated, and C_bis the discrepancy and d_bis degrees of freedom and the noncentrality parameter estimate for the baseline model.

NFI = normed fit index (NFI) or ∆₁ in the notation of Bollen (1989) can be written:

     

Where C = nF is the minimum discrepancy of the model being evaluated and Cb = nFb is the minimum discrepancy of the baseline model.

Structural Equations Modeling (SEM)

Finally, Structural Equations Modeling is used to indicate the relationship among the constructs and multiple variables. SEM is used to combine perspectives of multiple regressions and factor analysis to assess a series of interrelated dependence relationships at the same time through a multivariate technique. SEM’s characteristics derive from two basic components: the measurement model permits researchers to take several variables for a single independent or dependent variable, and the structural model explains the relationship of the independent to dependent variables. Compared to other statistical techniques, SEM has higher ascendancy because it has the capability to assess multiple and interrelated dependency relationships. It can also exhibit unobserved concepts or latent variables in those relationships and explain the measurement error in the business process of assessment. More specifically, SEM is the only multivariate technique that allows the simultaneous estimation of multiple equations. These equations represent the way constructs relate to measured indicator items as well as the way constructs are related to one another. Thus, when SEM techniques are used to test a structural theory, it is the equivalent of performing factor analysis and regression analysis in one step. Therefore, SEM has become an extremely popular technique to test a theory in the social sciences based on these key advantages.[13]

The correspondences of the actual or observed covariance or correlation matrix are measured through the method of goodness-of-fit. Goodness-of-fit tests can determine whether the model can be either accepted or rejected. There are three types of the goodness-of-fit measurements: absolute fit, incremental fit and parsimonious fit.[14] Absolute fit measures can test the overall model fit and there is no need to do any adjustment for the degree of over-fitting. Incremental fit measures can compare the research model to some model which was specified by the researcher. Parsimonious fit measures can offer a comparison between models that have different estimated coefficients. The purpose is to determine the amount of fit achieved by each estimated coefficient. This study follows the principles offered by Bagozzi and Yi[15]in order to evaluate the goodness-of-fit of the research model through the overall model fit.

By adopting SEM, this study tests the relationships between the demining factors, donor factors and government factors and demining success. Cheung and Lau[16] suggest that SEM has several advantages over the hierarchical regression approach to mediational analyses. First, SEM provides a better statistical tool to investigate latent variables with multiple indicators. Secondly, the measurement errors in the model can be controlled when relationships among variables are examined. Thirdly, the SEM approach allows for the analysis of a more complicated model. Finally, SEM depicts a clear model that helps ensure that all relevant paths can be included and tested, without omitting any.[17] The AMOS 18.0 version package software was used to perform SEM in order to verify the interrelationships in the entire research concept.

Research Results

This part is subdivided into four sections. The first section describes the Descriptive Analysis of the characteristics of respondents and the descriptive statistics of the questionnaire items. The second section presents the results of descriptive analysis, as well as the reliability test for the collected data, and the results of data analysis associated with the proposed research hypotheses. The third section offers the Exploratory Factor Analysis and reliability test of the collected data. The final section shows the results of the Structural Equations Modeling, which tests the overall fit of the model and relationships links of the hypotheses proposed in the conceptual model.

The Distribution Target Areas of the Questionnaires

City/Province	Unit	Staff	Section
Phnom Penh	CMAC HQ & CMAA	155	39
Kampong Chhnang	Training Center	54	14
Banteay Meanchey	Demining Unit 1	259	65
Battambang	Demining Unit 2	737	184
Kampong Cham	Demining Unit 5	147	37
Siem Reap	Demining Unit 6	305	76
Total		1,657	414

 Source: CMAC HR Department 2016

The Characteristics of Respondents

A survey design questionnaire was used to collect data from deminers and stakeholders in specific demining operation areas in Battambang, Banteay Meanchey, Pailin, Kampong Chhang, Kampong Cham, Oddar Meanchey, as well as at CMAC headquarters, for a six-month period from the end of September 2016 to March 2017. Applying the quota and sampling technique, a total set of 414 questionnaires were distributed, of which 320 sets were returned and eventually 300 sets were usable for testing, yielding a response rate of 77 percent.

The table below (Characteristics of Respondents) summarizes the profiles of the respondents, including gender, age, marital status, family background, work experience, and income level. Among the 300 usable samples, male deminers outnumbered female deminers 272 to 28 (male deminers 90.7 percent; female deminers 9.3 percent), and more than 47.3 percent of all respondents were aged over 45 years old. About 63.7 percent of research participants had over 7 years’ demining experience, followed by 31.3 percent with 4 to 7 years’ experience, and only about 5 percent of respondents had less than 3 years’ experience. By income-level, 52 percent of the respondents earned US$ 301 to US$ 500 per month of salary, 39.7 percent earned less than US$ 300, just over 3 percent earned over US$ 701, and about 5.3 percent earned between US$ 501 to US$ 700.

Characteristics of Respondents (n=300)

Categories	Frequency	Percentage
Gender
1.    Male	272	90.7%
2.    Female	28	9.3%
Age
1.    Less than 18 years old	0	0
2.    18-25 years old	8	2.7%
3.    26-35 years old	29	9.7%
4.    36-45 years old	121	40.3%
5.    Over 45 years old	142	47.3%
Marital Status
1.    Single	9	3%
3.    Married	291	97%
Major
1.    Operations	137	45.7%
2.    Support	141	47%
3.    Stakeholder	22	7.3%
Occupation
1.    Deminer	58	19.3%
2.    Mobile team	41	13.7%
3.    Supervisor	143	47.7%
4.    Management	21	7%
5.    Others	37	12.3%
Monthly Income
1.    $100 & below	-	-
2.    $101-$300	119	39.7%
3.    $301-$500	156	52%
4.    $501-$700	16	5.3%
5.    $701 and above	9	3%
Personal  Skill
1.    1- 3 years	15	5%
2.    4 - 7 years	94	31.3%
3.    Over 7 years	191	63.7%

Descriptive Statistics

The table below (The Results of Descriptive Statistics) shows the descriptive statistics of the questionnaire items of the four factors including the demining factors, donor factors, the government factors and the effective demining management factors. All research variables are measured by using 5-point Likert Scale. All research variables are satisfied with the level of agreement from the 300 respondents (i.e. mean scores range from 3.63 to 4.71).

The Results of Descriptive Statistics (n=300)

Questionnaire Items	Mean	Std. Deviat
Demining Factors
1. The deminers are offered good health care.	3.63	0.52
2. Salary and incentives for deminers are acceptable	3.86	0.47
3. Tools of demining are up-to-date.	4.19	0.56
4. Standard Operation Procedure is concise and clear.	4.21	0.57
Donor Factors
1. Pool funding is important for demining success in priority area.	3.73	0.62
2. Bilateral funding is important for demining success in targeted area.	3.88	0.54
3. Development project funding is important to fulfill demining project development.	4.16	0.55
Government Factors
1. Clear policy and regulation on mine action from government are important for effective management of demining.​	3.69	0.63
2. Sufficient funding on overhead cost (utilities, food,) from the government is important for effective management of demining.	3.88	0.61
3. The commitment and implementation of the National Action Strategy Plan are important for effective management of demining.	4.10	0.51
4. Ten-year extension request.	4.25	0.56
Effective Demining Management Factors
1. Casualties from mine-related accidents have steadily decreased in recent years.	4.53	0.52
2. The size of cleared land from mines has steadily increased in recent years.	4.71	0.46

• Mean scores range from 3.63 to 4.71 is acceptable (average 2.5).
• Deviation scores range from 0.46 to 0.63 is acceptable (under control).

The Exploratory Factor Analysis (EFA) and Reliability Test (RT)

The table below (Exploratory Factor Analysis) provides detailed statistical results of the EFA with all factor loadings ranging from 0.704 to 0.890, which are greater than the cutoff point (0.5), variance explained greater than 50 percent, and Cronbach’s Alpha greater than 0.7.

Exploratory Factor Analysis, EFA (n=300)

Factors /   Items	Factor Loading	Variance Explained (%)	Cumulative Variance Explained (%)	Cronbach’s Alpha
Demining Factors         >0.50        >50               >50              >0.70
1. The deminers are offered good    health care.	0.721	53.85	53.85	0.713
2. Salary and incentives for deminers are acceptable.	0.712
3. Tools of demining are up-to-date	0.758
4. Standard Operation Procedure is    concise and clear.	0.743
Donor Factors
1. Pool funding is important for effective management demining in priority area.	0.821	63.81	63.81	0.715
2. Bilateral funding is important for effective management demining in targeted area.	0.821
3. Development project funding is important to fulfill demining project development.	0.753
Government Factors
1. Clear policy and regulation on mine action from government are important for effective management of demining.​	0.799	56.73	56.73	0.745
2. Sufficient funding on overhead cost (utilities, food) from the government is important for effective management of demining.	0.765
3. The commitment and implementation of the National Action Strategy Plan are important for effective management of demining.	0.704
4. Ten-year extension request.	0.741
Effective Management of Demining Factors
1. Casualties from mine-related accidents have steadily decreased in recent years.	0.890	79.19	79.19	0.733
2. The size of cleared land from mines has steadily increased in recent years.	0.890

Factor loading scores ranging from 0.712 to 0.890 are acceptable.

• Variance Explained scores ranging from 85 to 79.19 are acceptable.
• Cumulative Variance Explained scores ranging from 85 to 79.19 are acceptable.
• Cronbach’s Alpha scores ranging from 713 to 0.745 are acceptable.

The Measurement Model

A two-step approach proposed by Anderson and Gerbing[18]was used to assess the proposed conceptual model in the study. First, a confirmatory factor analysis (CFA) with maximum likelihood estimation was first conducted to analyze the validity and reliability of the constructs in our conceptual model. Then, the structural equations modeling (SEM) was used to test the predicted relationships among the constructs in the proposed conceptual model.

The Confirmatory Factor Analysis (CFA)

In the previous section, EFA analyses were conducted to determine the correlations among the variables in a dataset and to delete the misfit variables. In this section, a preliminary CFA was conducted and the model adequacy was assessed by the fit indices, as suggested byJoreskog and Sorbom[19] (1996), and Hair et al[20] (2010).

Confirmatory Factor Analysis (CFA), n=300

The Convergent Validity

We evaluated the convergent validity of the study in three ways: by evaluating the strength and significance of the factor loadings, by examining the composite reliabilities, and by inspecting the average variance extracted (AVE) for each construct. As shown in the table above, the factor loading of each item was greater than 0.50 threshold, the construct reliability estimates of all the constructs exceeded the critical value of 0.70, and the values of the average variance extracted were well above the suggested value of 0.50.[21]These fit indices indicate the measurement model has a good convergent validity as our usable data is just 300, which is in the standard of the 150-400 sample size requirement for the structural equation modeling analysis as suggested by Hair, et al.[22]

The Confirmatory Factor Analysis, CFA (n=300)

Constructs	Standardized factor loading	Error variance	t-Value	AVE	Construct reliability
Demining Factors (DF) >0.50                <0.020             >+1.96        >0.50           >0.70
DF1	0.592	0.017	7.908	0.956	0.713
DF2	0.570	0.014	7.691
DF3	0.677	0.019	8.650
DF4	0.639	0.020	A
Donor Factor (DNF)
DNF1	0.716	0.024	8.213	0.956	0.715
DNF2	0.720	0.019	8.224
DNF3	0.600	0.020	A
Government Factor (GF)
GF1	0.736	0.023	9.605	0.940	0.729
GF2	0.477	0.050	6.931
GF3	0.735	0.019	9.599
GF4	0.691	0.021	A
Effective M   Management of Demining (EMD)
EMDF1	0.772	0.017	9.514	0.974	0.733
EMDF2	0.757	0.013	A

• Standardized factor loading score range from 477 to 0.772 is acceptable.
• Error variance score range from 013 to 0.050 is acceptable.
• t-value score range from 931 to 9.514 is acceptable.
• AVE score range from 940 to 0.974 is acceptable.
• Construct reliability score range from 713 to 0.733 is acceptable.

Model fit statistics: χ²/d.f. = 70.254/59 = 1.191, p-value = 0.150, CFI = 0.988, IFI = 0.989, NFI = 0.933, TLI = 0.985, RMSEA = 0.025; AVE = (Σλ2)/ [Σλ2+Σ(θ)], where Σ = summation over the indicators of the latent variable, λ = indicator loadings, θ = indicator error variances.

Note: AVE = average variance explained, CR = construct reliability, df = degree of freedom, CFI = Comparative Fit Index, IFI = Incremental Fit Index, NFI = Normed Fit Index, TLI = Tucker Lewis Index, A regression weight was fixed at 1.

The Structural Equation Modeling (SEM)

A maximum likelihood estimation method with AMOS 18.0 was used to test the predicted relationships among the constructs in the proposed conceptual model. The overall model achieves a good fit with χ² (300) = 151.4000 (p = 0.000), χ²/df = 151.4000 /60= 2.523, meeting the criteria of value of less than 3 (χ²/df < 3), GFI = 0.931, AGFI = 0.895, CFI = 0.905, IFI = 0.907, TLI = 0.877, which basically satisfied the threshold as suggested by Hair, et al (2010).[23]Accordingly, the proposed conceptual model of entrepreneurial success produced a moderate model-fit.

The results of the structural equation modeling produced four supported hypotheses and two partially supported hypotheses. H1, H2 and H4 are positively supported because the capability ratio t. value is greater than the cut-off point of +1.96 as suggested by Hair et. al (2010). Hypotheses 1 and 3 are partially supported because the C.R. value is somewhat smaller than the cut-off value of +1.96 as our 300 samples data is in the norm of the requirement to perform the Structural Equation Modeling analysis using AMOS software. The next two tables report the proposed direct paths of variables in the model.

Structural Equations Modeling (SEM), N=300

Model = Standardized estimates, Chi-square/df (151.400/60) = 2.523, Goodness of fit Index (GFI) = 0.931; Adjusted Goodness of Fit Index (AGFI) = 0.895; Comparative fit index (CFI) = 0.905; Incremental Fit Index (IFI) = 0.907; Tucker Levis Index (TLI) = 0.877, p = 0.000. DNF = Donor Factor; GF = Government Factor, and EMDF = Effective Management of Demining Factor.

Structural Equation Modeling, SEM (N=300)

Path	Direct Effect	C.R    (t-value)	p-value	Hypotheses Testing
H1: Demining Factor → Effective Management of Demining	0.131	7.523	0.001	Fully Support
H2: Donor Factor →Effective Management of Demining	0.171	4.158	0.001	Fully Support
H3: Government Factor →Effective Management of Demining	0.100	2.117	0.034	Fully Support

• 171 is very significant, 0.131 is significant.
• Direct effect score-range from 0.100 to 0.171 is acceptable.
• Construct Reliability (t-value) score-range from 2.117 to 7.523 is acceptable.
• P-value score-range from 0.001 to 0.034 is acceptable.

Discriminant Validity

In addition to convergent validity, discriminant validity was assessed. The dataset is confirmed if the Average Variance Explained (AVEs) are larger than the squared correlation coefficients between the constructs (Fornell & Larcker, 1981).[24] From the table above, it is clear that the AVEs of all variables are higher than the squared correlations of any pair of variables, which supports the discriminant validity of all measures. The AVE is calculated using the formula where AVE= (Σλ2)/ [Σλ2+ Σ(θ)], where Σ=summation over the indicators of the latent variable, λ= indicator loadings, θ=indicator error variances. Diamantopoulos and Siguaw (2000).[25]

Discriminant Validity

Constructs	Mean	SD	DF	DNF	GF	EMDF
DF	3.9750	0.3940	0.956
DNF	3.9256	0.4602	.437**	0.956
GF	4.0733	0.4987	-.084	-.017	0.940
EMDF	4.6217	0.4397	.524**	.411**	-.001	0.974

Note: The bold numbers in the diagonal row are square roots of the average variance extracted (AVE); inter-construct correlation is shown off the diagonal; **p < 0.01, *p < 0.05. AVE = (Σλ2)/ [Σλ2+ Σ (θ)], where Σ = summation over the indicators of the latent variable, λ = indicator loadings, θ = indicator error variances DMF = Demographic Factor; DF = Demining Factor; DNF = Donor Factor; GF = Government Factor and EMDF = Effective Management of Demining Factor. SD denotes Standard Deviation.

A Discussion of the Research Findings

This section is composed of three subsections: a discussion of research findings; research contributions; and research limitations and future research. The research discussion explains the findings of study and compares the current findings with previous literature review. The research contributions and recommendations discuss how the current findings contribute to the effective management of demining both theoretically and managerially. (This section is divided into two sub-sections. The first one discusses the theoretical aspects and provides recommendations to the Royal Government of Cambodia and related donors and institutions. The second one discusses the managerial aspects and provides recommendations to deminers and the demining institutions). The final part of this section summarizes and concludes the findings of this study along with the limitations of the research and offers avenues for future research.

The Research Results

This study investigates the relationships between the demining factor, donor factor, government factor, and the factor of effective management of demining. The results of the structural relationship analysis show that the donor factor is the most important factor leading to the effective management of demining (0.171, t=4.158), followed by the demining factor (0.131, t=7.523), and finally the government factor (0.100, t=2.117). The results show that the Donors Factor is the most significant because without donor funding there would be no demining. The Demining Factor plays the second most significant role after the Donor Factor. The last one is the Government Factor which studies the role of the state as the coordinator and cooperator. Accordingly, this study suggests that policymakers such as donors, the Royal Government of Cambodia and all related partners, should consider these findings as very important in order to reflect upon the effective management of demining activities in the future.

The research findings also explain the benefits of the effective management of demining. It demonstrates that the human causalities from mines and explosive remnant of war would decline very rapidly and local people could start working safely on the cleared land, and that economic growth would follow naturally, if the deminers could remove all mines and explosive remnant of war from Cambodia. It would contribute greatly to the MAPUTO+15 declarations made by the State Parties to the Anti-Personnel Mine Ban Convention to free the world of mines by 2025.

The donor factor was the most important to lead the effective management of demining. The donor factor has three dimensions: pool funding, bilateral funding and the development projects. As we can see from the results of our data analysis, the donor factor has a strong influence on the effective management of demining. So, the donors who have supported the demining activities should reconsider the importance of pool funding, bilateral funding and development projects. The donors should provide enough financial support to all these three forms of funding in order to effectively remove mines from Cambodia.

This study shows the demining factor to be the second most important factor affecting the effective management of demining. The demining factor has four dimensions: good healthcare for deminers, salary and incentives for deminers, up-to-date tools, and excellent standard operating procedure (SOP). Therefore, in order to be more successful and effective in removing mines from Cambodia, all the related partners need to seriously reconsider the demining factors very carefully and provide the necessary healthcare, adequate salary and high incentives to support deminers’ livelihood, as well as up-to-date tools for demining activities, and very good standard operating procedures to the deminers, enabling them to work tirelessly to remove the mines. One of the riskiest jobs in the world, demining does not offer a good salary.

Last but not the least, the government factor also positively affects the effective management of demining. The government factor has four dimensions: clear policy and regulations, sufficient funding for overhead costs, commitment and implementation of the National Mine Action Strategy Plan, and the ten-year extension request. The Royal Government of Cambodia should set clearer policy and regulations, stipulating the priority areas for removal of mines, the timeframe to finish those tasks, and identify the next target areas. In addition, the government should provide sufficient funding to cover not only overhead costs and but should also have a strong commitment to implement the three types of demining: Emergency, Humanitarian and Developmental. The emergency demining stage lasted from 1992 to 1995 when the United Nations Transitional Authority in Cambodia and the United Nations High Commissioner for Refugees hired Halo Trust and Mines Advisory Group to clear the mines in order to resettle more than 300,000 Cambodian refugees who were arriving in Cambodia from the Thai border area. Then, in 1995, the stage of humanitarian demining began. Finally, the process of demining for development purposes started when the government set up the Provincial Mine Action Commission and Mine Action Planning Units in 2004. In fact, the government covered the overhead cost of humanitarian demining till now and employed the Royal Cambodian Armed Forces to clear the mines from all national road construction projects.

As the trend of donors suffering from funding fatigue is prevalent, the government should prepare to take over the donors’ role after 2025. For almost thirty years, the donors exercised power based on their financial contribution, but now the time has come for the government to take more responsibility for humanitarian demining.

Significance of this Three-Part Research Study

In conclusion, the first and most important theoretical contribution of this three-part research study lies in its development of the model of the effective management of demining. This study has generated important recommendations for the partners involved in clearing mines in Cambodia as well as for the donors who have committed to the Anti-Personnel Mine Ban Convention (APMBC). It has also contributed to assisting new researchers entering the field to conduct further analysis of the effective management of demining. It paves the way for new researchers to carry out deeper analysis of what may happen after the turning point in 2025

The Research Limitations and Future Research

While this research study has developed a model to investigate the effective management of demining in Cambodia, it is not a formula for clearing mines in all contexts or in other territories or countries. Potential limitations include scope, influence of variables and publication bias. We, therefore, took several measures to counteract potential problems. First, this research was conducted with deminers, team leaders and high ranking officers in five provinces of Cambodia, which may limit the generalizability (the extension of research conclusions from a study conducted on a sample population to the population at large) of the findings to demining activities in other specific countries.

Secondly, this study employed quota sampling methods, so the sample may not reflect the entire population of demining individuals of the whole country. Future research empirically testing this model with larger random samples or samples in other contexts, for example with demining individuals in different countries, would increase our understanding of this important research model of effective management of demining.

Thirdly, since most of the respondents in this study were male deminers, the research findings may not be generalized to all deminers, especially female ones. Accordingly, a future research study empirically testing the model with larger sample size is vital for the generalization of the current study.

Mined Areas Cleared by Different Deminers

Source: CMAA database.

The Recommendations

This research study offers a sequence of recommendations to mine action policymakers, donors, the Royal Government of Cambodia and other related partners, as follows:

1. The Development Partners (DPs): Prior to commissioning new research and the development projects, donors should ensure that there is both an operational need and no duplication of existing initiatives. Bilateral and multilateral donors should provide the bulk of resources needed to undertake much of the policy development and clearance operations throughout the country. Without their past contribution, Cambodia would not have made such remarkable achievements. Financial and technical support from DPs is still required to help Cambodia fulfill its obligation to the APMBC, and hopefully to achieve its National Mine Action Strategy (NMAS) 2018-2025 target.
2. The Expert Organizations: Cambodia receives periodic expert advice from the mine action organizations such as the Geneva International Centre for Humanitarian Demining (GICHD), James Madison University (JMU), and the United Nations Mine Action Service (UNMAS), etc. They should continue to provide technical assistance in a range of areas such as land release methodologies, gender mainstreaming, the UN mine action standards, and database and ISO standards. They can provide independent sector reviews. An emerging regional organization, the ASEAN Regional Mine Action Centre (ARMAC), which has been established in Phnom Penh, can ensure technical cooperation and information exchange on mine action among the ASEAN member states.
3. The Demining Operators: They should release land in locations where land is cultivated or inhabited by farmers, and those lands that have not been released in the database. Where necessary, national standards and approaches should be amended to make full use of non-technical and technical surveys, based on international standards and concepts of reasonable effort and tolerable risk. Capacity development of the national authorities should prioritize the application of land release approaches in order to assist states to draw fully on international standards and sector good practices. Innovative clearance approaches should respond to operational needs and should continue to draw on approaches used in sectors beyond mine action. Demining is one of the pillars of mine action, whose other components are mine clearance, victim assistance, mine risk education, advocacy, and stockpile destruction as well asenabling activities such as assessment and planning, mobilization and prioritization of resources, information management, human skills development and management training, quality management, and application of effective, appropriate and safe equipment. The demining operators should assign the highest priority to the welfare of deminers, with a view to improving their salary, incentives, and healthcare.
4. The Royal Government of Cambodia: The RGC should strengthen the Technical Working Group on Mine Action (TWG-MA), which isa consultative and coordinative mechanism between the government and the Development Partners to discuss policy and strategy issues. The Cambodia Mine Action Authority (CMAA) should look for new donors such as China, Russia, India, South Korea, and the ASEAN Regional Mine Action Centre to get more funding before 2025.
5. Zoning for Mine Clearance: The author of this study concludes that not a single demining operator or the CMAA has claimed that even one district or one commune is free from mines despite almost twenty-eight years of demining which cleared 1,893 square km of land. What has been happening behind the scenes? Perhaps it was due to the erroneous policies of the government, or the orientations of the donors, or the operators’ initiatives. The author proposes dividing the areas for clearance into three zones:

Zone A: Consists of seven eastern provinces which were the site of intensive U.S. aerial bombing. The responsibility of clearing the approximately 10 percent remaining unexploded remnants of war (ERW) in these areas should be left to the United States to fund. For the last ten years, the United States has diverted all funding from mine clearance in the northwestern part of the country to clear U.S. mines in the eastern provinces. There are some critical questions: Why does Cambodia spend other donors’ funds to compete with U.S. policy? For example, since 2018 CMAC has divered some funds,donated by China, away from high mine-risk areas to clear mines from non-risk areas in the eastern provinces. And why are scarce resources not spent in twenty-two highly mine-contaminated districts in CMAA databases?

Zone B: Located in various mine-contaminated areas spread across twenty-two high mine-risk districts in the northern and western parts of the country. All the demining operators should keep working with the donors to clear mines in these areas, as long as the funds last.

Zone C: Are the high mine risk districts/communes which were selected for clearance by the government using government funding (clearing of contaminated Base Line Survey areas shown in the CMAA database). In order to succeed, the Ministry of Economy and Finance should support and co-lead this mechanism along with the CMAA, under a clear procurement process for all operators.

6. MAPUTO+15 Declaration: This research study’s findings are that Cambodia has used best practices of clearance to support the MAPUTO +15 Declarations made by State Parties of the APMBC to free the world of mines by 2025.
7. Government Funding: The Royal Government of Cambodia is advised to provide more funding to the mine action sector in Zone C because after 2025 the Development Partners and foreign deminers will move away from Cambodia and the management of CMAC will shift under the umbrella of the Defense Ministry. Facing a short period of just five years to free the land of mines, the government should start preparing from now.
8. Police (future role in demining tasks): The existing police EOD informant budget of around US$ 500,000 per annum should be terminated and diverted to be spent on police EOD teams instead. Police EOD teams should be trained and equipped by the CMAC for emergency response at the grassroots’ level. At least one police EOD team should be set up in each province to perform these tasks in the future when external funding will end. The police are best suited for the task because their network is deployed from the commune level to the top, implying that the police possesses relevant information about the villages. Such a networked EOD team would be more effective than the existing model under which this task is managed by the demining operators.
9. The Provincial Mine Action Commission and Mine Action Planning Unit: The PMAC and MAPU mechanism, under sub-decree 70, needs to be amended to delegate all leadership authority to the sub-national level, including financial management, whereas the CMAA should only provide training, monitoring of technical issues and collection of data.

Conclusions

Planning and prioritization of land release operations is an essential process that directly influences the strategic achievements of national development and, therefore, the efficiency of the sector.[26] Cambodia has adopted a combination of top-down and bottom-up approaches.

The current mine action mechanism and spending is no longer effective and efficient. The government is advised to restructure the mine action framework and funding to fit into the new context.

Just five years are remaining to meet the deadline for mine-clearance under the Anti-Personnel Mine Ban Convention. The country is squeezed for time and money. There is a huge funding requirement of about US$ 400 million to meet the mine-clearance target. Without the contribution of donors, Cambodia would not have made such remarkable achievements. Financial and technical support of the DPs is still needed to help the country fulfill its obligation to the APMBC and accomplish its long-term strategy.

The flow of funds from the donors to the mine action sector represents 95 percent of financial assistance, whereas the government’s contribution is only 5 percent. Based on this research study, the donors perform a critical role, and they can exercise their money power to change the mindset of demining leaders to focus on the proposed zoning area policy in order to reduce the areas not yet cleared. Failure to adopt this policy will afflict the donors with symptoms of “funding fatigue” and drive out of sight the target to free the country of mines in 2025.

From a strengthened Technical Working Group on Mine Action (TWG-MA), the government should hive-off an ad-hoc team to work closely with donors to undertake the Zone C clearance in order to gradually erase the safari color from the mine map.

Early consideration should be given to the mine action arrangements that would exist after 2025. A transitional exit strategy should take into account several considerations and principles, such as:

Preparation: Establishment or strengthening over time of new or existing institutions ready to assume the responsibility for the remnant contamination.

Implementation: Gradual exit of the international demining operators and NGOs as the contaminated areas diminish.

Transfer of responsibility to the Cambodian institutions and agencies for any residual threat.

Transfer of assets, equipment, and personnel to replacement organizations.

Following completion of demining in 2025, a residual risk will remain because it is impossible to clear every piece of mine/ordnance. A national capacity is required to deal with any mine/ordnance causing local interference. The Cambodian institutional framework will determine the organizations needed and their role and address the residual mine/ERW threats. By 2025 Cambodia should have an appropriate national operational framework, infrastructure, resources and capacity available to address any residual threat risks. Cambodia should continue to provide a full disclosure of the monies it contributes to the various aspects of the mine action in a more transparent and accountable manner. In this way, it is hoped that a smooth transition will happen from international to full domestic responsibility for mine action, which will go a long way in reassuring international donors from the State Parties that Cambodia is ready to assume responsibility by itself.

The mine action sector needs to resolve the two key issues. First is an internal issue that needs the CMAA leadership to adapt both to the new arrangements and to the reallocation of scarce resources from now till the end state in 2025. The second is the external issue of fundraising for which the country has been reliant on the donors, who have been critical to the success of the sector. But if these issues are not solved in time, the country will not achieve its goal as stated in the Maputo Review Conference on a Mine-Free World in 2014. Cambodia had endorsed the Maputo +15 Declaration with an overarching ambition to intensify efforts to complete clearance to the fullest extent possible by 2025. The country is close to the winning post. It needs to keep going.

LengSochea holds a PhD (2018) in Business Administration from Asia Europe University, Phnom Penh, Cambodia; an MBA degree (2005) from the National University of Management, Phnom Penh; a Bachelor’s degree in Business Administration (2002) from the National Institute of Management, Phnom Penh; and a Bachelor of Law degree (1999) from the Royal University of Law and Economy, Phnom Penh. His varied career began as a worker making fishing nets at a Khmer Rouge factory from 1975 to 1979. Currently, he is an Adviser to the Chairman of the Cambodian National Election Committee since 2016, a rank equal to the Secretary General of the NEC. Concurrently, he is the Permanent Vice-Chairman of the Governing Council of the Cambodia Mine Action Centre, since 2011, ranking equal to a minister. He was the Vice-Chairman of the Governing Council of the Cambodia Mine Action Centre in 2011-2012, and he served as the Deputy Secretary General of the Cambodia Mine Action Authority during 2000-2011. He attended the Senior Mine Action Management course at Cranefield Army University in the UK in 2002, a Law and Economics Awareness course at L’Ecole Royale Administration, as well as training courses at the Ministry of Foreign Affairs. Earlier, he was the Head of the Public Information Office and the Spokesperson of the National Election Committee in 1998-2006, and the Deputy Director General in charge of Audio Visual and Media Centre at the Ministry of Information in 1994-2000. He began a career in government working as a Department Director at the Media Centre of the Ministry of Information between 1992 and 1994. Earlier, he was a journalist for the Japanese newspaper, Mainichi Shimbun, in 1991-92, reporting on the United Nations’ Transitional Authority in Cambodia.

END NOTES

[1]Jennifer Bilec-Sullivan, “Technical and non-technical considerations when developing and implementing new technology for the humanitarian mine action community,”https://benetech.org/wp-content/uploads/2017/12/Landmine_Report_2007-10-22.pdf  (Accessed 20 July 2016).

[2]Jean Devlin and Sharmala Naidoo, “Mine-action Funding: GICHD Survey of Donor Countries,” The Journal of ERW and Mine Action 14, no. 3, Article 8 (2010). https://commons.lib.jmu.edu/cisr-journal/vol14/iss3/8.

[3]Bilec-Sullivan,“Technical and non-technical considerations when developing and implementing new technology for the humanitarian mine action community.”

[4]Ibid.

[5]Sharmala Naidoo,Linking Mine Action and Development Guidelines for Policy and Programme Development 2010, http://www.gichd.org/lmad; and Vanna Mao, “Study on Enhancing Aid Effectiveness & Harmonization in Mine Action,”Cambodian Rehabilitation and Development Board/Council for the Development of Cambodia, April 2010.

[6]Bilec-Sullivan,“Technical and non-technical considerations when developing and implementing new technology for the humanitarian mine action community.”

[7]Joseph F. Hair, William C. Black, Barry J. Babin, and Rolph E. Anderson, Multivariate Data Analysis (7th ed.) (New Jersey: Prentice-Hall, 2009). 

[8] Ibid.

[9]Ibid.

[10]Karl G. Jöreskog and Dag Sörbom, LISREL 8: User’s Reference Guide (Chicago, IL: Scientific Software International, 1996).

[11]Hair, et al., Multivariate Data Analysis.

[12]P.M. Bentler, “Comparative Fit Indexes in Structural Models,” Psychological Bulletin 107, no. 2: 238-246.

[13]Hair, et al., Multivariate Data Analysis.

[14]Ibid.

[15]Richard P. Bagozzi and Youjae Yi, “On the Evaluation of Structural Equation Models,” Journal of the Academy of Marketing Sciences 16 (1988): 74-94.

[16]G.W. Cheung and R.S. Lau, “Testing Mediation and Suppression Effects of Latent Variables: Bootstrapping with Structural Equation Models,” Organizational Research Methods 11, no. 2 (2008): 296–325.

[17]Reuben M. Baron and David A. Kenny, “The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations,” Journal of Personality and Social Psychology 51, no. 6 (1986): 1173-1182.

[18]James C. Anderson and David W.  Gerbing, “Structural Equation Modeling in Practice: A Review and Recommended Two-Step Approach,” Psychological Bulletin 103, no. 3 (1988): 411-423.

[19]Jöreskog and Sörbom, “LISREL 8: User’s Reference Guide.”

[20]Hair, et al., Multivariate Data Analysis.

[21]ClaesFornell and David F. Larcker, “Evaluating Structural Equation Models with Unobservable Variables and Measurement Error, Journal of Marketing Research 18, no.1 (Feb. 1981): 39–50. 

[22]Hair, et al., Multivariate Data Analysis.

[23]Ibid.

[24]Fornell and Larcker, “Evaluating Structural Equation Models with Unobservable Variables and Measurement Error.”

[25]Adamantios Diamantopoulos and Judy A. Siguaw, Introducing LISREL (Sage, 2000).

[26]National Mine Action Strategy, 2018-2025, Royal Government of Cambodia.