Journal Search Engine
Search Advanced Search Adode Reader(link)
Download PDF Export Citaion korean bibliography PMC previewer
ISSN : 1225-8504(Print)
ISSN : 2287-8165(Online)
Journal of the Korean Society of International Agriculture Vol.27 No.1 pp.7-14

Improvements of Efficiency of Analytical Hierarchy Process (AHP) for Project Priority in International Rural Development

Jong-San Choi†
College of Agriculture & Life Science, Chonbuk National Univ., JeonJu 561-756, Korea
Corresponding author: (Phone) +82-10-8267-2918
May 12, 2014 January 6, 2015 February 26, 2015


This research was designed to review improvements of efficiency of Analytic Hierarchy Process (AHP) as a way to facilitate decision-making on project priority in international rural development in the following ways; to apply an AHP method; to discover issues that might arise in the research preparation, conduct, and analysis; and ultimately to expand a practical use of AHP in recipient countries. The survey questioned residents of the Dadeldhura district in Far-Western Region of Nepal, on the assumption that one donor country has a plan to implement projects for rural development in a community. This research collected 104 questionnaires. 55 questionnaires with a Consistency Ratio (CR) greater than 0.2 were excluded and 49 questionnaires were accepted for AHP analysis. The rejection rate reached 52.9%. The results of AHP analysis showed that residents considered Number of Beneficiaries the first priority in terms of Criteria. They selected Education as the first priority project that most reflected their opinion. The top three priority projects were Education, Public Health/Hygiene/Sanitation, and Income Generation. Although the AHP method can be a useful tool in deciding project priority, some problems were revealed during the process of preparing and carrying out a survey. By addressing these problems and improving on its application, the AHP method can be expected to contribute to the academic field of international rural development and to allow for a more accurate selection of projects that fully reflect residents’ needs at the early stages of implementation in rural communities of recipient countries.


    When a donor country promotes a specific project for rural development in a recipient country, all processes are a series of decisions from the establishment in the donor country to the completion in the recipient country. At the initial planning stage, the donor country should decide which country will be selected to support the rural development project. After selecting the recipient country, the donor country should decide which sites within the country will be chosen for project implementation. After the completion of these steps, decisions need to be made as to which of the available rural development projects will take priority. Having the project priority appropriately and correctly reflecting the needs of the project site’s residents would be one of the important factors improving the efficiency and effectiveness of rural development. If a specific project-in-progress is reconsidered due to inappropriately managing a priority at a baseline study to analyze the current situation and residents’ needs in the project site, it would cause the overall reduction of efficiency and effectiveness to a project. Most of all, some capital losses would not be avoided. Hence the good grasp of a priority in the community becomes a key process to a successful project. However, it is difficult to find literature addressing project priority decision-making from a mathematical and scientific approach in an international rural development sector. We conducted a questionnaire survey among residents of the Amargadhi municipality, Dadeldhura district, in the Far-Western Region of Nepal to apply Analytic Hierarchy Process (AHP) as a mathematical approach. Since this research is took place in a foreign country, we encountered difficulties during the research preparation, conduct, and analysis phases. Therefore, we tried to find improvements of efficiency of the AHP method for decision-making on a project priority in international rural development to finally expand and utilize it in other recipient countries.


    Nepal is located in South Asia and includes eight of the existing 14 summits in the world. Geographically, it is an isolated country surrounded by the Tibetan Autonomous Region of China and India with roughly rectangular in shape. Based on the District Development Profile of Nepal 2010/11, Nepal is divided into three major ecological regions: Mountain, Hill, and Plain. These regions occupy 35%, 42%, and 23% of the area, respectively. The altitudes range from 70 to 305 meters for the Plain region; from 305 to 4,877 meters for the Hill region; and from 4,877 to 8,848 meters for the Mountain region. Administratively, Nepal consists of 5 regions: Eastern, Central, Western, Midwestern, and Far-Western. The study was accomplished in the Amargadhi municipality Dadeldhura district, one district forming the Far-Western region. It is situated at 1,848 meters above sea level and approximately 770 Kilometers away from Kathmandu, the capital of Nepal. Wheat is the most widely cultivated crop in the Dadeldhura district. Paddy, maize, millet, and barley follow, in order. The Amargadhi municipality is the center of transportation and administration in the Dadeldhura district, and agriculture and business are the chief industries in this region.Fig .1

    Analytic Hierarchy Process (AHP) is known as a multiple- criteria decision analysis method and has been applied to such areas as selection, evaluation, benefit-cost analysis, priority, development, resource allocation, decision-making, forecasting, and medicine (Vaidya & Kumar, 2006). To apply AHP to international rural development, a hierarchy structure needs to be created (see Fig. 2 as an example). This structure is divided into the levels of Alternatives, Criteria, and Final Goal. The Alternatives level is the target of a project priority and includes 8 rural development projects:

    1. Infrastructure supports making the establishment of facilities such as rural roads, irrigation canals/ponds, community buildings and others.

    2. Income generation includes all activities and projects that generate capital.

    3. Public health/hygiene/sanitation provides all activities and projects to maintain the residents’ health conditions.

    4. Education refers to all activities and projects that promote students’ abilities to learn and think.

    5. Vocational training concerns all activities and projects designed to create jobs in a community.

    6. Social service support includes all activities and projects to improve accessibility to fundamental services by supporting health centers, banks, administrative agencies and non-government organizations (NGOs).

    7. Residential environment improvement comprises all activities and projects to create a healthy and comfortable residential environment, such as replacing and maintaining hygienic kitchen, developing a regular supply of clean drinking water, cultivating flowers and fruit trees in community areas, and establishing recreation spots within the community.

    8. Tourism development refers to all activities and projects to promote tourism.

    Economic Impact, Environmental Consideration, and Number of Beneficiaries were selected as the categories in the Criteria level. Economic Impact refers to an appraisal standard on how much a project affects a regional economic development. Environmental Consideration makes a valuation basis on the level of influence a project has on the natural environment in the study area. Number of Beneficiaries is based on how many residents can share in the benefits of a project. The Final Goal level describes a priority decision of international rural development projects.

    In cases where there are many alternatives, the number of pairwise comparisons also increases. When a pairwise comparison is applied at the Alternatives level in this study, it caused a time-consuming analysis and made the Consistency Index (CI) worse (Eizo & Taka, 2012). CI is used as a yardstick for judgment on how much consistency there is between two comparative factors in a pairwise comparison. To supplement these potential problems, an absolute measurement is utilized to evaluate all projects with regard to criteria. However, the pairwise comparison that makes a quantitative evaluation possible is applied to find the weight of fundamental scales of the absolute measurement and the weight of the individual criterion by comparing each other at the Criteria level.

    The survey questionnaire consisted of three sections. The first section consisted of eight questions focusing on resident demographics, the second included three questions for measuring the weight of criteria by pairwise comparison with nine scales, and the final section organized three questions for measuring the absolute measurement among eight projects as alternatives with regard to three criteria with four scales. The question was first written in English by the researcher before being translated into Nepali.


    As mentioned above, the weight of scales of the absolute measurement and the weight of the individual criterion at the Criteria level are found by a pairwise comparison. By comparing all the elements of the same level, the weight relative to each other can be estimated.Table 1

    Suppose that w1, w2, ···, wn are given, A = (aij) is the same as the below.


    where, aij = wi/wj (i, j = 1, 2, ···, n)

    The weights are perfectly consistent if aij·ajk = aik for all i, j, and k.

    When the Eq. (1) is multiplied by the vector w of order n, the new equation is formed such as the Eq. (2).

    Aw = λ w

    The Eq. (2) can be transformed into the Eq. (3),

    A λ I w = 0

    The Eq. (2) can be transformed into the Eq. (3),

    where, I is a unit matrix For this matrix, λ is called an eigenvalue of A and w is also said to be an eigenvector of A to make w ≠ 0. When λ is equal to n (λ = n), this matrix has a consistency. The sum of eigenvectors which become the final weights should be normalized into 1 (EQ =1) by numerical conversion of the estimated vector of weights such as the column of Weight in Table 2.

    Since the matrix for a pairwise comparison includes human judgments, the Consistency Index (CI) needs to be checked. In the case of Aw ≠ λmaxw and λmaxn, the difference between λmax and n indicates the inconsistency of the judgments. If λmax is equal to n (λmax = n), it means that the judgments have a consistency. CI can be calculated from (λmaxn)/(n – 1). To extend further, Consistency Ratio (CR) can be derived by dividing CI by the Random Index (RI). In this research, the value of RIs for pairwise comparisons of Criteria (n = 3) and the scales of the absolute measurement (n = 4) used 0.58 and 0.9 respectively. Satty, the developer of AHP, suggested a CR less than or equal to 0.1 is acceptable to be consistent (Satty, 1990). Eizo and Taka (2012) suggested that, when a CI exceeds 0.1 ~ 0.15, a pairwise comparison needs to be revised. If his suggestion will be applied to this study, a CR could be calculated into approximately 0.17 ~ 0.26 in terms of the values of CI. However, up to 0.2 of CR is allowed to be acceptable in this study. Table 2 shows the 104 respondents’ pairwise comparison matrixes and the weight of 3 criteria.

    Unlike the calculation of weights in Table 2, the weight of scales of an absolute measurement is normalized so that the maximum value among scales can be 1, as indicated in the Weight column in Table 3. The weights of Excellent, Good, Average and Poor were calculated into 1, 0.595, 0.343 and 0.204 respectively. Since CR is less than 0.1, this pairwise comparison can be accepted as reliable.

    These weights were used to indicate the intensity importance corresponding to each criterion of the 104 responses. All respondents’ weights under criteria by an absolute measurement are shown in Table 4. In the case of the economic impact criterion for the first respondent (No 1), he/she weighted the intensity importance of 8 different projects under that criterion as average, excellent, good, good, good, good, average, and excellent. These scales can be transformed into the numeric value matching the Weight column as listed in Table 3. The weights expressed by respondent No. 1 on the column of Economic Impact are listed in Table 4 such as 0.343, 1, 0.595, 0.595, 0.595, 0.595, 0.343, and 1. Other respondents were also analyzed as above.


    This research collected 104 questionnaires from 1 March to 5 March, 2014. Among them, 55 questionnaires with a CR greater than 0.2 were excluded and 49 questionnaires were accepted for the AHP analysis. The rejection rate of this survey reached 52.9% indicated in Table 5. Gender ratios were 56.7% male and 43.3% female in collected data and 46.9% male and 53.1% female in accepted data. The rejection rate of male responses was approximately twice as large as that for female responses. Age ratios in the collected data were 51.9% for respondents under 30 years of age, 39.4% for respondents between the ages of 31 and 50, and 8.7% for respondents over 51 years of age. Age ratios in the accepted data were 44.9% for respondents under 31 years of age, 44.9% for respondents between the ages of 31 and 50, and 14.3% for respondents over 51 years of ages. The highest and lowest rejection rates were recorded for the age groups under 30 and over 51, respectively. Employment was classified between three categories: No Job, Agriculture, and Others. Ratios were respectively 17.3%, 26.5% and 51.9% in the collected data and 18.4%, 26.5% and 55.1% in the accepted data. The rejection rates of respondents listing their employment as no job was the lowest, while that of those listing other as their employment was the highest. Education also was classified under three categories: No/Elementary, Middle/ High School, and College/college degree. Within the education category, respondents were identified as follows: No/Elementary (13.5% in the collected data, 16.3% in the accepted data), Middle/High School (42.3% in the collected data, 42.9% in the accepted data), and College/College degree (40.4% in the collected data, 40.8% in the accepted data). The rejection rate of respondents who have college/college degree was the highest, and the rejection rate of respondents with no education/elementary school was the lowest. The Period of Residence ratios were 24.0% of respondents reporting under 10 years, 13.5% of respondents listing a time frame of between 11 and 20 years, and 62.5% of respondents reporting residency of over 20 years in the collected data, and 30.6% of respondents for under 10 years, 18.4% of respondents for between 11 and 20 years, and 51.0% of respondents for over 21 years in the accepted data. The rejection rate of respondents over 21 years and between 11 and 20 years occupied the highest and lowest ratios, respectively. As a result from Table 5, the male respondents over 51 years old with no job and low education living in the study area for between 11 and 20 years have relatively the highest consistency in their answers on the survey.

    From the respondents’ pairwise comparison matrixes shown in Table 2, 49 matrixes less than or equal to 0.2 CR were selected and analyzed. These matrixes should aggregate one pairwise comparison matrix by the geometric mean. Table 6 indicates the aggregated pairwise comparison matrix and weights of 3 criteria. The weights derived from the aggregated pairwise comparison matrix were 0.326 for Economic Impact, 0.314 for Environmental Consideration, and 0.359 for Number of Beneficiaries. The priority of these 3 criteria can be decided based on these weights. 49 respondents consider Number of Beneficiaries the first priority among the 3 criteria.

    After selecting the 49 respondents’ weights from all respondents’ weights under criteria by the absolute measurement shown in Table 4, these weights also become the aggregated weight under each criterion by the geometric mean in Table 7. The weight of each project under the 3 criteria can be multiplied by the weight of each criterion indicated in Table 6 to generate a single weight of each project, such as the values in the Weight column in Table 7. These values were used to make a decision on a project priority. The calculation of the weight for the individual project is the same as below:


    From the results of Eq. (4) and Table 7, the education project was found to be the highest priority of the residents. Public health/Hhygiene/Sanitation ranked as the second priority and Income generation was the third priority.

    This research was designed both to find a project priority for residents in the study area and to suggest improvements of efficiency of the AHP method as a way to facilitate decision-making on project priority on the field of international rural development when supposing that a donor country has a plan to implement rural development projects. As illustrated by Table 7, residents considered the Number of Beneficiaries the first priority among criteria. This result means that donor countries need to focus on designing projects that benefit, without partiality, as many as residents of the site as possible. The residents’ opinions were reflected in their choices of a project priority. They selected education as the project that most matched their opinion. In addition, the top 3 priority projects chosen were Education, Public health/Hhygiene/Sanitation, and Income generation. If the donor country designs rural development projects for this area, it needs to promote these projects to meet residents’ needs. These results may not apply to other regions or recipient countries because they have different environments, economic conditions, priorities and cultural attributes. However, the AHP method is worth applying and proved to be useful in discovering residents’ priorities in spite of a somewhat high questionnaire rejection rate. Although this questionnaire was adjusted several times by pre-testing, using an absolute measurement at the alternatives level, and setting up the minimum criteria selection, it is important to devise other proper and complementary measures to reduce the rejection rate. Moreover, it was found that the concept of Criteria was complicated for some residents, especially uneducated respondents during the survey process. A more detailed explanation should be added to a survey questionnaire to help in understanding. It also seemed to be difficult for some respondents to answer the questionnaire on a pairwise comparison. Even though some words and sentences in English have clearer meanings than in Nepali, this survey was performed with only questionnaires in Nepali. Since some residents were ambiguous in seizing meanings, it is clear that the research would have benefited from both a Nepali translation faithful to the original English questionnaire as well as the simultaneous use of questionnaires in both languages. Using two questionnaires, one in English and one in Nepali, and allowing the residents to state a preference would be an effective method to get more accurate ideas.

    Normally, when donor countries research the residents’ needs in recipient countries, this research has been achieved through using the qualitative research method of Focus Group Discussion (FGD) with leaders of the recipient community. FGD has difficulty in preventing arbitrary decisions being made by some influential leaders taking an active hand in FGD. FGD also has the disadvantage of only depending on the opinions and decisions of a small number of community residents. Moreover, FGD cannot make a decision on a project priority based on an objective assessment. However, the projects derived from FGD can provide the basic information which is able to form the Alternative level in a hierarchy structure, as illustrated in Fig 2. FGD would be a valuable tool if it is used as a preliminary stage of survey creation for the quantitative method of AHP. A Mixed method research approach using FGD and AHP is expected to produce more accurate results that can promote the reliability of the research.

    Since this research was carried out in a foreign country, some problems were found in the process of survey creation and implementation. Coping with these issues became another problem confronting us. A series of these research studies would be a great contribution to the evolution of the academic field of international rural development. As the highest priority of rural development projects sufficiently reflecting the needs of residents through an improvement of the AHP method is promoted in the community of recipient countries, it will also contribute efficiently, effectively and practically to the economic, social and cultural development of those countries. Therefore, it is important to realize the importance of improving the AHP method because it will help guide the prioritization of projects to meet recipient needs.

    적 요

    해외 농촌개발의 계획수립부터 사업완료까지 과정은 의사결 정의 연속이며 수혜국에서 사업지역이 선정 되면, 농촌개발 프 로젝트 우선순위 결정을 위한 과정이 필요하다. 본 연구는 우 선순위 의사결정방법인 AHP를 네팔 극서부 지역의 다델두라 주민을 대상으로 적용하여 준비·실행·분석 과정에서 나타나 는 문제점 발견 및 개선방안을 제시하여 AHP 적용 효율성을 증대시키고자 수행하였다.

    AHP 계층구조는 우선순위 결정을 위한 3가지 평가기준과 8가지 대체요소(프로젝트)로 구성되었다. 평가기준는 쌍대비교 를 통해 우선순위를 결정하였고 대체요소(프로젝트)는 요소간 의 쌍대비교 경우의 수가 많아지고 일관성이 떨어지는 문제점 을 극복하기 위해 절대평가법을 적용하였다.

    분석결과, 우선순위 결정 평가기준으로 프로젝트를 통한 수 혜자 명수(주민수)를 가장 중요시하는 것으로 나타났으며 교 육, 공중보건/위생, 소득증대를 순서적으로 가장 선호하는 프 로젝트로 꼽았다.

    문제점으로는 여러 번의 사전 설문조사, 절대평가법의 적용, 최소한의 평가기준 설정을 통한 설문지보완이 이루어졌으나 주민들의 쌍대비교 설문에 대한 이해 부족으로 응답의 비일관 성 비율이 높아 분석에서 제외되는 설문지 비율이 높았다. 연 구 지역과 같은 저개발국가의 농촌지역에서 해외농촌개발 우 선순위 결정을 위한 AHP을 적용시 비슷한 현상이 발생할 것 으로 예상된다.

    AHP 적용 효율증대를 위한 세가지 방안을 제안한다. 첫째, 설문 전에 응답대상자에게 평가기준과 대체요소에 대한 충분 한 설명을 제공해 준다. 둘째, 영어가 가능한 대상자를 위해 영어와 현지어 설문지를 동시활용하여 응답대상자의 이해도를 높인다. 셋째, 설문지 구성 전에 지역단위 또는 마을단위의 지 도자를 소집하여 포커스 그룹 토의에 의한 질적 연구방법을 적용하여 현지의 실제적인 요구를 반영하는 설문지를 구성한 다. 이 방법을 통해 도출된 주민요구는 대체요소(프로젝트)를 구성할 수 있는 현실적인 정보를 제공하고 있어 AHP분석의 사전 단계로 활용하는 혼합방법은 분석의 신뢰성을 높여줄 것 으로 기대된다.



    Research Area.


    Hierarchy Structure.


    Fundamental Scale of Pairwise Comparison and Absolute Measurement.

    All Respondents’ Pairwise Comparison Matrixes and Weight of Criteria

    Pairwise Comparison and Weight of Scales of Absolute Measurement.

    All Respondents’ Weights under Criteria by Absolute Measurement.

    Information of Respondent’s Demographic Features.

    *Total Accepted Data means data having values of CR less than and equal to 0.2.
    **Rejection Rate (%) was calculated by Frequency of Collected Date−Accepted DateFrequency of Collected Date×100
    ***Other includes business, public officer/UN staff, teacher, student, NGO staff etc.
    ****Gross indicates the gross value classified by an individual demographic feature.

    Aggregated Pairwise Comparison Matrix and Weights of Criteria by Geometric Mean.

    Aggregated Weight under Criteria by Geometric Mean and Project Priority.


    1. District Development Profile of Nepal 2010/11 (2010) Mega Research Centre & Publication,
    2. Kwon Jae Hyun (2012) Strategic Making -Decision Method, AHP, Chung Ram Publishing Co., Ltd, [ Korean book title translated to English]
    3. Satty TL (1990) How to make a decision: The Analytic Hierarchy Process , European Journal of operational research, Vol.48; pp.9-26
    4. Vaidya OS , Kumar S (2006) Analytic hierarchy process: An overview of applications , European Journal of operational research, Vol.169 (1) ; pp.1-29