WEB SURVEYS FOR OFFLINE RURAL COMMUNITIES Prepared by Beatrice W. Gichohi, Chief Manager Research Planning and Development, Tana and Athi Rivers Development Authority, Nairobi, Kenya [email protected] Tana and Athi Rivers Development Authority (TARDA) •Is a statutory corporation in Kenya with a mandate for regional development, organized around Tana and Athi Rivers Basin. •Tana and Athi Rivers Basins comprise of 19 counties (out of the 47 counties in Kenya) •According to 2009 Census report Kenya 14.3 ( 37%) of Kenya's population of about 38.6 million people lived in the region •Of these ,about 11 million (78%) lived in the rural areas while about 3.3 million lived in Nairobi Outline • • • • • • • Introduction Survey Design Methodology Benefits Challenges Conclusion Thinking Future Introduction • TARDA has been formulating a Regional Development Plan for Tana and Athi Rivers basin. • This is a 20 year plan which identifies and prioritizes strategies, programs and projects to be implemented in the region for sustainable equitable development and improved livelihoods. • The plan has involved resource mapping as well as identification of development needs and priorities of the communities. Introduction • In March 2014, a Household survey was conducted to evaluate the perception of the community on Resources in the region and their potential , Levels of resource utilization Socio- economic Challenges development priorities • A total of 3,400 households were selected for the study • A questionnaire was used to collect data • This paper looks at how the questionnaire was administered online. Internet Access and Mobile phone Usage in Kenya According to the communications commission of Kenya, (CCK) 20112012 Report, • Internet usage via traditional users grew by 19.2% • Kenya has about 7 million users of internet - About 72% of Kenyans lack internet access • 29.7 million mobile phone users- 11% lack mobile telephone access • Disparity in access to high speed internet remains a large obstacle • Majority of internet users are located in Nairobi. Lack of access to internet and mobile telephones has mostly affected communities in the rural areas. Enumerators in a training session in in a previous data collection exercise in Kajiado In the past TARDA used conventional methods of pen and paper data collection methods. Characterized by •High cost of administering •A lot of time required to collect , enter , clean and analyze data •A lot of errors in data especially due to missing and incomplete data •Other logistical problems There was need to leverage on technology and devise a method which would minimize the above problems Survey Design • The data collection method designed was a hybrid of Web surveys and the enumerator assisted conventional field data collection. Methodology- Data Collection Tools Online questionnaire • A questionnaire with 120 variables ( Closed ended and open ended questions) was created and uploaded to the organizations website. • The questionnaire was organized by the six thematic areas of study ;Economic, social cultural , demographic, Physical features, infrastructure and human settlement Mobile phones; 3G enabled mobile phones were used to access and respond to the online survey Methodology - Data collection Organization • Tana and Athi Rivers Basin TANA & ATHI RIVER BASINS • 3 Catchments • 1 Coordinators Per Catchment CATCHMENT COUNTY • 17 COUNTIES • 4 enumerators per county • Each enumerator was in charge of a county ward Methodology -Sampling • A directory obtained from the Kenya Population and Housing census 2009 was used as the sampling frame. • 17 counties studied. From each 200 house holds would be selected. • Multistage Random sampling was used to select the households • Simple random Sampling was used to select four wards per county • From each ward a cluster sample of 50 household’s was selected. Each cluster was a sub village. A total of 3,400 households were selected for the survey. Methodology- Data Collection • To qualify as an enumerator a person had to Be a resident of the County ward have a 3G enabled mobile phone have at least secondary school level of education. • The catchment coordinators trained the community enumerators during a one day training session. Methodology- Data Collection •Bulk internet bundles was bought and distributed to the which were distributed to the enumerators. •Each of the enumerators would visit the respondents at a pre arranged convenient location (at home, on their farms, shopping center, and market). •Respondents with previous exposure to web applications would respond with minimum assistance •Where there was no access to internet, an offline copy which had been saved in the phone would be used •Each enumerator would complete an average of 15 questionnaires each day •Internet allocation was 100 MB per day/ Enumerator . Data collection at a shopping Center in Machakos Methodology- Data Cleaning • Back in the office, the catchment coordinators were tracking the responses as they came in. • Data from completed questionnaires would be downloaded at regular intervals and subjected to cleaning. • The enumerators would be informed of any errors identified and would verify accuracy of data with respondents • The data collection exercise took five days within which all data had been collected, downloaded and cleaned. Benefits Cost effective; • Cost of web survey was about -KES 300,000 ($3,500) • Initial budget of about- KES 1,000,000 ($11,500) for data collection and entry conventional methods. • Minimized Sampling Error ; Web surveys are characterized by big sampling errors. However this being a hybrid of the conventional survey methods and web surveys probability methods of sampling were incorporated. Clean Data; eliminated errors resulting from data entry. The coordinators were also able to clean the data in their offices as data collection was going on in the field. • High Response rate. Out of the Targeted 3,400 households, 3,284 responded translating to a response rate of 96.5%. • Respondent Guidance; Respondents were guided in taking the Web survey. This ensured that everybody regardless of their levels of education had the same interpretation of the questions. Quick Results; It took four days to collect, enter and clean data from 3,200 respondents. Challenges • Data Security; with rapid growth of incidents of cyber crime in Kenya, Some respondents were uncomfortable with their responses being uploaded on line. • Internet Access; The initial assumption that internet would be accessible everywhere, any time was proved wrong. Enumerators in a few areas were unable to access internet. They however were able to use the offline version of the survey which they would later upload • Mobile Phone Battery; Some mobile phones batteries could not last the whole day. This posed a problem in some rural areas where there was no access to power • Literacy levels and previous exposure; Literacy levels vary greatly with from one county to the other. Questionnaires should be designed to fit the varied literacy levels. There was a significant association between Respondent’s Levels of Education and their catchments of residence Conclusion • This was a hybrid of the web survey and the conventional enumerator assisted data collection method . • The Hybrid benefited from the strengths of each of these two methods • offered solutions to the initial skepticism of using the web survey which revolved around issues of low literacy levels, and internet access. And Thinking Future • Kenya’s population is ready and enthusiastic about adoption of technology • In future, with more exposure to mobile phone technology and increasing access to internet, communities in Kenya will be able to respond to self administered online surveys. • Issues of sampling, internet access and ability of the respondents to self administer a questionnaire should be considered.