October 2009
AFRICAN COMMISSION ON AGRICULTURAL STATISTICS
Twenty-first Session
Accra, Ghana, 28 – 31 October 2009
Price Data Collection Using Personal Digital Assistant (PDA)-EthiopiaIntroduction
Monthly retail price data collection of urban areas first started in the capital city of the country, Addis Ababa, together with the Household Income, Consumption and Expenditure Survey of 1963 with the objective of producing consumer price index (CPI). However, due to the ever increasing demand for the data, the scope and coverage of price survey has increased stage by stage. Hence, the geographical coverage of the survey was expanded to cover all regional capitals to be included but the CPI was limited only to Addis Ababa until 1997.
The retail price survey of the rural areas was launched together with the Rural Integrated Household Survey Program (RIHSP) in 1980/81. As part of the RIHSP, the price survey was carried out on quarterly basis starting from May 1981 to July 1987 and on monthly basis since September 1987. Between 1986/87 and 1996/97 the survey has been conducted in about 760 EAs. In the year 1997/98 the numbers of EAs to be covered by the survey were substantially increased to 1420. Considering the substantial number of EA's to be covered and thereby huge amount of data to be collected, the processing and timely publishing of price data were not manageable. After conducting an intensive exercise on the quality of the data, the CSA decided to reduce the sample size substantially without affecting the quality of the resulting Consumer Price Index (CPI). Consequently, the numbers of rural EAs were cut down from 1420 to 446 EAs starting from September 1998 producing national, urban and rural level CPIs.
Moreover, further improvement has been made starting from July 2001 on the number of market outlets to be covered for this exercise. As a result, the price survey data collection had been restricted and basically focused on a market based survey approach by selecting 119 representative market outlets. The shift to more representative markets was implemented with the aim of achieving acceptable coverage of urban markets for each Region by dropping many remote rural markets. As a result, unnecessary time lag in the production of the CPIs has been achieved with eleven regional and country level indices.
It is to be noted that the retail price survey, covers the collection of prices of major agricultural and industrial goods including food, drinks, drink constituents, tobacco, clothing and footwear, building materials, energy, household equipment, medical care, transportation, education, recreation, farm equipment, hotel services (food and accommodation) and other goods and service obtained from various outlets, Hence, about 400 items are covered by the survey monthly.
Collecting Retail Price Data Using PDA
The CSA has been collecting and disseminating price data using the traditional means of data collection i.e. questionnaire based system where field enumerators use to collect the price data. The CSA conducts two types of price surveys, namely; Monthly Retail Price of Goods and Services and Monthly Producers’ Prices of Agricultural Products. With these surveys CPI and Producers’ Price Indices (PPI) are compiled every month.
The CSA has deployed PDAs (Personal Digital Assistants) to facilitate the price data collection and electronic transfer from the field to its head office. Personal Digital Assistant (PDA) is a type of palm held mini-computer that has a complete set of computer key board functions. The data is entered in to the PDA by means of a touch screen using a specially designed touch pen for the purpose. Files are created, saved and retrieved like any other computer. The questionnaire of price data format has been pre-loaded on the PDA. It was possible even to load a local font that enables the enumerator to retrieve the questionnaire written in Amharic (local language). Excel PDA version software was customized for the application in price statistics, including data entry form, min-max validation tools to reduce data entry errors and introduction of Amharic fonts. A PDA user guide has been fully developed in English and translated into Amharic. A total of 138 PDAs are being deployed (including back-up devices to replace those requiring maintenance).
During the pilot exercise both the traditional hard copy questionnaire and PDA were used simultaneously. The data collected on PDA and the hard copy questionnaires were compared to check whether the entry errors were tolerable and enumerators could handle it effectively. The result of the comparison revealed some minor problems that need be corrected before applying the PDA as a standard data collection tool. Three comparative tests of data entry on PDAs by enumerators and desk top computers by data entry clerk have shown that data quality between different systems is comparable. The tests showed that data quality improves when managed closer to data generation in the field.
The PDA system satisfies both the statistical standards’ requirements for CPI and PPI calculation and also the requirements of timeliness for price data users. The full deployment of PDAs to 119 retail price markets in all 25 CSA branch offices is expected to reduce data processing time by three weeks. In the first implementation phase price data from five CSA branch offices has been transferred to head office within two working days after collection. Full deployment of PDAs could make CSA’s price data be available on-line within the same month in which data are collected.
As such, the traditional paper based data collection system has been fully replaced by a digital system. In order to further reduce data processing time an automated excel spreadsheet has been developed. This facilitates market price data cleaning and further reduces the time from data collection to data dissemination on the web. A manual on the utilization of the automated spreadsheet has been developed for CSA subject matter specialists. Data quality is a critical issue to meet the statistical objectives for PPI and CPI calculation.
Recently CSA’s price publications are becoming timely due to the saved time and we are expecting a dramatic reduction of the time needed for the dissemination of price data once all market places are covered by the PDA. It is to be noted that if properly used, the PDA is much more advantageous than the traditional paper based questionnaires in terms of cost, timeliness and the quality of data.
It envisaged that this experience will also be extended to other data collection activities of the CSA in the future. In this regard, countries like Brazil, Mexico and Cape Verde have used the PDAs for other surveys including for collecting data for their Population and Housing Censuses. It is also planned that the latest version of PDA called Ultra-mobile computers will be used.
Furthermore, since all the basic edits like range checks can be developed in the system, it gives an opportunity to make all the necessary corrections right in the field while the enumerator is doing data collection. This in turn has a greater advantage in reducing the non sampling error compared with the paper based data collection where by such a non sampling errors have to be dealt with other imputation techniques.
Advantages of Using PDA
· It drastically reduces time needed for collection, analysis, and dissemination of data since the time used for data entry and editing is automatically saved.
· The PDA and efficient computer network between CSA and its branch offices will revolutionize the data transmission process and saves a lot of time, manpower and money.
· A lot of paper work such as duplicating thousands of pages of questionnaires will become history once questionnaires are directly loaded on the PDA and sent to all data collectors. The data collector need not carry a questionnaire and a manual since these are already saved on the PDA. He can open the files and refer to them any time.
· Minimizing the non sampling errors
Some Challenges in Using PDA
· Currently, the PDAs need to be charged after every two hours of work. In places where electricity is not available it may pose a problem to work for longer hours in a day.
· The other challenge was to obtain solar powered PDA in which the data collectors need to carry extra batteries to continue the day’s work.
· Since the technology is imported and is not a local material, availability of PDA could be a problem unless a good stock is purchased.
· The enumerator’s capability in terms of modest mathematics and computer skills is very important.
Price Data Dissemination
Besides posting monthly CPI and PPI on its website, the CSA has made available price database online. This will help data users to have access to any particular commodity price data dating back from 1997 in a given market in any particular year and month. This online system, we believe, provides an easy access to a time series price data collected by the CSA.
This system runs on an SQL based database and is made available online through CSA’s official website (www.csa.gov.et). Through its simple search facility, a user can easily get a time series data on prices of commodities within and across years, market places or specific months of interest.
The online retail price database is currently presented as three surveys.
• Survey-1 lists price data for the year 1989-1993 Ethiopian Calendar
• Survey-2 lists price data for the year 1994-1999 Ethiopian Calendar
• Survey-3 lists price data for the year 2000-onward Ethiopian Calendar
To obtain retail price data from the database follow the following steps.
1. Brows to www.csa.gov.et/consumerprice
2. Select one of the surveys (figure 1)
Figure 1: Retail price website screenshots of theprice survey selection
3.Select the Year, Month, item and market from the next screen (figure 2). This screen has two selection mechanisms.
Simple selection allows you to select a single year and month. Multiple items and market can be selected.
Advanced selection allows you to make multiple selections with all the criteria. Markets and items are categorized for ease of access
Figure 2: Screen shots illustrating a query of white Teff in 1 market of Addis (Addisu Gebeya) for all months covering the years 2001 and 2002 EC
4. Click ‘submit’ and the data that meet the search criteria will be displayed (figure 3). The data screen also allows you to sort the result by any column. Click on the column headers to sort the data.
Figure 3: Screen shots illustrating the result of the query on white Teff in 1 market of Addis (Addisu Gebeya) for all months covering the years 2001 and 2002 EC. The data queried includes 3 price quotations and mean price.