Kidanemariam G.egziabher Mekelle University
The opinion and ideas expressed in this correspondence are entirely of the author, and should not be attributed in any manner to the Bureau of Finance and Planning of the Tigray regional state or Mekelle University.
I came across an article posted in the aigaforum, under the title My reflection on Tigray Poverty Status Reports and Early Warning” By Hailai Abera Weldeslassie posted on 2018/10/21.
In the first place, I would like to appreciate Hailai for his concern, follow-up and reflection to what is going-on in the development of Tigray and trying to contribute to this end. I wish the criticism Hailai has tried to establish was based on evidence and concrete ground level knowledge. Hailai in his brief article has touched so many things, but my focus will confine regarding his comment targeted at the estimated poverty level of the region. Before explaining the overall methodology and research protocols the study has followed, I would like to raise the following questions to Hailai:
1. When was the study that you mentioned that come-up with REGIONAL poverty status of 39-45% (headcount ratio) for Tigray Region was conducted? Save a study conducted by Nega et al., (2011). Baseline Socioeconomic Survey Report of Tigray Region, Mekelle; which reported a regional poverty level of 41%, just 7 years ago. 2. Which institution conducted that study (if any)? 3. Which areas of the region were covered by study what sample sizes were used?
If there are some other studies there with such high poverty level (HPL) of 39-45%, then issue becomes why the current study with a modest poverty level (MPL) of 29% headcount ratio, differ from, and see if there is serious methodological fallacy in either of the studies. Unfortunately, nothing has been shared in support of the HPL or concrete technical argument to prove the MPL has such and such mythological fawns.
Having said this as introductory remark let me explain the
methodologies adopted both in the survey administration and poverty
1.2. Population and sampling - frame The population of this study was all households in all woredas and urban administrations of Tigray regional state. The survey was designed to consider tabias/kebelle as primary sampling unit and households as secondary sampling unit. The sampling frame for the primary sampling unit was all lists of tabias/ kebelle in rural woredas and all lists of kebelles in urban administrations of the region. The sampling frame for the secondary sampling unit was the lists of all households in the selected tabias/kebelle of the region.
1.3. Sample size determination - The survey sample size was determined by using a single population proportion formula, with 95% confidence interval, 1% level of precision, design effect of 1.5 and 10% non-response rate. Accordingly, the minimum sample size for each woreda was set to be 229 households (including 21 households for non response rate). As there are 52 woredas in Tigray region, the total sample survey for the region was 11,908 households. To the best of my knowledge there is no any other study with such huge sample size and a survey designed administered scientifically (CSA and Mekelle University joint work).
1.4. Sampling technique - A multistage sampling technique was employed to select the 11,908 households using systematic random sampling method. See the Figure below. To select households from rural woredas, a multistage random sampling strategy was employed. The rural woredas were divided into two categories namely with tabias and infant towns. Further, tabias were stratified/categorized in to three agro-ecological zone (Kola, Dega, and weyna Dega). Using simple random sampling two tabias were selected from each agroecologically zones. A representative sample size of 265 tabias and 58 infant towns were selected randomly from rural woredas and 114 infant towns, respectively. The primary sampling units (PSUs) were tabias and infant towns. At the second stage, the required number of households per selected PSUs was randomly chosen. Households were selected from the selected 187 tabias and 34 infant towns based on systematic random sampling (SRS) technique proportion to the size of population (PSP). Thus, total sample of 7,165 households were selected from rural woredas. The list of households was obtained from tabia administration office.
Similar approach was followed to select households in urban areas as well.
1.5 Involvement of the Federal Central Statistics Agency - (CSA) CSA is a federal agency established with a proclamation no.442/2005, mandated to oversee and ensure data collection, compilation, and analysis activities are up to the required standard and can be used for planning, monitoring social and economic development programs in the country. Pursuant to that mandate CSA has actively participated from inception to the final stage of the study (survey result presentation). So share us your evidence and then our discussion becomes a productive engagement.
2. Poverty estimation
Consistent with the Federal Government’s, method of measuring poverty and poverty line we used the consumption approach and the cost of basic needs approach to determine a regional poverty line. The estimation methodologies as well as the result were presented at regional workshop; the concern (especially from CSA) was the other way round instead.