African Cigarette Prices 2016-2020
Africa, 2016 - 2020
Research Unit on the Economics of Excisable Products
Created on
March 18, 2021
Last modified
March 18, 2021
Page views
28427
Download
1799
Metadata
xml (ddi standard)
json
Identification
IDNO
afr-reep-acp-2016-2020-v1.4
Title
African Cigarette Prices 2016-2020
Country
| Name |
|---|
| Africa |
Study type
Price Survey [hh/prc]
Series Information
Data on Alcohol and Tobacco in Africa Project. African Cigarette Prices 2016-2020. [dataset]. Version 1.4. Cape Town: Research Unit on the Economics of Excisable Products [producer], 2021.Cape Town: DataFirst [distributor], 2021. DOI: https://doi.org/10.25828/nvz2-ah77
Abstract
The Data on Alcohol and Tobacco in Africa (DATA) Project is based at the University of Cape Town. The DATA project collected prices of cigarettes sold at retail outlets and from street vendors in a number of African countries. The aim is to create a retail cigarette price database to enable researchers to make estimations about price differences in these countries, between brands, in urban and rural settlements, and over time.
Kind of Data
Sample survey data
Unit of Analysis
Other
Version
Version Description
v1.4: Edited, anonymised and serialized data for public distribution
Version Date
2021-03-17
Version Notes
Version 1.2: All previous rounds and datafiles of the ACP were harmonized and combined into a single data series. Version 1.2 covered rounds 1-7. Prior to version 1.2 rounds were released separately.
Version 1.3: Round 8 was added. It was undertaken between June and August 2019.
Version 1.4: Round 9 was added. It was undertaken between November 2019 and Feburary 2020.
Version 1.3: Round 8 was added. It was undertaken between June and August 2019.
Version 1.4: Round 9 was added. It was undertaken between November 2019 and Feburary 2020.
Scope
Notes
The study collected data on prices of cigarettes from retail outlets and street vendors in several African countries.
Coverage
Geographic Coverage
The countries included in the study are:
Botswana, Chad, Eswatini, Etiopia, Ghana, Kenya, Lesotho, Madagascar, Malawi, Mauritius, Mozambique, Namibia, Nigeria, South Africa, Swaziland, Tanzania, Uganda, Zambia, Zimbabwe
Botswana, Chad, Eswatini, Etiopia, Ghana, Kenya, Lesotho, Madagascar, Malawi, Mauritius, Mozambique, Namibia, Nigeria, South Africa, Swaziland, Tanzania, Uganda, Zambia, Zimbabwe
Geographic Unit
The data was collected at the level of retailer (often incliuding GPS coordinates), but the lowest level of geographic detail available in the public use dataset is Suburb.
Universe
The study covered cigarette prices of retailers (including street vendors) in selected African countries.
Producers and sponsors
Primary investigators
| Name | Affiliation |
|---|---|
| Research Unit on the Economics of Excisable Products | University of Cape Town |
Funding Agency/Sponsor
| Name | Abbreviation | Role |
|---|---|---|
| International Development Research Centre | IDRC | Funder |
Data Collection
Dates of Data Collection
| Start | End | Cycle |
|---|---|---|
| 2016 | 2020 | All Rounds |
Data Collection Mode
Computer Assisted Personal Interview [capi]
Data Collection Notes
For this project, the Data on Alcohol and Tobacco in Africa (DATA) Project organised for University of Cape Town students from different African countries to visit retail outlets and street vendors in their countries, where cigarettes were sold, and obtained permission to take photographs of cigarette prices, using their cellphones.
Fieldworkers working in the same areas were put in contact to avoid duplication of effort. As such, it is reasonable to assume that duplication was largely avoided and that a store being interviewed twice in one round occured quite rarely if at all. Students were somewhat self-governing in their choice of areas to visit and stores to gather data from. How the student groups covering an area divided up that area has not been well documented. As such, there may be georgraphic bias, for example against poorer areas that are more difficult to interview. This also means that a given interviewer might repeat whatever bias/mistakes they tend to make over all the rounds (if they remained on the project) making it hard to detect.
In some cases the same store (retailer/spaza shop/street vendor) was interviewed in more than one round of data collection.This means it is possible (in theory) for an analyst to create a panel of cigarette vendors. However, this requires using GPS locations which are only available in the secure (disclosive) verison of the data.
Fieldworkers working in the same areas were put in contact to avoid duplication of effort. As such, it is reasonable to assume that duplication was largely avoided and that a store being interviewed twice in one round occured quite rarely if at all. Students were somewhat self-governing in their choice of areas to visit and stores to gather data from. How the student groups covering an area divided up that area has not been well documented. As such, there may be georgraphic bias, for example against poorer areas that are more difficult to interview. This also means that a given interviewer might repeat whatever bias/mistakes they tend to make over all the rounds (if they remained on the project) making it hard to detect.
In some cases the same store (retailer/spaza shop/street vendor) was interviewed in more than one round of data collection.This means it is possible (in theory) for an analyst to create a panel of cigarette vendors. However, this requires using GPS locations which are only available in the secure (disclosive) verison of the data.
Data Collectors
| Name | Abbreviation | Affiliation |
|---|---|---|
| Research Unit on the Economics of Excisable Products | REEP | University of Cape Town |
| Economics of Tobacco Control Project (prior) | ETCP | University of Cape Town |
Data Processing
Data Editing
Data collected by REEP was in excel format and not serialized. Datafirst has serialized and cleaned the data into a research-ready series.
Appending the rounds required harnomizing variable names which was straightfoward.
The brand names (eg Marborough) and sub-brands (eg gold) required significant cleaning due to typos and errors in data collection.
Some observations with bad values (around 20) were dropped from the final dataset.
Appending the rounds required harnomizing variable names which was straightfoward.
The brand names (eg Marborough) and sub-brands (eg gold) required significant cleaning due to typos and errors in data collection.
Some observations with bad values (around 20) were dropped from the final dataset.
Data Appraisal
Data Quality Notes
Store IDs. Store IDs were created based on what is known about data collection design. Specifically, that each fieldworker took one photo of each store they enumerated per round. Fieldworkers in the same area were put in contact; making the assumption that they managed to avoid any overlap, stores can be uniquely identified by their fieldworker-photoID-round. This is how the StoreIDs in version 1.3. were created. There are two consequences of this. Firstly, if the same store was visited twice in a round, it will be recorded as two separete stores. Secondly, this design unfortunately means that the same phyicsal store will get different IDs in each round of data collection. One can only indentify stores across rounds in the disclosive version of the dataset using GPS information.
StoreIDs in Round 1. It has been noted that some of the PhotoIDs in round 1 did not map 1-to-1 with the phyiscal stores. By example, two different enumerators used the same photoID to record photos of two different stores. As such, the storeIDs for round 1 are not relaible. An analyist would need to look at store name and suburb to get a better idea of how to idenitfy distinct phyiscal stores in round 1.
There are other types of tobacco products still in the data. The analyst can use the variable "Product" to remove such non-cigratte products (eg cigars). For the same reason, the analyst may also want to check the fieldworker's comments. Brand and sub-brands including "King Size" might affect analysis as this is a bigger size of cigarette and thus should cost more.
It is slightly unclear how to treat the sub-brand "filter" or "filters". In general a filter cigarette is a standard (filterless are very rare these days), and many data collectors may have opted not to include the word "filter" as it is implied. If the sub-brand is captured as "filter 20" this generally just means a standard pack of 20. Despite this, the data cleaning policy was not to remove the word "filter" but rather leave it up to the analyst. The recommendation would be for the analyst to consider these as being equivalent to no sub-brand. For example "Camel Filters" is really the same thing as "Camel". However, in the case of specfic flavours, such as "Camel - Blue Ice", this is not the standard Camel cigarrette but a particular flavour. If filter/filters appeared here, the policy was to remove the word.
StoreIDs in Round 1. It has been noted that some of the PhotoIDs in round 1 did not map 1-to-1 with the phyiscal stores. By example, two different enumerators used the same photoID to record photos of two different stores. As such, the storeIDs for round 1 are not relaible. An analyist would need to look at store name and suburb to get a better idea of how to idenitfy distinct phyiscal stores in round 1.
There are other types of tobacco products still in the data. The analyst can use the variable "Product" to remove such non-cigratte products (eg cigars). For the same reason, the analyst may also want to check the fieldworker's comments. Brand and sub-brands including "King Size" might affect analysis as this is a bigger size of cigarette and thus should cost more.
It is slightly unclear how to treat the sub-brand "filter" or "filters". In general a filter cigarette is a standard (filterless are very rare these days), and many data collectors may have opted not to include the word "filter" as it is implied. If the sub-brand is captured as "filter 20" this generally just means a standard pack of 20. Despite this, the data cleaning policy was not to remove the word "filter" but rather leave it up to the analyst. The recommendation would be for the analyst to consider these as being equivalent to no sub-brand. For example "Camel Filters" is really the same thing as "Camel". However, in the case of specfic flavours, such as "Camel - Blue Ice", this is not the standard Camel cigarrette but a particular flavour. If filter/filters appeared here, the policy was to remove the word.
Access policy
Access authority
| Name | Affiliation | URL | |
|---|---|---|---|
| DataFirst | University of Cape Town | support@data1st.org | support.data1st.org |
Contacts
| Name | Affiliation | URL | |
|---|---|---|---|
| DataFirst Support | University of Cape Town | support@data1st.org | www.support.data1st.org |
Access conditions
Public use files, available to all
Citation requirements
Data on Alcohol and Tobacco in Africa Project. African Cigarette Prices 2016-2020. [dataset]. Version 1.4. Cape Town: Economics of Tobacco Control Project [producer], 2021. Cape Town: DataFirst [distributor], 2021. DOI: https://doi.org/10.25828/nvz2-ah77
Metadata production
Producers
| Name | Affiliation | Role |
|---|---|---|
| DataFirst | University of Cape Town | Metadata producer |
Date of Metadata Production
2021
DDI Document version
Version 4