Methodology for NBS National Corruption Survey 2017


Friday, August 18, 2017 1:45PM / NBS

Part 1 –National Survey on the Quality and Integrity of Public Services

Introduction - background
Data presented in this report were collected in the National Survey on the Quality and Integrity of Public Services, otherwise known as the Nigerian Corruption Survey, a project funded by the European Union and implemented by the UNODC in collaboration with the National Bureau of Statistics of Nigeria (NBS).

The Nigerian Corruption Survey was designed as a large-scale household survey, representative at the level of the Nigerian states, with the aim of collecting baseline information

  • on direct experiences of corruption events, as victims, by citizens
  • on the experience of reporting corruption and other crimes to the public authorities
  • on opinions and perceptions of citizens concerning recent trends, patterns and state responses to corruption

Evidence derived from the survey is used to calculate benchmark indicators that can be used to inform relevant policies and track future progress, while ensuring international comparability with surveys of similar nature carried out in other countries.

Survey contents and instruments
Following previous international experiences and extensive consultations with all major stakeholders in a nationwide corruption survey, the survey instrument was structured into several parts that best addressed the information requirements identified: starting with a part required to randomly select a household member and a section on attitudes and opinions on corruption, the main part of the questionnaire addresses experiences with public services and in particular experiences of bribery. The subsequent parts of the questionnaire deal with bribery in the private sector, corruption in employment and the awareness of and opinions about specific anti-corruption agencies.[1] 

In building the survey instrument, much time and effort was invested in adapting a standard survey methodology already tested and refined in other national contexts (such as Afghanistan, Iraq and the western Balkans) to the specific Nigerian context and to ensure national ownership of the survey results. From the early stages of the survey development, roundtable discussions were held with the National Steering Committee (NSC) of the survey – consisting of 14 beneficiary agencies of the European Union-funded UNODC-led Support to Anti-Corruption Project in Nigeria, including all the major anti-corruption agencies in Nigeria ― for overall guidance on survey implementation.  

In addition, several working group meetings were held  with the National Technical Committee (NTC) for technical guidance in the implementation of the survey, and for consultation and inputs on successive versions of the draft survey instrument. The NTC was composed of a sub-set of the NSC and included UNODC, the Inter-Agency Task Team (IATT), Action Aid Nigeria, the CLEEN Foundation and NBS as implementing partner (IATT in turn includes other beneficiary organizations within the anticorruption sector of Nigeria). 

In addition to national consultations with stakeholders, extensive qualitative research on patterns and modalities of corruption in Nigeria was carried out to further the understanding of patterns, forms and manifestations of corruption in different parts of the country. [2] Taken together, the experience from other national corruption surveys, the national consultations and the in-depth qualitative research contributed to developing a survey instrument that best captured the complex reality of bribery in Nigeria.

Survey preparation
To test the adequacy of the survey instrument and the integrity of survey operations and logistics before the main survey, several preparatory activities were conducted. In particular, the two most important activities for testing the survey instrument included

  • cognitive testing
  • a pilot survey

Cognitive testing
To explore the adequacy of question formulations in the Nigerian multi-cultural and multi-lingual context, a cognitive testing was conducted on 40 respondents recruited from two different Nigerian states (20 from Lagos and 20 from Kaduna state) in order to examine the ways in which respondents interpret, consider and ultimately answer the questions as well as to identify potential response errors. One-on-one in-depth interviews were used to test the questions with respondents. The interviews were conducted on the four steps in the cognitive process of responding to questions: understanding the question and response categories, recalling or searching for the requested information, thinking about the answer and making a judgment about what to report, and reporting the answer.

Pilot survey

A pilot survey is the process of testing all stages of data production that is carried out during the main survey, but on a smaller scale. It helps to perfect all survey instruments as well as to ensure adequate planning for all logistics that will be required for the main survey. Using two slightly different survey instruments that tested alternative question formulations, the pilot survey was carried out in Oyo, Kwara, Delta and Katsina states with a total sample size of 480 households (around 120 in each state). The results of the pilot surveys were documented in a technical report and used to improve survey procedures, question formulations (especially where two different versions were tested), instructions for interviewers in the interviewer guidelines, sampling procedures applied by interviewers and other technical issues. 

Main survey

Sample design and weighting procedure
For the main survey, a total of 33,067 interviews were conducted across the 37 states with persons aged 18 and older.[3] The sampling methodology adopted is a stratified random sampling, with a stratification process conducted at the state level. Stratified random sampling is a sampling method that requires dividing the population into smaller groups (or strata) and drawing a randomized sample within each stratum. In this case, the strata are represented by the 37 Nigerian states (36 states and the Federal Capital Territory) and within each state, the (adult) population was extracted randomly.  

In each of the 37 states, between 779 and 900 interviews were conducted. To draw the sample, the population in each state is divided into Enumeration Areas (EA).[4] From this process, a total of 23,280 EAs (4 per cent of the overall total) were randomly selected for the sample frame. From this frame, in a first stage sampling procedure (selection of EAs) around 60 EAs were selected in each of the 37 states to be canvassed. The EAs will be selected such that they will cut across both urban and rural areas. A total of 2,220 EAs were selected throughout the country.  

For the Second Stage sampling, which is the household (HH) selection, 15 HHs were systematically selected per EA, resulting in a total of around 900 HHs per state. In each household, the interviewer then randomly selected one individual respondent. The interviewer was required to return to the same HH for at least 3 visits (i.e., three call backs) if the household member could not be located  before the respondent was treated as a missing/non-responding unit. The overall response rate was 85 per cent. Households or respondents were not replaced on the basis of refusal. Rather, allowance for non-response was built into the estimation of sample size. However, sample EAs or households were replaced in case of inability to locate them or if they “moved away” recently. 

To make survey results at the state level representative at the national level, for the analysis of the data after the survey the sample size was adjusted (“weighted”) for the size of the population in each state. Weighting ensures that the distribution of the sample across states reflects the actual distribution of the population. The latest population data were provided by the National Bureau of Statistics in 2016 and refer to 2015.[5] 

Weights calculation:

s = 1,…, S  Nigerian states (including the Federal Capital Territory)

Ns = sample size for state s

Ps = (adult) population of state s

The weight (W) for the state s-th is defined as the ratio between the (adult) population and the sample size in the same state.

In the database, such weighting factor is included in order to align all the findings to actual population figures.

Quality control and questionnaire translation
Considering the large scope of the study and the nature of extended and sensitive field work, an external quality assurance monitoring of survey operations was contracted. Following a public tender procedure Practical Sampling International (PSI), a Lagos-based company specialized in marketing, social and opinion research, was tasked with the responsibility of ensuring quality of survey activities and survey outputs at all stages of the survey implementation.  

Based on the outcome of the quality control report of the pilot survey, a number of quality control measures were further adopted in order to ensure reliability of the data collection, such as the translation of the questionnaire into the local language, updating the household listing frame to include extra households to serve as possible replacement for non-effective interviews and additional logic and consistency checks on the electronic device adopted for fieldwork (Computer Assisted Personal Interview device).  

In order to ensure that the survey questionnaire appropriately addressed the research objective, the research instrument was checked to ensure the questions were arranged in logical sequence, to check that the routing and interview instructions were in logical order and that the questions were consistent with the research objectives. 

Before the main survey, the questionnaire was translated into Pidgin English and the three main languages spoken in the geo-political zones of Nigeria, namely Yoruba, Igbo and Hausa. The translations were checked to ensure that the questions were correctly interpreted. To ensure the quality of the translations, relevant parts of the translated questionnaires were also translated back to the original English. 

Training and field work
The first level of preparation ― the Training of Trainers (TOT) ― included Trainers, Monitors and Coordinators and was held in Abuja. Participants were staff of NBS, UNODC and PSI. The second level training, or zonal training, was held in 12 states, each of the 6 geo-political zones duly represented. During this training, a day apiece was fully dedicated to training on Paper and Pencil Interviewing and the Computer-Assisted Personal Interview device, to ensure all field interviewers had a full understanding of all the modules in the questionnaire.  

Another day was used for field practice to prepare interviewers for the full exercise at various designated clusters.  For the field work, two teams of five people each (four interviewers and one supervisor) were constituted in each state, for a total of 74 teams with 296 interviewers and 74 supervisors. In addition to supervisors, two field monitors per state and one coordinator per zone ensured proper conduct of interviews and quality control. Field teams completed each EA before moving to the next EA to enable close contact and feedback from the supervisor.  

In addition, PSI deployed its own team of 74 quality control officers (two per state) plus six coordinators and three managers, who provided an additional layer of supervision and quality monitoring. For quality control, PSI conducted 3,391 back checks (interviews are repeated and compared) and 1,719 accompanying interviews (PSI observed the interview in situ), corresponding to 10 per cent and 5 per cent of all interviews, respectively. In its comprehensive final quality report, PSI stated that generally there were no major issues that could have derailed the data-gathering process.

Language versions
Interviews were conducted in five main languages (English, Pidgin English, Yoruba, Igbo and Hausa), with additional language requirements covered by qualified interviewers in their native languages. The following map shows the main languages in which the interviews were conducted by zone.

Data entry and data cleaning
In most cases, the Computer Assisted Personal Interview with a hand-held device was used to capture the data during the interview. Consequently, immediately after the interview, or during the debriefing of interviewers by the supervisor at the end of the working day, data were automatically uploaded to the central NBS Data Processing Centre in Abuja.

In a first stage, the raw data were processed and cleaned for obvious errors by qualified NBS staff. In a second stage, the data were transformed to SPSS and sent to UNODC Vienna for further analysis and processing. Detailed cleaning of inconsistent data points and obvious outliers resulting from erroneous data entries was performed and a final data set prepared for analysis by UNODC, before data were analysed for tabulation and report writing.

Part 2 – Key Indicators
In this report, three key indicators are used to understand the extent to which bribery is affecting the lives of Nigerians

  1. prevalence of bribery
  2. frequency of bribery (or average number of bribes paid)
  3. average and total annual amount paid in bribes

All formulas described below have been multiplied by the weighting factor (W) illustrated above in order to align all the findings to actual population figures. 

I. Prevalence of bribery
The estimate of the prevalence of bribery (P) is calculated as the number of adult Nigerians who had at least one contact with a public official and who paid a bribe to a public official, or were asked for a bribe by a public official (B), on at least one occasion in the 12 months prior to the survey, as a percentage of all adult Nigerians who had at least one contact with a public official (C).

In other words, letting

i = 1, …., N  Nigerian adult (in the sample)

c =  1 ,…., C  Nigerian adult who had at least one contact with a public official in the 12 months in the 12 months prior to the survey

b = 1, ….., B  Nigerian adult who had at least one contact with a public official and who paid, or was asked for a bribe by a public official in the 12 months prior to the survey

k = 1, …., K  Nigerian adult who had at least one contact with a public official and who paid a bribe by a public official in the 12 months prior to the survey

j = 1, ….., J  Public official receiving the bribe

with N C B K 

The prevalence of bribery is obtained as:

This definition of the prevalence of bribery follows the formula/is valid for the calculation of the SDG indicator 16.5.1. “Proportion of persons who had at least one contact with a public official and who paid a bribe to a public official, or were asked for a bribe by those public officials, during the previous 12 months”. 

II. Frequency of bribery
The frequency of bribery (F) (also called “average number of bribes paid” in the report), is calculated as the average number of times Nigerian bribe-payers paid a bribe in the 12 months prior to the survey. 

Vol. 1 and Vol.2 PDF Reports Here

Related News
Sobering Facts from 2016 Bribery in Nigeria Report
Global Economic Crime and Fraud Survey
Corruption in Nigeria - Bribery as Experienced by the Population
Skye Bank Plc Denies Concealing TSA Funds
US SEC Bars Lawyer Who Committed Fraud
BREAKING: False Asset Declaration Charge: Saraki Discharged, Acquitted
125,790 Cases of Reported Offences Recorded in 2016 - NBS
1MDB-Related Transactions - Financial Penalties Imposed On Credit Suisse & UOB by MAS

Related News