Introduction

This study is about coverage vaccination of Nigeria. After cleaning data, we start by doing univariate analysis, bivariate analysis, partial correlation, multivariate analysis than logisitc regression. We use some test collerationshep withe two variables such as chi2. Some findings are given in the report.

Please see atached word file, Excel file for the results and do file stata

Please see attached the stata15 do file

1. Data cleaning plan

The dataset must cleaning before all calculation of indicators and before interpreting it. So, I should know all my variables and find all errors:

-number of cases or observations by counting; storage variable name; storage type; variable label, etc.

• check all duplicates cases in terms of all variables. I drop for all duplicates cases if they exist
• check duplicates cases in terms of keys variables (Stratum ID, Cluster number, Household number, Child's line number, region ID, etc.) for merge dataset.
• check the missing value with tabulating all variables, e.g. sex 1. M 2. F, I shouldn’t have missing value for it
• check the range of all variables, e.g. age of children: 12-23 months
• check the outlier for the variables with
• Skip patterns or respecting for all shift

2. Weighting plan

Generally, for complex surveys, data should be weighted to improve the estimation of indicators. There are four steps for calculating weights .(section 6.2 manuel WHO and annex J).

• Step1- Sampling weights

The fist step in calculating weights is to calculate the probability with which each respondent was selected into the survey sample. It is the inverse of the probability of selection.

Sampling weight for Respondent i=1/Probability Respondant i was selected into the sample.

In a one-stage cluster survey, this figure is related solely to the probability that the cluster has been selected. If the cluster needs to be segmented, or if it is a multi-stage cluster sampling design, then the probability will equal the product of the probability of selection at each stage.

Probability Respondent I was Selected= (Stage1 Probability) (Stage2 Probability) …

• Step 2- Interviewing respondents within a household

All eligible should be taking account in the calculation of the weight, so the probability of selection for an individual is equal to the probability of selection for his or her household.

• Step 3- Adjusting for non-response

In case of missing data, it is necessary to use collected information and reports on non-respondents and in function of this information that we will make the adjustment

• Step 4 Post-stratifiation to re-scale survey weights

Survey bases are often obsolete or include cluster size estimates based on the total population rather than the eligible population (for example, all residents rather than only the 12 to 23 months), so that the sum of the weights will in general be not equal to the size of the total eligible population from which the results of the survey will be generalized.

3. Table shells

See attached Excel file

4. Completed Table(s)

see attached Excel file

5. Table-filling syntax

see attached do file

6. Graphic Example(s):variable: valid_penta3_age1_card

See attached with Excel file

7. Methods & Results & Strengths & Limitations Text

The dataset concerned Nigeria and we have 12 state (Adamawa,Bauchi,Borno,Gombe,Taraba,Yobe,Akwa Ibom,Bayelsa,Cross River,Delta,Edo and Rivers).devided into two zones (North East,South East). We have 1728 children age 12-23 months in dattset.

To conduct this study, we use STATA15. Before calculation of indicators, data cleaning is conducted by using simple tools such as tabulation of variables, by checking range of variables, checking missing value, checking duplicates cases, etc. Some results are coping in Excel file. We calcule crude coverage penta1 and penta3 then we calculate drop out between penta1 and penta3 by using the VCQI formula.

Comparing the result by zones, noticed that South East zone seems high than North East zone. There is some discrepency between defferent states. The graph show this discrepency in the Excel file.

The use of probabilist sampling (WHO manual 2018) is one of the strenghting of the study because indicators have confidence intervals and weignted. But we don't have or conducted health faciliy or visit for register checking if home based record is not avalaible at the household.

8. Keys finding

-there is a relationship between educationnal leval of the caretaker or mather and practice of vaccination,

-There is a relationship between education and practice of vaccination. The result above shows tha more the level education is high more the caretaker vaccine the child.For exemple the educational level of caretaker is Higher the rate or proportion of children vaccinated for PENTA3 is 79.45%. The seem for BCG trend and OPV0,(cf word file)

-There is a significant relationship between zone (urban or rural) and practice of vaccination. The results below show the correlation between the variables. The children habit in zone urban is more vaccinated then their homolgue habit in rural (cf word file)

-No significant relationship between sex of caretaker and practice of PENTA3 (at level 5%) (cf word file)

-There is a significant relationship between the place « variable : Place of immunization (PENTA3 BGC, OPV0) too far » and immunization. All results are significant even at the level 1% (cf word file).

-

-No obvious significant relationship between age category of the caretaker and vaccination PENTA3

-there is a significant relationship between the variable Government hospital and the immunization (PENTA3,BGC, OPV0). We exclud here the missing value of the variable (im0ba) (cf word file).

-There is a relationship between welfare or Wealth index quintile (MICS_5_windex5) and vaccination or immunization (PENTA3,BGC and OPV0).More the walfare increases more the proportion of the chidren vaccinated becomes high. We notice that all test (Chi2 and Fisher exact) are valid. The correlation between PENTA3 and welfare index is significant at the level 5% as the resultat bellow shows. We note the seem trend for the BGC and OPV0 (cf word file).

-There is a partial correlation between PENTA3 and the following varaiables :

• caretaker_education (Highest level of education attended)
• MICS_5_windex5(Wealth index quintile)
• im0ba (Vaccinated at Government hospital)
•  im0bb (Vaccinated at Government health centre)
• im0be (Vaccinated at Campaigns / Supplementary immunization activities)