Home » Countries » Kenya » Issues with Education Data in DHS Dataset for Kenya (2022 and 2014)
| Issues with Education Data in DHS Dataset for Kenya (2022 and 2014) [message #29862] |
Wed, 14 August 2024 09:45  |
n.borgmann
Messages: 10
Registered: August 2024
|
Member |
|
|
Dear DHS Team,
I am reaching out with a question regarding the DHS dataset for Kenya, specifically about the variables V107 (years of schooling) and V106 (education level) for the years 2022 and 2014.
In the 2022 dataset, there seem to be only a very small number of individuals with more than 8 years of schooling--just 7 individuals. Considering that Kenya's education system typically includes both primary and secondary schooling, which should exceed 8 years, I am puzzled by this finding.
I have seen discussions in the forum suggesting that there were issues with this variable in previous datasets and that the revised dataset will be released by the end of this week. However, the dataset for 2014 was referenced as being accurate. Upon reviewing the 2014 data, I found that the maximum value for years of schooling is also 8 years. This seems inconsistent with the data for variable V106, where many individuals fall into the third category, which is supposed to be higher than secondary education.
Could you please provide an explanation for these results? Is there a specific reason for this discrepancy, or could the data potentially be incorrect? Understanding this would be crucial for my work.
Thank you very much for your time and assistance. I greatly appreciate your help.
Best regards,
Nils
[Updated on: Wed, 14 August 2024 09:45] Report message to a moderator |
|
|
|
|
|
|
|
|
|
|
|
| Re: Issues with Education Data in DHS Dataset for Kenya (2022 and 2014) [message #30603 is a reply to message #30594] |
Mon, 06 January 2025 09:45  |
Bridgette-DHS
Messages: 3230
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS staff member, Tom Pullum:
This is a very interesting observation, but before trying to understand it, I first need to know how you got the estimates of the medians.
Did you use hv108 in the PR file? If so, did you have lower or upper cutoff ages? Tables 2.13.1-2 in the final report on the 2014 survey, for example, include ages 6+, with no upper age cutoff. Or did you use v133 in the IR file, which would apply to age 15-49 (for women) and mv133 in the MR file, which would apply to age 15-54 (for men)? Did you modify the upper age limit for men to match that of women?
I can suggest a strategy to understand this pattern, which I agree is unexpected, if you will clarify how you got the numbers.
[Updated on: Mon, 06 January 2025 09:46] Report message to a moderator |
|
|
|
|
|
| Re: Issues with Education Data in DHS Dataset for Kenya (2022 and 2014) [message #30716 is a reply to message #30604] |
Fri, 24 January 2025 09:27 |
Bridgette-DHS
Messages: 3230
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS staff member, Tom Pullum:
I apologize for the long delay in this response. I have been able to confirm that these medians were calculated from hv108 (in the PR file) in both surveys. hv108 is constructed hv106 (highest level attended) and hv107 (years attended at that level). In the following table, using the 2014 survey, hv106 is the column variable, hv107 is the row variable, and hv108 is the constructed variable, in each cell for the total number of years of schooling (ignore the totals row and the totals column).
. tab hv107 hv106, summarize(hv108) means
Means of education completed in single years
highest |
year of | highest educational level
education | attained
completed | primary secondary higher | Total
-----------+---------------------------------+----------
0 | 0 8 12 | 3.5867717
1 | 1 9 13 | 4.831788
2 | 2 10 14 | 6.9003889
3 | 3 11 15 | 6.5741577
4 | 4 12 16 | 9.3242178
5 | 5 13 17 | 5.2350961
6 | 6 14 18 | 6.3245768
7 | 7 . 19 | 7.0337612
8 | 8 . 20 | 8.0081755
9 | . . 21 | 21
10 | . . 22 | 22
don't know | 98 98 98 | 98
-----------+---------------------------------+----------
Total | 4.8757119 10.761662 14.517886 | 6.9356705
Here is the corresponding table for the 2022 survey:
. tab hv107 hv106, summarize(hv108) means
Means of education completed in single years
highest |
year of | highest educational level
education | attained
completed | primary secondary higher | Total
-----------+---------------------------------+----------
0 | 0 8 12 | 5.6417091
1 | 1 9 13 | 5.6580055
2 | 2 10 14 | 7.7659006
3 | 3 11 15 | 8.8481961
4 | 4 12 16 | 10.427846
5 | 5 13 17 | 5.2305875
6 | 6 14 18 | 6.6000514
7 | 7 . 19 | 7.0442526
8 | 8 . 20 | 8.0260387
9 | . . 21 | 21
10 | . . 22 | 22
11 | . . 23 | 23
12 | . . 24 | 24
don't know | 98 98 98 | 98
-----------+---------------------------------+----------
Total | 5.3123704 11.025853 15.361057 | 8.1386855
There were some changes between 2014 and 2022, which added 23 and 24 years to the distribution, but only a total of 9 people in the 2022 survey have those values so that is not an issue. The "don't know" cases were omitted from the calculations.
I believe that what you see for median years of schooling in the top quintile could be understood by disaggregating birth cohorts of women and men, or equivalently age groups. It's an interesting question, as I said, but unfortunately I have not been able to find time to do the kind of disaggregation that I just described. Perhaps other users can make suggestions or follow up.
|
|
|
|
Goto Forum:
Current Time: Fri Jan 31 22:43:49 Coordinated Universal Time 2025
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
