CSP Endline Bi-Weekly Report
Lucy Right
September 02, 2024
Survey Completion
So far, 1059 surveys have been completed. There are 3426 surveys to go!
| Oblast | Aug 26 | Aug 27 | Aug 28 | Aug 29 | Aug 30 | Aug 31 | Sep 01 | Total This Week | Total to Date | Female |
|---|---|---|---|---|---|---|---|---|---|---|
| Batken | 20 | 12 | 12 | 43 | 41 | 43 | 19 | 190 | 396 | 51% |
| Jalal-Abad | 6 | 4 | 1 | 0 | 1 | 2 | 0 | 14 | 114 | 56% |
| Osh | 45 | 44 | 40 | 30 | 20 | 6 | 19 | 204 | 549 | 49% |
Village-Level Completion and Village Issues
Update on village-level completion:
- Data collection has been finished in 9 villages, but there are issues with 5 villages. Please check below!
- Data collection is ongoing in 38 villages
- Data collection has not yet started in 65 villages
| Oblast | AA | Village | Baseline | Endline | Duplicated ID | Gender Matches Baseline | Note |
|---|---|---|---|---|---|---|---|
| Баткен | Ak-Turpak | Tokoy | 9 | 9 | 0 | Yes | Data collection complete |
| Баткен | Birlik | Bel | 20 | 20 | 0 | Yes | Data collection complete |
| Баткен | Leylek | Kara-Suu | 29 | 31 | 0 | No | Too many surveys in this village AND gender balance incorrect. Please check. |
| Баткен | Moldo Niyaz | Shybran | 8 | 10 | 0 | Yes | Too many surveys in this village. Please check. |
| Баткен | Sulukta-city | Kolco | 16 | 16 | 2 | Yes | Data collection complete but duplicate IDs. Please check. |
| Баткен | Tort-Gul | Ak-Otok | 10 | 10 | 0 | Yes | Data collection complete |
| Ош | AA named after Toktomat Zulpueva | Ak-Chabuu | 17 | 17 | 2 | Yes | Data collection complete but duplicate IDs. Please check. |
| Ош | Kenesh | Ak-Terek | 10 | 10 | 0 | Yes | Data collection complete |
| Ош | Papan | Papan | 17 | 24 | 3 | No | Too many surveys in this village AND gender balance incorrect AND duplicate IDs. Please check. |
UniqueID Errors
Please refer to file uniqueID_issues_20240902.csv for the complete list of surveys with Unique ID errors.
- There are 80 entries with duplicated unique IDs. Some of these may be duplicated submissions, some might be incorrect IDs.
- The teams in Osh have the most duplicates at 47!
- There are 8 entries with a Unique ID not found in the baseline data.
Attrition and Recontacting
So far, the overall recontact rate for households is 69%, and the recontact rate for individuals is 61%.
| Respondent Sex | Week 1 | Week 2 | Week 1 | Week 2 |
|---|---|---|---|---|
| Female | 69% | 71% | 61% | 63% |
| Male | 70% | 68% | 62% | 58% |
| Total | 69% | 70% | 62% | 60% |
| Oblast | Recontact Rate (HH) | Recontact Rate (Resp) | Replaced Wrong Gender | Replaced Too Young |
|---|---|---|---|---|
| Batken | 68% | 58% | 5 | 0 |
| Jalal-Abad | 63% | 46% | 17 | 0 |
| Osh | 72% | 67% | 22 | 0 |
| Total | 69% | 61% | 44 | 0 |
Poor Replacements
The following enumerators replaced the baseline respondent with someone of the incorrect gender.
| ono | fi | Week 1 | Week 2 |
|---|---|---|---|
| Batken | Altynai_B | 1 | 2 |
| Osh | Nazgul_P | 1 | 2 |
| Batken | Zhiidegul_M | 0 | 1 |
| Osh | Dilbara_A | 2 | 1 |
| Osh | Tynychbek_A | 1 | 1 |
| Batken | Omurbek_A | 1 | 0 |
| Jalal-Abad | Aibolot_K | 1 | 0 |
| Jalal-Abad | Aisha_M | 3 | 0 |
| Jalal-Abad | Baiel_A | 4 | 0 |
| Jalal-Abad | Beksultan_J | 2 | 0 |
| Jalal-Abad | Erlan_M | 3 | 0 |
| Jalal-Abad | Kadyrali_J | 2 | 0 |
| Jalal-Abad | Nursultan_T | 2 | 0 |
| Osh | Azamat_T | 3 | 0 |
| Osh | Elmira_B | 3 | 0 |
| Osh | Indira_K | 3 | 0 |
| Osh | Kunduz_K | 1 | 0 |
| Osh | Карабаев | 2 | 0 |
| Osh | Сайфудинов Алихан | 2 | 0 |
Replacements by Enumerator
| Oblast | Enumerator | Surveys | Recontact Rate (HH) | Recontact Rate (Resp) | Bad Replacement |
|---|---|---|---|---|---|
| Batken | Cholpon_K | 30 | 40% | 23% | 0 |
| Jalal-Abad | Aibolot_K | 9 | 67% | 33% | 1 |
| Jalal-Abad | Nursultan_T | 3 | 33% | 33% | 2 |
| Batken | Busadat_S | 26 | 35% | 35% | 0 |
| Batken | Turgunai_S | 11 | 36% | 36% | 0 |
| Jalal-Abad | Aisha_M | 27 | 48% | 37% | 3 |
| Batken | Omurbek_A | 8 | 62% | 38% | 1 |
| Jalal-Abad | Kadyrali_J | 25 | 64% | 44% | 2 |
| Osh | Kunduz_K | 59 | 47% | 44% | 1 |
Time Stamps
Time Statistics
| Section | Surveys | Median | 25th Percentile | 75th Percentile |
|---|---|---|---|---|
| time_sec_a | 1058 | 0.2 | 0.2 | 0.6 |
| time_sec_b | 1058 | 5.3 | 4.0 | 7.4 |
| time_sec_c | 1058 | 1.2 | 0.8 | 1.6 |
| time_sec_conjoint | 1057 | 0.7 | 0.4 | 1.3 |
| time_sec_d | 1057 | 11.6 | 8.5 | 15.5 |
| time_sec_d1_d26 | 1058 | 4.1 | 3.0 | 5.4 |
| time_sec_d30_d60 | 1058 | 5.5 | 4.2 | 7.5 |
| time_sec_d61_d74 | 1057 | 1.2 | 0.3 | 2.4 |
| time_sec_e | 1057 | 11.2 | 8.3 | 15.1 |
| time_sec_f | 1058 | 3.6 | 2.5 | 5.0 |
| time_sec_g | 1058 | 4.9 | 3.5 | 7.1 |
| time_sec_g1_g14 | 1058 | 2.3 | 1.5 | 3.3 |
| time_sec_g15_g31 | 1058 | 2.4 | 1.7 | 3.7 |
| time_sec_h | 1058 | 3.5 | 2.7 | 5.2 |
| time_sec_h1_h9 | 1058 | 1.1 | 0.7 | 1.6 |
| time_sec_h10_h40 | 1058 | 3.5 | 2.7 | 5.2 |
| time_sec_i | 1058 | 1.2 | 0.8 | 1.9 |
| time_sec_k | 1032 | 0.9 | 0.7 | 1.6 |
| time_sec_list | 1057 | 0.2 | 0.1 | 0.5 |
| time_sec_total | 1058 | 46.9 | 36.5 | 59.2 |
Change in Duration Over Time
Duration by Enumerator
The table below shows median section times for enumerators with surveys with shorter than average median survey times.
| Enumerator | ono | Surveys | Median | Median: Sec B | Median: Sec C | Median: Sec D Electricity | Median: Sec D Oth Services | Median: Sec D CSP | Median: Sec E | Median: Sec E Conjoint | Median: Sec F | Median: Sec G COVID | Median: Sec G Climate | Median: Sec H Gender | Median: Sec H Disability | Median: Sec I | Sum |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Erlan_M | Jalal-Abad | 3 | 25 | 4 | 1 | 2 | 3 | 0 | 5 | 0 | 2 | 2 | 1 | 0 | 2 | 1 | 23 |
| Tynychbek_A | Osh | 32 | 25 | 4 | 1 | 3 | 2 | 0 | 6 | 0 | 2 | 1 | 1 | 1 | 2 | 1 | 24 |
| Azamat_T | Osh | 11 | 30 | 4 | 1 | 3 | 4 | 0 | 8 | 0 | 3 | 1 | 1 | 1 | 2 | 1 | 29 |
| Baktybek_M | Batken | 17 | 32 | 4 | 1 | 2 | 3 | 2 | 6 | 0 | 2 | 2 | 2 | 1 | 3 | 1 | 29 |
| Nazgul_P | Osh | 27 | 34 | 4 | 1 | 3 | 4 | 0 | 8 | 1 | 2 | 2 | 2 | 1 | 3 | 1 | 32 |
| Kadyrali_J | Jalal-Abad | 4 | 36 | 3 | 1 | 3 | 4 | 0 | 10 | 1 | 3 | 2 | 3 | 2 | 3 | 1 | 36 |
| Busadat_S | Batken | 13 | 37 | 6 | 1 | 2 | 2 | 1 | 4 | 0 | 2 | 1 | 1 | 0 | 2 | 1 | 23 |
| Dilbara_A | Osh | 2 | 41 | 7 | 1 | 7 | 6 | 3 | 16 | 0 | 7 | 2 | 2 | 1 | 3 | -15 | 40 |
| Gulmira_Im | Batken | 24 | 41 | 10 | 1 | 3 | 5 | 1 | 7 | 0 | 2 | 1 | 1 | 1 | 2 | 1 | 35 |
| Aisha_M | Jalal-Abad | 6 | 43 | 4 | 1 | 4 | 5 | 1 | 11 | 1 | 4 | 2 | 2 | 2 | 4 | 1 | 42 |
| Omurbek_A | Batken | 5 | 45 | 4 | 1 | 3 | 4 | 2 | 13 | 0 | 5 | 2 | 2 | 1 | 3 | 1 | 41 |
| Kunduz_K | Osh | 28 | 46 | 7 | 1 | 3 | 5 | 0 | 12 | 1 | 4 | 2 | 2 | 1 | 3 | 1 | 42 |
| Zhiidegul_M | Batken | 15 | 46 | 5 | 1 | 3 | 5 | 2 | 14 | 1 | 3 | 2 | 2 | 1 | 3 | 1 | 43 |
Answer Quality
Don’t Knows
Statistics:
- The median survey has 7 “don’t know” answers.
- The 75th percentile survey has 15 “don’t know” answers.
- The median enumerator has 8 “don’t know” answers on average.
The enumerators below have recorded a very large number of Don’t Knows on many surveys! We expect some respondents will say don’t know, but enumerators should not consistently report more Don’t Knows than other enumerators!
| Enumerator | ono | Surveys | Surveys > 75 perc DKs | Median | Max |
|---|---|---|---|---|---|
| Altynai_B | Batken | 22 | 9 | 14 | 31 |
| Gulmira_Im | Batken | 24 | 9 | 14 | 32 |
| Minavar_T | Batken | 23 | 6 | 11 | 30 |
| Ainur_T | Batken | 18 | 3 | 4 | 49 |
Agree / Disagree
This table shows the average HHI, by enumerator, of the responses to 25 agreement-style questions in the survey. The HHI is a metric that indicates the ‘diversification’ in responses to the agreement questions.
An average HHI of 1 indicates that the enumerator records the same answer for all 25 agreement questions on every survey they enumerated. As the average HHI approaches 0, it indicates that the enumerator tends to record the same number of responses at each level of agreement for each survey.
Average HHIs above 0.70 indicate the enumerator is very often recording the same answer. Average HHIs below 0.3 indicate the enumerator is recording a lot of different options.
| Oblast | Enumerator | Surveys | Avg HHI | Avg % Str. Agr | Avg % Agr | Avg % Neither | Avg % Dis | Avg % Str. Dis | Avg % DK |
|---|---|---|---|---|---|---|---|---|---|
| Batken | Gulburak_T | 7 | 0.810 | 0.042 | 0.887 | 0.042 | 0.024 | 0.012 | 0.000 |
| Jalal-Abad | Erlan_M | 3 | 0.797 | 0.014 | 0.886 | 0.014 | 0.085 | 0.029 | 0.000 |
| Osh | Dilbara_A | 2 | 0.726 | 0.108 | 0.827 | 0.065 | 0.000 | 0.000 | 0.000 |
| Jalal-Abad | Kadyrali_J | 4 | 0.708 | 0.829 | 0.032 | 0.011 | 0.043 | 0.085 | 0.032 |
| Osh | Indira_K | 8 | 0.696 | 0.076 | 0.804 | 0.005 | 0.125 | 0.000 | 0.000 |
| Osh | Tattybubu_M | 21 | 0.684 | 0.023 | 0.794 | 0.006 | 0.202 | 0.000 | 0.000 |
| Osh | Adilet_S | 5 | 0.681 | 0.042 | 0.792 | 0.000 | 0.167 | 0.000 | 0.008 |
Summary of problematic interviews
How problematic interviews are determined:
- Duration : the number of sections with time stamps below 25 percentile. The following sections are assessed: B, D1 (electricity), D2 (other services), E, F, G1 (covid), G2 (climate change), H2 (disability), Total survey (sections B-I). Since survey time can sometimes be unreliable if the respondent goes back to the beginning of the survey, I look at the number of sections that are very short - not just the total survey time.
- Don’t Knows : how many don’t know responses are in the survey
- Similarity in Question Response : how similar answers are to agree/disagree questions (for example, if the enumerator selects ``somewhat agree” in 95% A/D questions)
Please check the file issues_20240902.csv to see all problematic interviews. Below is just a summary by oblast.
| Date | Batken | Jalal-Abad | Osh |
|---|---|---|---|
| Aug 20 | 1 | 5 | 7 |
| Aug 21 | 0 | 5 | 12 |
| Aug 22 | 4 | 4 | 3 |
| Aug 23 | 0 | 1 | 3 |
| Aug 24 | 0 | 1 | 2 |
| Aug 25 | 0 | 2 | 3 |
| Aug 26 | 0 | 1 | 3 |
| Aug 27 | 2 | 0 | 4 |
| Aug 28 | 0 | 0 | 5 |
| Aug 29 | 3 | 0 | 2 |
| Aug 30 | 3 | 0 | 4 |
| Aug 31 | 1 | 1 | 0 |
| Sep 01 | 2 | 0 | 3 |
| Total | 16 | 20 | 51 |
Summary of enumerator performance
The following table shows indicators for enumerator performance. You can sort the table to see how enumerators perform.
Please pay special attention to the following enumerators: Tynychbek_A, Nazgul_P, Gulmira_Im, Busadat_S!!