EXAMINING NONRESPOSE OF LABOR FORCE SURVEY AND FAMILY

EXAMINING NONRESPOSE OF LABOR FORCE SURVEY AND FAMILY INCOME AND EXPENDITURE SURVEY IN THE PHILIPPINES by Maribel L. Bernardo, Hiroshige Furuta Presented by Maribel L. Bernardo (PSA MIMAROPA) 13 th National Convention on Statistics October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City

OUTLINE OF PRESENTATION I. III. IV. Introduction Methodology Findings of the Study Recommendations 13 th National Convention on Statistics October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City

I. Introduction • Nonresponse is defined as failure to get all of the desired information on the sample households in a sample survey. • Nonresponse is treated in the estimation procedures of surveys by having the weighting adjustment due to nonresponse. • Applying weighting adjustment to compensate for nonresponse is likely to increase the variance as well as bias of the estimates derived from the survey. 13 th National Convention on Statistics October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City

I. Introduction • minimize nonresponse cases to prevent reduction in the number of elements on which the estimates for analyses are based. • know the characteristics of nonrespondents so that appropriate survey methodology can be devised according to the subject matter • characteristics of nonrespondents are usually not available because they are not interviewed 13 th National Convention on Statistics October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City

I. Introduction • uses the 2009 Family Income and Expenditure Survey (2009 FIES) data characterization of the nonrespondents. • FIES is a rider survey of LFS. It is done in two separate visits, visit 1, in July round and visit 2, in January round of survey. The characteristics of nonrespondents were taken from successful interview of visit 1 or visit 2. • nonrespondents for the purpose of this study are defined as those households who refused to be interviewed during 2009 FIES field enumeration. For the households where outcome of interview is ‘refusal’, the characteristics of the household heads are used to characterize the household. 13 th National Convention on Statistics October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City

I. Introduction Outcome of Interview Status Code Description 1 2 3 4 5 Completed interview Refusal Temporarily away/Not at home/On vacation Vacant housing unit Housing unit demolished, destroyed by fire, typhoon, etc. 6 Others, such sample household moved-out in the housing unit, housing unit converted into establishment 7 Critical area, flooded 13 th National Convention on Statistics October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City

I. Introduction Outcome of Interview Classification of Respondents Eligible Ineligible Interview Status Code 1 -Completed 2 -Refusal 3 -Temporarily away 7 -Critical area/flooded 4 -Vacant 5 -H. U. demolished 6 -Hhld moved-out 13 th National Convention on Statistics October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City

I. Introduction Based on MS 2003 Sample Size Reason for Adjustment of Sample Size 43, 882 Initial 46, 961 Based on October 2000 LFS response rate of 93. 4% 51, 000 2003 Listing of Households 13 th National Convention on Statistics October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City

I. Introduction Nonresponse statistics Table 1. doc Proportion to Total Sample of Successful and Unsuccessful Interview of LFS: July 2003 - January 2010 Figure 1. xlsx Total Eligible Sample Households and Response Rates of LFS Table 2. xls MS 2003 Monitoring and the Replicate Used in Survey Table 3. doc Total Sample and its Percent Distribution by Outcome of Interview: FIES 2009 Table 4. doc Combination of Interview Status Code of July 2009 and January 2010 FIES 13 th National Convention on Statistics October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City

I. Introduction Table 1. Proportion to Total Sample of Successful and Unsuccessful Interview of LFS: July 2003 to January 2010 Total Households Reference Period Sample Eligible 1/ Responding Nonresponding of which code 4, 5, 6 2, 3, 7 (1) (2) (3) (4) (5) (6) (7) 2/3/ July 2003 50, 922 46, 729 45, 242 5, 680 4, 193 1, 487 October 2003 3/ 52, 207 47, 106 45, 242 6, 965 5, 101 1, 864 January 2004 2/3/ 52, 207 46, 691 44, 862 7, 345 5, 516 1, 829 3/ April 2004 52, 207 46, 349 44, 249 7, 958 5, 858 2, 100 July 2004 3/ 52, 207 45, 675 43, 041 9, 166 6, 532 2, 634 October 2004 52, 207 45, 482 43, 310 8, 897 6, 725 2, 172 January 2005 52, 207 45, 574 43, 744 8, 463 6, 633 1, 830 April 2005 47, 767 45, 701 43, 668 4, 099 2, 066 2, 033 July 2005 47, 062 44, 922 42, 875 4, 187 2, 140 2, 047 October 2005 45, 869 44, 471 42, 332 3, 537 1, 398 2, 139 January 2006 46, 512 44, 372 42, 273 4, 239 2, 140 2, 099 April 2006 46, 892 44, 762 42, 253 4, 639 2, 130 2, 509 July 2006 2/ 47, 018 43, 780 41, 013 6, 005 3, 238 2, 767 October 2006 46, 552 43, 872 41, 581 4, 971 2, 680 2, 291 2/ January 2007 45, 497 42, 928 40, 689 4, 808 2, 569 2, 239 April 2007 53, 379 44, 115 41, 776 11, 603 9, 264 2, 339 July 2007 49, 551 43, 096 40, 318 9, 233 6, 455 2, 778 October 2007 46, 311 44, 097 41, 890 4, 421 2, 214 2, 207 January 2008 56, 414 44, 410 42, 161 14, 253 12, 004 2, 249 April 2008 53, 653 43, 488 41, 228 12, 425 10, 165 2, 260 July 2008 49, 829 43, 016 40, 662 9, 167 6, 813 2, 354 October 2008 50, 027 43, 822 41, 955 8, 072 6, 205 1, 867 January 2009 50, 027 43, 804 42, 063 7, 964 6, 223 1, 741 April 2009 49, 754 43, 634 41, 818 7, 936 6, 120 1, 816 July 2009 2/ 52, 233 42, 813 41, 044 11, 189 9, 420 1, 769 October 2009 50, 224 43, 870 42, 049 8, 175 6, 354 1, 821 January 2010 2/ 48, 441 42, 282 40, 619 7, 822 6, 159 1, 663 13 th National Convention on Statistics Notes: 1/ - Eligible respondents are those with interview result codes 1 , 2 , 3 , and 7. October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City 2/ - Labor Force Survey (LFS) has rider survey Family Income and Expenditure Survey (FIES) Response Rate col 4/col 3 (8) 96. 8 96. 0 96. 2 95. 5 94. 2 95. 2 96. 0 95. 6 95. 4 95. 2 95. 3 94. 4 93. 7 94. 8 94. 7 93. 6 95. 0 94. 9 94. 8 94. 5 95. 7 96. 0 95. 8 95. 9 95. 8 96. 1 Proportion to Total Sample Household of Successful Unsuccessfu whichco Interview l Interview de 2, 3, 7 (9) (10) (11) 88. 8 11. 2 2. 9 86. 7 13. 3 3. 6 85. 9 14. 1 3. 5 84. 8 15. 2 4. 0 82. 4 17. 6 5. 0 83. 0 17. 0 4. 2 83. 8 16. 2 3. 5 91. 4 8. 6 4. 3 91. 1 8. 9 4. 3 92. 3 7. 7 4. 7 90. 9 9. 1 4. 5 90. 1 9. 9 5. 4 87. 2 12. 8 5. 9 89. 3 10. 7 4. 9 89. 4 10. 6 4. 9 78. 3 21. 7 4. 4 81. 4 18. 6 5. 6 90. 5 9. 5 4. 8 74. 7 25. 3 4. 0 76. 8 23. 2 4. 2 81. 6 18. 4 4. 7 83. 9 16. 1 3. 7 84. 1 15. 9 3. 5 84. 0 16. 0 3. 6 78. 6 21. 4 3. 4 83. 7 16. 3 3. 6 94. 7 16. 1 3. 4

I. Introduction Figure 1. Total Eligible Sample Households and Response Rates of LFS 13 th National Convention on Statistics October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City

I. Introduction Table 2. 2003 Master Sample Survey Monitoring and the Replicate Used in Surveys Conducted Survey Round Main Survey Rider Survey Replicates 1: 4 Replicate 1 Replicate 2 Replicate 3 July 2003 LFS 2003 FIES V 1 October 2003 LFS 2003 SOF January 2004 LFS 2003 FIES V 2 April 2004 LFS 2004 FPS July 2004 LFS 2004 APIS 2004 SOF October 2004 LFS 2004 HECS January 2005 LFS April 2005 LFS 2005 FPS July 2005 LFS 2005 SOF October 2005 LFS 2005 HSDV January 2006 LFS April 2006 LFS 2006 FPS July 2006 LFS 2006 FIES V 1 October 2006 LFS 2006 SOF January 2007 LFS 2006 FIES V 2 April 2007 LFS July 2007 LFS 2007 APIS October 2007 LFS 2007 SOF January 2008 LFS April 2008 LFS 2008 ISS July 2008 LFS 2008 APIS October 2008 LFS 2008 SOF January 2009 LFS April 2009 LFS July 2009 LFS 2009 FIES V 1 October 2009 LFS HSDV (Core) January 2010 LFS 2009 FIES V 2 13 th National Convention on Statistics Source: NSO Philippines October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City Replicate 4 2003 FIES V 1 2003 SOF 2003 FIES V 2 2004 APIS 2004 SOF 2004 HECS 2005 FPS 2005 SOF 2005 HSDV 2006 FPS 2006 FIES V 1 2006 SOF 2006 FIES V 2 2007 APIS 2007 SOF 2008 ISS 2008 APIS 2008 SOF 2009 FIES V 1 HSDV (Core) 2009 FIES V 2

I. Introduction Table 3. Total Sample and its Percent Distribution by Outcome of Interview: 2009 FIES Outcome of Interview Total Sample Completed interview (code 1) July Perce 2009 nt 52, 233 100. 0 January Perce 2010 nt 48, 441 100. 0 40, 957 78. 4 40, 521 83. 7 274 0. 5 304 0. 6 Temporarily away, not at home, on vacation (code 3) Vacant housing unit (code 4) 1, 491 2. 9 1, 390 2. 9 3, 036 5. 8 3, 771 7. 8 Housing unit demolished, destroyed by fire, typhoon (code 5) Sample household moved-out of the housing unit (code 6) Critical area, flooded area (code 7) 1, 495 2. 9 509 1. 1 4, 889 9. 4 1, 879 3. 9 91 0. 2 67 0. 1 Refusal (code 2) 13 th National Convention on Statistics October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City

I. Introduction Table 4. Combination of Interview Status Code of July 2009 and January 2010 FIES Visit 2 FIES Visit 1 Code 1 - Code 2 -Refusal Completed Code 1 - Total of FIES Code 3 -7 Interview Status 38, 400 158 2395 40, 953 Code 2 -Refusal 113 115 45 273 Code 3 -7 520 24 4, 113 4, 497 39, 033 279 6, 411 45, 723 Completed Total of FIES Interview Status 13 th National Convention on Statistics October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City

I. Introduction Research Objectives To identify the characteristics of nonresponding sample households and their housing units based from 2009 FIES in the Philippines. Specifically, the variables under study are the following: 1) age 2) sex 3) household size 4) total income 5) marital status 6) highest grade completed 7) industry 8) occupation 13 th National Convention on Statistics October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City

II. Methodology Nonresponse rate was computed by getting the proportion of nonresponding to the total cases and by sex. Decomposition of nonresponse rate was also computed to get the proportion of each level of variable to the total nonresponse rate. In this way, each level shows to what extent the variable was influenced by the change of the factor of nonresponse. 13 th National Convention on Statistics October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City

III. Findings of the Study 1. Proportion of cases of unsuccessful interview to total sample is quite large in household sample surveys in the Philippines. There are many causes to nonresponses. However, as the first step, this research focused on the most crucial causes of refusal. It is recommended that characteristics that affect refusal as outcome of interview be given consideration during field operations whether census or survey. 2. The same method of analysis can be applied on cases on interview code 3, temporarily away, which also have large proportion. 13 th National Convention on Statistics October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City

III. Findings of the Study 3. Declining count of eligible respondents • As shown in table 1, the number of eligible respondents is declining over time. This can be attributed to the not updated sampling frame. The MS 2003 uses housing units in the sample selection because of its stability. However, due to lack of budget for updating the list of housing units and households, the newly built housing units was not updated in the sampling frame of MS 2003. As a result not all potential samples are considered in the sample selection. It is recommended in the redesign of new master sample to allocate budget for the annual updating of sampling frame. 13 th National Convention on Statistics October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City

III. Findings of the Study 4. Characteristics of nonrespondents • high proportion of nonrespondents were those household heads that are 65 years and over, with family size of one member, with family income of 500, 000 Ph. P and over, college graduate, engaged in industry of health and social work, and their major occupation are professional workers. 5. The results of the study can be validated using census data, which can provide characteristics of nonrespondents. It is possible to get the nonrespondents characteristics if there will be unique identifier of households that will link master sample and census data. The characteristics of households of nonrespondents in surveys can be verified to census data. 13 th National Convention on Statistics October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City

IV. Recommendations • annual updating of sampling frame • improve field operations through: tailor made interview techniques extensive information campaign drive competency of interviewers timing of interview employed self-administered questionnaires • establish linkage between MS and census data • study characteristics of temporarily away household 13 th National Convention on Statistics October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City

Thank you very much. 13 th National Convention on Statistics October 3 -4, 2016, EDSA Shangri-La Hotel, Mandaluyong City