Pan-African Socio-Economic Status measures (PA-SES)

Pan-African Socio-Economic Status measures (PA-SES) Three levels of sophistication on both household and individual versions Neil Higgs & Heidi Swanepoel TNS South Africa 29 May 2014 Upd 18 November 2014 Upd 1 Aug 2015

Contents 1 Background 03 2 The process 07 3 Short-form versions for PAMRO and 25 ESOMAR 4 The questions and the scoring system 38 2

Contents 1 Background 03 2 The process 07 3 Short-form versions for PAMRO and 25 ESOMAR 4 The questions and the scoring system 38 3

Why did we tackle this? Three main reasons: 1 2 3 ESOMAR approached SAMRA late 2013 SAMRA asked for our help due to our expertise in this arena. We agreed to do it pro-bono. We were keen to do this also for our region and, recently, interest in this mushroomed wildly from many other parts of the TNS world. Then we heard in early 2014 that PAMRO had also been interested in this for some time already.

The initial problem and specifications ESOMAR had no funds for this It needed to be very short (current SA LSMs are generally felt to be too long and cumbersome) Short is also good for mobile Short is also good from the point of view of very short questionnaires arising out of neuroscience thinking and the need to tap into the System 1 part of our brains PAMRO needed something unique for their members 5

We needed (free) data Nothing common across the TNS AMME world UL has a proprietary measure not available Came across Afrobarometer dataset across 20 countries in Africa we were able to obtain the relevant components for mahala: the q re is almost identical across all countries 6

Contents 1 Background 03 2 The process 07 3 Short-form versions for PAMRO and 25 ESOMAR 4 The questions and the scoring system 38 7

What goes into the pot? Segmentations = (A/B, C1, C2, D, E) Scoring systems and partitions (LSM) 8

What drives SES? What do you think of when you think of higher SES? Nice house, modern materials, good infrastructure, durables in home, access to communications/www/ mobile, car, better job, more education That makes sense and how about lower SES? Poor people in smaller/not so good houses, less formal materials (shacks), fewer amenities, fewer durables, less access to what the world has to offer, unskilled, poor education 9

We decided on three tiers of measure A best in class measure (in fact, two) from which shorter versions could be derived with confidence Long versions A shorter derivative exclusively for PAMRO A very short one for ESOMAR and mobile Our initial brainstorm with ESOMAR suggested Reasonably short versions Dwelling type Neighbourhood services Number of people responsible for (e.g. a headman or chief)(!!) Very short version 10

The Afrobarometer dataset Country West Africa East Africa Southern Africa Benin 1200 The dataset contains data for three African regions West Africa, East Africa and Southern Africa. The twenty countries covered in each region with their sample sizes are as follows: Burkina Faso 1200 Cape Verde 1208 Ghana 2400 Liberia 1199 Mali 1200 Sierra Leone 1190 Togo 1200 Burundi 1200 Kenya 2399 Tanzania 2400 Uganda 2400 Botswana 1200 Lesotho 1197 Malawi 2407 Mauritius 1200 Namibia 1200 South Africa 2399 Zambia 1200 Zimbabwe 2400 11

Decide on the system and select the variables A scoring system is more flexible (can partition in different ways) and better reflects reality Individual level/household level (discuss) We investigated the common variables available in the Afrobarometer database and identified those that could correlate with SES Our first draft included 23 variables Type of shelter Roofing material Presence of electricity grid Presence of piped water in the neighbourhood Presence of sewerage system in neighbourhood Presence of a cell phone service Presence of a post office Presence of a school Presence of a police station Presence of a health clinic Presence of market stalls Ownership of a TV Presence of a car or motorcycle Cellular ownership and usage Frequency of use of a computer Frequency of use of the internet Frequency of usage of a cell phone Frequency of sending or receiving text messages Frequency of sending or receiving money Source of water Location of toilet Level of employment Level of education 12

Dimension 2 (6.0% variance explained) Create a big square matrix (Burt matrix) Use correspondence analysis to reveal the continuum Individual measure Reduced variables - Individual - Inertia = 0.33 1.000 EDU Post-graduate 0.800 USE COMPUTER Every day USE INTERNET Every day 0.600 CELLULAR USAGE AND OWNERSHIP USE CELL No, PHONE I never Never use EDU University completed a mobile phone 0.400 ROOF Tiles Some other material SOURCE OF WATER Inside LOCATION the house OF TOILET Inside the house ROOF Concrete SEND/RECEIVE EDU No formal TEXT schooling EDU Some university 0.200 MESSAGE Never CAR/MOTORCYLCE Yes, do LOCATION OF TOILET None, own SHELTER ROOF Traditional Thatch or house grass / USE INTERNET A few times a SHELTER Flat in a block of no latrine available USE COMPUTER hut SEWAGE A few SYSTEM times SEND/RECEIVE Yes TEXT EDU EDU Some Informal primary schooling schooling only week flats a week EDU SHELTER Post-secondary MESSAGE Room PIPED Five in or a WATER more hotel, times TV Yes, do own or SYSTEM Yes a residential hotel SHELTER Non-traditional / POST POLICE ROOF ROOF OFFICE STATION Plastic Shingles sheets No ELECTR GRID No qualifications, POST not OFFICE university USE ELECTR per CELL day Yes PHONE 0.000 GRID Yes Five or CAR/MOTORCYLCE LOCATION SOURCE SEWAGE TV SYSTEM No, PIPED No, CELLULAR don't OF OF don't WATER WATER No own TOILET USAGE Outside AND SYSTEM No -1.400-1.200-1.000-0.800-0.600 ROOF -0.400 Asbestos POLICE STATION Yes USE INTERNET COMPUTER formal house SHELTER ROOF Single Multiple room materials own Never more times -0.200 CELLULAR per day 0.000 USAGE AND 0.200 Outside 0.400 CELLULAR OWNERSHIP the compound 0.600USAGE Yes, I AND use 0.800 a in a SHELTER Other OWNERSHIP larger Yes, SHELTER ROOF dwelling I use Hostel Metal, a USE CELL mobile OWNERSHIP PHONE owned Less by Yes, someone than I use a else EDU Secondary SEND/RECEIVE in structure tin an or zinc or mobile owned outside by my someone hhold else USE backyard USE INTERNET COMPUTER A few A times few times a mobile USE school EDU phone USE that CELL I own PHONE One or two one time per day industrial CELL Some TEXT PHONE secondary LOCATION compound Three EDU MESSAGE SEND/RECEIVE Three TEXT SHELTER or farming or Primary OF TOILET school Inside completed completed/high school four school/high or times four times per school the compound day per day Temporary structure in my household month SOURCE MESSAGE One or two SEND/RECEIVE times TEXT a month per -0.200 day OF WATER Inside compound the compound per day MESSAGE Less / shack than one time per day USE USE INTERNET COMPUTER Less Less than than once once a month a month -0.400-0.600 Dimension 1 (41.4% variance explained) 13

Dimension 2 (6.8% variance explained) Use correspondence analysis to reveal the continuum Household measure Reduced variables - Household - Inertia = 0.38 0.600 SOURCE OF WATER Inside the 0.400 SHELTER compound Single room in a larger dwelling structure or SHELTER Hostel in an backyard SHELTER Temporary structure industrial compound or farming / shack compound LOCATION OF TOILET Inside SHELTER Other ROOF Metal, tin or the zinc compound 0.200 POST OFFICE POLICE Yes STATION Yes ROOF Multiple materials SHELTER CELLULAR Non-traditional USAGE / AND ROOF Plastic sheets CELLULAR USAGE AND ELECTR GRID Yes formal OWNERSHIP house Yes, I use a ROOF AsbestosPIPED WATER SYSTEM mobile Yesphone that I own LOCATION OWNERSHIP Yes, I use a SOURCE ROOF SEWAGE TV Shingles SYSTEM No, don't OF PIPED No own WATER TOILET Outside SEWAGE SYSTEM 0.000 TV Yes, do own Outside mobile owned the by WATER someone SYSTEM else the compound No -1.200-1.000-0.800-0.600-0.400-0.200 0.000 0.200 POST OFFICE 0.400 No in SHELTER Flat in a block of CELLULAR my household 0.600 ELECTR USAGE GRID AND 0.800 POLICE STATION No No flats OWNERSHIP Yes, I use a mobile owned by someone else CELLULAR outside my USAGE hhold AND -0.200 LOCATION OWNERSHIP SHELTER Traditional OF No, TOILET I never house None, use / SHELTER Room in a hotel, or no a latrine mobile hut available phone a residential hotel ROOF Thatch or grass ROOF Tiles LOCATION OF TOILET Inside the house SOURCE OF WATER Inside the house ROOF Concrete ROOF Some other material -0.400-0.600-0.800 Dimension 1 (46.9% variance explained) 14

Look for main differentiators and stable items and finalise variables The relative differentiating power of each of the final items in both the individual-level measure (17 items) and the household-level measure (10 items) are as follows: Item Maximum score and % differentiation - Individuals Maximum score and % differentiation - Households Education 10.1 Freq of use of a computer 7.7 Freq of use of internet 7.5 Roof material 7.1 13.6 Location of toilet 7.0 12.4 Source of water 6.8 12.2 Type of shelter 6.4 10.9 Freq of use of cell phone 5.9 Freq of send/receive text msg 5.6 Sewage system 5.3 9.9 TV 5.1 8.7 Electric grid 4.9 9.2 Usage of cell phone 4.7 Piped water 4.5 8.6 Post office 4.1 8.1 Car/motorcycle 4.0 Police station 3.2 6.4 15

Distribution of the two SES measures across 32,400 people class intervals of ten 20.0% 18.0% Individual vs household 20.0% 18.0% Individual vs household 16.0% 16.0% Indiv 14.0% Indiv 14.0% HH 12.0% HH 12.0% 10.0% 10.0% 8.0% 8.0% 6.0% 6.0% 4.0% 4.0% 2.0% 2.0% 0.0% 0 5 15 25 35 45 55 65 75 85 95 0.0% 0 5 15 25 35 45 55 65 75 85 95 16

0 2.5 7.5 12.5 17.5 22.5 27.5 32.5 37.5 42.5 47.5 52.5 57.5 62.5 67.5 72.5 77.5 82.5 87.5 92.5 97.5 Distribution of the two SES measures across 32,400 people finer class intervals (five) 14.0% 12.0% 10.0% 8.0% Finer class intervals Indiv HH 6.0% 4.0% 2.0% 0.0% 17

Scores by country Country Individual score Household score Benin 31.2 35.1 Botswana 50.2 55.2 Burkino Faso 27.7 27.3 Burundi 22.7 28.2 Cape Verde 54.7 66.4 Ghana 37.6 41.2 Kenya 39.7 41.2 Lesotho 32.8 35.2 Liberia 25.8 21.3 Malawi 19.4 18.9 Mali 29.6 33.8 Mauritius 65.0 79.3 Namibia 38.9 34.1 Sierra Leone 30.6 28.8 South Africa 58.0 69.1 Tanzania 26.9 24.7 Togo 34.3 36.9 Uganda 25.8 24.9 Zambia 36.0 37.4 Zimbabwe 41.5 43.9 Overall mean 36.2 38.6 Standard deviation 22.6 27.4 18

Household Scores plotted by country 90 80 y = 1.3169x - 8.8295 R² = 0.9599 Mauritius 70 South Africa Cape Verde 60 50 Botswana 40 30 20 Lesotho Benin Liberia Ghana Total Togo Zambia Mali Burundi Sierra Leone Uganda Burkino Faso Tanzania Malawi Zimbabwe Kenya Namibia 10 0 0 10 20 30 40 50 60 70 Individual 19

Distributions for six selected countries 35.0% 30.0% 25.0% 20.0% Individual-level SES in six countries Botswana Cape Verde Kenya Malawi Mauritius South Africa 45.0% 40.0% 35.0% 30.0% 25.0% Household-level SES in six countries Botswana Cape Verde Kenya Malawi Mauritius South Africa 15.0% 20.0% 10.0% 15.0% 10.0% 5.0% 5.0% 0.0% 0 5 15 25 35 45 55 65 75 85 95 0.0% 0 5 15 25 35 45 55 65 75 85 95 20

Generate scoring system, rescale 0 to 100, establish validity and reliability Reliability - Cronbach s is 0.893 for the individual level measure and 0.863 for the household level measure both excellent results Construct validity via theory, sense-check, prior experience and determination of shape of and trace functions 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 Elec grid Sewage Police station Car/motorcycle Freq internet Freq cell phone Source water Educ Roof Piped water Post office TV Use computer Usage of cell phone Freq of text msg Loc toilet Shelter 0.0 5 15 25 35 45 55 65 75 85 95 21

Implications for the future based on these versions There are three prime advantages to this system: From a practical point of view, TNS has developed two scoring systems, one for individuals and one for households, that have potential application across Africa, allowing comparisons to be made that make sense. The approach adopted largely equates infrastructure, ownership of key durables and, in the case of the individual scoring system, education and personal use of technology with socioeconomic status. The measures help marketers and communicators understand much more about how people actually live at either household or area level, as well as understand their access to wider horizons through the use of technology and use of transport. Education levels, if used, imply potential employment. Hence, there will be a better understand of people s needs. 22

Implications for the future based on these versions The algorithm is a work-in-progress as there may be additional variables that might be useful to include. In particular, there may be other durables and agricultural items that are good indicators of wealth. Additional primary research in a small selection of countries has been undertaken and some potential improvements identified. Of importance will be the need to co-ordinate this research across more countries so that a common set of variables emerges. Also of importance is the need not to include too many variables simplicity commensurate with a system that has good properties is critical. Further, some key economies such as Nigeria, Angola and Mozambique were not available in the dataset: some work in these countries has been done but we need a co-ordinated approach across more PAMRO countries to do this properly. This will be discussed in our panel discussion in Session eight External validation at country level is also needed so as to relate scores on these SES measures with incomes or other measures of wealth that might be in use in various countries (such as the A/B/C1/C2/D/E system) We have profiles of the variables across the 0 to 100 spectra but we do need to drill down within each country to understand the patterns of the actual variables along the scoring distributions within country. For example, what are the characteristics of people in the class 10.01 to 20.0 in Malawi, or in Mauritius? 23

Implications for the future based on these versions Limitations continued: The scoring systems given represent the best that can be achieved with the current (free) dataset. However, they are too long for PAMRO and ESOMAR purposes, especially given the rapid rise in very short questionnaires that, from our new knowledge of how the brain works, are on the increase and also with the rapid increase in research on mobile devices. Hence, TNS has developed two much shorter forms of the household measure with this in mind. Of the two outlined below, the very short form is in line with the original request from ESOMAR whilst the longer integer form is more suitable for PAMRO s needs. Both yield very similar results but are not exactly the same as the long-form version discussed above. TNS has now also developed a short-form version of the individual measure due to many requests to calibrate against the A/B/C1/C2/D/E systems present in many countries (though not applied in the same way in different countries) 24

Contents 1 Background 03 2 The process 07 3 Short-form versions for PAMRO and 25 ESOMAR 4 The questions and the scoring system 38 25

Short-form pan-african SES measures: Reducing the variable set household measure An inspection of the relative differentiating power of the variables, as well as the correlations between them, led to the household variable set being reduced to the following: 1. Roof material 2. Type of shelter 3. Electricity grid 4. Source of water 5. Post office 6. TV 26

Dimension 2 (8.0% variance explained) Use correspondence analysis to reveal the continuum six variables only household measure Short-form SES (six variables) - Inertia = 0.71 0.800 SHELTER Single room in a larger dwelling structure SHELTER or Temporary structure backyard SHELTER / shack 0.600 Other SHELTER Hostel in an industrial SOURCE compound OF 0.400 WATER or farming Inside the compound ROOF Metal, ROOF tin or Multiple zinc materials 0.200 ROOF ROOF Plastic Shingles sheets SHELTER Non-traditional / SOURCE OF WATER Outside formal house TV No, don't own ELECTR GRID Yes 0.000 POST OFFICE No the compound POST OFFICE Yes ELECTR GRID No -1.500-1.000 TV Yes, -0.500 SHELTER Flat in a block of SHELTER do own 0.000 0.500 1.000 Room in a flats hotel, or ROOF Asbestos a residential hotel -0.200 SOURCE OF WATER Inside the house ROOF Tiles ROOF Concrete -0.400-0.600 SHELTER Traditional house / hut ROOF Thatch or grass ROOF Some other material -0.800-1.000 Dimension 1 (35.0% variance explained) 27

Differentiating power of the short-form items (household measure) As before, by examining the maximum values of each item, the relative differentiating power is determined and shown below: Item Maximum score and % differentiation - One decimal place form Maximum score and % differentiation - Integer form Maximum score and % differentiation - Shortest form Roof material 24.2 24 24 Source of water 17.9 18 18 Type of shelter 17.3 17 17 TV 14.3 14 14 Electric grid 14.4 15 15 Post office 11.8 12 12 28

Take most differentiating variables and reduce duplication short form versions (household measure) Reliability- Cronbach s is 0.798 for the household level measure very good Construct validity via correlation with long versions 0.96 All three short-form versions have a correlation of 0.9998 with themselves Important note: all versions do measure the construct of SES but the short versions do not yield exactly the same scores as the long version. Hence, once a short or long version has been chosen, it should not be used interchangeably with a version of different length. The three short versions have 97% level of agreement between themselves, however. 29

Characteristics of the measures It is evident that the three measures are indeed almost identical. Descriptive statistics are shown below: Item One decimal place form Integer form Shortest form Mean 43.1 43.0 43.1 Median 42.1 42.0 42.0 Mode 0 0 27.0 Range 97.9 97.0 98.0 Minimum 0.0 0.0 0.0 Maximum 97.9 97.0 98.0 Standard deviation 28.5 28.3 28.6 Kurtosis -1.1-1.1-1.1 Skewness 0.1 0.1 0.1 30

Comparison of distribution across the short-form measures The following chart confirms that the three measures are virtually identical. Note that the extreme left point (0) represents those with a score of exactly zero. 20% 18% 16% 14% 12% 10% 8% 6% 4% 2% 0% Short-form distribution comparisons One decimal Integer Shortest 0 5 15 25 35 45 55 65 75 85 95 31

Scores by country Country One decimal version Integer version Shortest version Benin 39.6 39.6 39.3 Botswana 60.1 60.0 60.2 Burkino Faso 34.1 34.2 33.9 Burundi 31.7 31.8 31.4 Cape Verde 75.3 74.5 76.3 Ghana 47.1 46.9 46.9 Kenya 46.4 46.4 46.3 Lesotho 38.7 38.8 38.8 Liberia 26.6 26.7 26.2 Malawi 20.2 20.2 20.2 Mali 36.8 36.7 36.7 Mauritius 88.0 87.4 88.4 Namibia 35.2 35.1 35.1 Sierra Leone 33.3 33.5 32.9 South Africa 72.4 72.1 72.6 Tanzania 28.6 28.7 28.2 Togo 41.7 41.6 41.5 Uganda 28.6 28.6 28.3 Zambia 40.7 40.6 40.7 Zimbabwe 48.0 47.7 48.2 Overall mean 43.1 43.0 43.1 Standard deviation 28.5 28.3 28.6 Original longfrom HH version 35.1 55.2 27.3 28.2 66.4 41.2 41.2 35.2 21.3 18.9 33.8 79.3 34.1 28.8 69.1 24.7 36.9 24.9 37.4 43.9 38.6 27.4 33

Original long-form household version Relationship between shortest form and original long form by country Country scores: shortest vs original household versions 100.0 80.0 60.0 y = 0.9153x - 0.7639 R² = 0.991 40.0 20.0 0.0 0.0 20.0 40.0 60.0 80.0 100.0 Shortest version 34

Developing the short form version of the individual level measure 35

Differentiating power of the short-form items (individual measure) As before, by examining the maximum values of each item, the relative differentiating power is determined and shown below: Item Maximum score and % differentiation - Integer form Roof material 13 Source of water 11 Type of shelter 12 Use of computer 17 Use of internet 17 Use of cell phone 11 Education level 21 36

Scoring systems for both measures (and see manual) Variable Level H/hold Individ Roof Thatch or grass 0 0 material Shingles 9 3 (obsvn) Plastic sheets 10 4 Multiple materials 11 6 Metal, tin or zinc 13 7 Asbestos 21 12 Concrete 23 12 Tiles 21 13 Some other material 24 12 Source Outside the compound (far away) 0 0 of water Inside the compound (nearby) 10 5 Inside the house 18 11 Type of Traditional house/hut 0 0 shelter Other 7 5 (obsvn) Temporary structure/shack 8 5 Single room in a lgr dwelling/byard 12 5 Hostel in an indus/farming compound 15 7 Non-traditional formal house 15 8 Flat in a block of flats 15 11 Room in a hotel/residential hotel 17 12 TV Yes 14 Elec grid Yes 15 Post Off Yes 12 Variable Level H/hold Individ Com- Never use 0 putor/ Use less than once a month 8 laptop Use a few times a month 11 Use a few times a week 13 Use every day 17 Internet Never use 0 Use less than once a month 9 Use a few times a month 11 Use a few times a week 13 Use every day 17 Mobile Never use 0 Less than once a day 2 One or two times a day 5 Three to four times a day 7 Five or more times a day 9 Educ- No formal schooling 0 ation Some primary schooling 1 Informal schooling only 2 Primary schooling complete 3 Some secondary/high school 6 Secondary/high school completed 9 Post-secondary quals, not university 13 Some university 16 University completed 19 Post-graduate 21 37

Contents 1 Background 03 2 The process 07 3 Short-form versions for PAMRO and 25 ESOMAR 4 The questions and the scoring system 38 38

Appendix 2 Scoring system on short version of the SES measures (PAMRO version) Variable Level Household version Individual version Roof material Thatch or grass Shingles Plastic sheets Multiple materials Metal, tin or zinc Concrete Asbestos Some other material Tiles 0 9 10 11 13 23 21 24 21 0 3 4 6 7 12 12 13 12 Source of water Outside the compound Inside the compound Inside the house 0 10 18 0 5 11 Type of shelter Traditional house/hut Other Temporary structure/shack Single room in a larger dwelling or backyard Hostel in an industrial or farming compound Non traditional formal house Flat in a block of flats Room in a hotel/residential hotel 0 7 8 12 15 15 15 17 0 5 5 5 7 8 11 12 43

Appendix 2 Scoring system on short version of the SES measures (PAMRO version) - continued Variable Level Household version Individual version TV No Yes 0 14 Electricity grid No Yes 0 15 Post Office No Yes 0 12 Computer Internet Cellphone use Education Never use Use less than once a month Use a few times a month Use a few times a week Use every day Never use internet Use internet less than once a month Ise internet a few times a month Use internet a few times a week Use internet every day Never Less than once per day One or two times per day Three or four times per day Five or more times per day No formal schooling Some primary schooling Informal schooling only Primary schooling completed Some secondary school/high school Secondary school/high school completed Post-secondary qualifications, not university Some university University completed Post-graduate 0 8 11 13 17 0 9 11 13 17 0 2 5 7 9 0 1 2 3 6 9 13 16 19 21 44

Appendix 3 Questions used from the Afrobarometer questionnaire (full list use as appropriate) EA-SVC. Are the following services present in the primary sampling unit / enumeration area? Yes No Can t determin e A. Electricity gird that most houses could access 1 0 9 B. Piped water system that most houses could access 1 0 9 C. Sewage system that most houses could access 1 0 9 D. Cell phone service 1 0 9 EA-FAC. Are the following services present in the primary sampling unit / enumeration area or within easy walking distance? Yes No Can t determin e A. Post office 1 0 9 B. School 1 0 9 C. Police station 1 0 9 D. Health clinic 1 0 9 E. Market stalls (selling groceries and/or clothing) 1 0 9 47

Appendix 3 Questions used from the Afrobarometer questionnaire - continued 90. Which of these things do you personally own? No (don t own) Yes (do own) Don t know A. Radio 0 1 9 B. Television 0 1 9 C. Motor vehicle, car or motorcycle 0 1 9 91. How often do you use: Every day A few times a week A few times a month Less than once a month Never Don t know A. A computer? 4 3 2 1 0 9 B. The internet? 4 3 2 1 0 9 48

Appendix 3 Questions used from the Afrobarometer questionnaire - continued 92. Do you ever use a mobile phone? If so, who owns the mobile phone or cellphone that you use most often? No, I never use a mobile phone or cellphone 0 Yes, I use a mobile phone or cellphone that I own 1 Yes, I use a mobile phone or cellphone owned by someone else in my household 2 Yes, I use a mobile phone or cellphone owned by someone outside my household 3 Don t know 9 93. How often do you normally use a mobile phone or cellphone to: Never Less than one time per day One or two times per day Three or four times per day Five or more times per day Don t know A. Make or receive a call? 0 1 2 3 4 9 B. Send or receive a text message or SMS? 0 1 2 3 4 9 C. Send or receive money or pay a bill? 0 1 2 3 4 9 C-SAF. Access the internet? 0 1 2 3 4 9 49

Appendix 3 Questions used from the Afrobarometer questionnaire - continued 95. Please tell me whether each of the following are available inside your house, inside your compound, or outside your compound: None, no latrine available Inside the house Inside the compound Outside the compound Don t know A. Your main source of water for household use 1 2 3 9 B. A toilet or latrine 0 1 2 3 9 96. Do you have a job that pays a cash income? [If yes, ask:] Is it full-time or part-time? [If no, ask:] Are you presently looking for a job? No (not looking) 0 No (looking) 1 Yes, part time 2 Yes, full time 3 Don t know 9 50

Appendix 3 Questions used from the Afrobarometer questionnaire - continued 97. What is the highest level of education you have completed? No formal schooling 00 Informal schooling only (including Koranic schooling) 01 Some primary schooling 02 Primary school completed 03 Some secondary school / high school 04 Secondary school / high school completed 05 Post-secondary qualification, other than university, e.g. a diploma or degree from a polytechnic/collage 06 Some university 07 University completed 08 Post-graduate 09 Don t know 99 51

Appendix 3 Questions used from the Afrobarometer questionnaire - continued 104. In what type of shelter does the respondent live? Non-traditional / formal house 1 Traditional house / hut 2 Temporary structure / shack 3 Flat in a black of flats 4 Single room in a larger dwelling structure or backyard 5 Hostel in an industrial compound or farming compound 7 Other 8 52

Appendix 3 Questions used from the Afrobarometer questionnaire - continued 105. What was the roof of the respondent s home or shelter made of? Metal, tin or zinc 1 Tiles 2 Shingles 3 Thatch or grass 4 Plastic sheets 5 Asbestos 6 Multiple materials 7 Some other material 8 Could not tell / could not see 9 53

Pan-African Socio-Economic Status measures (PA-SES) Three levels of sophistication with an emphasis on household versions Heidi Swanepoel Neil Higgs TNS South Africa 1 Aug 2015