1. Summary of conclusions

2. Prior Questions Questions for DiscussionSome Answers How interested are government s in Strand C This is a PILOT; high-level political considerations may determine level and nature of country participation in Strand C What is the policy question  Principally OECD/PISA intention to cover all 15 year olds  Secondarily, country-specific, why are they out of school?, why did they drop out?, etc Who wants to do what with information? Combining in-school and OOS samples not very sensible:  difficult to get different contractors to collaborate;  Areas with schools will have different OOS profile than non-school areas

3. 2a Counting and Locating Questions for Discussion Some Answers What is our cohort? 15.3- 16.2? 1. extending age range especially for contextual questionnaire. 2. interviewing and testing all 15 yo children What is an Out- of-School 15 year old? Take care when labelling a 15 year old as in-or out- of school. Reality is more complex Some alternative instruction patterns may be leading to a recognised examination Is our overall picture correct? Consensus was that the overall picture was largely correct. To be confirmed at country level Where are out- of-school 15 year olds Consensus was that identification of rural poor and urban slums was about right. Importance of ‘missing’ vulnerable groups; possible use of estimation procedures

4. 2b Finding and Identifying (1) Questions for Discussion Some Answers How to survey OOS 15 year olds 1.Most solutions have started from school sample and searched for OOS 15yo; also search for OOS 15 yo at home and then go to school 2.Area-census may be problematic Advantages and disadvantages of Options 1 (RPS) and 2 (Targeting) Nearly everyone agreed, with partial exception of Senegal, they would or are already planning a targeted approach because an RPS would be inefficient Are these options mutually exclusive or complementary NO: e.g. Senegal will go for an RPS with overweighting of sample in vulnerable areas.

5. Finding and Identifying (2) Questions for DiscussionSome Answers ADDITIONAL How do we identify marginal and vulnerable goups Local knowledge, poverty maps, include an educational anthropologist/ sociologist on national team Can you think of any other options Ask the following questions  Which are most useful questions  Which 15 year olds can I help  Who are the 15 yos I should interview to reach those conclusions If there is a functioning EMIS over several years then, conditional on migration patterns, one can estimate from cohort How about synergies with other strands If comparability is the aim, then that defines the appropriate sample

6. 2c. Administering a Survey (1) Questions for Discussion Some Answers How to administer the survey? Don’t use term ‘child’ use ‘young adult’ or ’15 year old whether married, in school or working What can we learn from other experiences? ASER selects enumerators from Districts, UWEZO from local village; latter preferred? What do you think of the recommended approach? Consensus on two stage approach with option for interviewing all youths, say between 14-16 and only testing 15 year old

7. 2c Administering a Survey (2) Questions for Discussion Some Answers Any other alternatives to be considered? Use of mobile, etc. DHS and MICS have experimented with tablets; to be explored by contractors Some of f2f questions could be movd to observation checklist ADDITIONAL How to tackle possible difficulty of mini-census Local knowledge and support What are minimum numbers to be in Conformity with PISA Standards Need 200-400 responses from each response item

8. Options for Testing (1) Questions for Discussion Some Answers How to administer the survey? If administered in evening be aware of possible bias from other members of the household being present What can we learn from other experiences? In small rural areas, pre contact with head of village, head Buddha or Iman or priest essential; any child welfare officer or equivalent will also be very useful What do you think of the recommended approach? Generally agreed

9. Options for Testing (2) Questions for Discussion Some Answers Any other alternatives to be considered? Adaptive computer testing is a possibility (LAMP) ADDITIONAL Are these PISA scores? NO; but some wanted a couple of PISA low level items in order to provide a baseline for trends. How will security of PISA items be ensured? 1.Security only an issue with cognitive instrument 2.Only professional interviewers administer test not citizen volunteers

10. Designing the Instrument Questions for Discussion Some Answers What is the policy issue In taking disability seriously? Issue of inclusion because they are a higher proportion of Out-of-School What are implications of taking disability seriously? Difficult and costly to assess disabled Items would have to be properly adapted to accommodate range of disabilities Other Issues Can responses of Strand C be scaled with standard PISA Yes using available mathematical/ statistical procedures; but not from commercial software

11. Contextual Questionnaire Questions for Discussion Some Answers How much integration should there be between Strands B and C Strong feeling that the contextual questionnaire should be as similar as possible, with exception of school-based questions (resources on the one hand and impediments on the other) What mechanisms can be put in place to encourage collaboration? Contract of ToR Anything else?

12. Contextual Questionnaire Questions for Discussion Some Answers How much should we ask about Special Needs? Probably best limited to their identification and collection o contextual data rather than attempting to assess competencies? How much should we collect information about the community? Simple observable indicators relevant to the factors that might influence the 15 yo to drop-out or to re-enter school Who should be respondent? Must be in conformity with Strands A and B But for knowing about process of drop- out, parent has to be involved