COUNTRY: KENYA

1

• COUNTRY KENYA
• NAME OF PRESENTER FRANCIS M. MUNENE
• TITLE CHIEF ECONOMIST, MINISTRY OF JUSTICE AND
  CONSTITUIONAL AFFAIRS.
• QUESTION SELF CARE QUESTIONS DOMAIN

2
2. METHODOLOGY

• 2.1 The sampling design
• The sampling design for the pre-test was
  purposive implying that the selection of
  respondents was not based on any authoritative
  method. Interviewers had full freedom to select
  their respondents from households, disability
  institutions, through friends and even relatives.
  Consequently, no inferences can be made from the
  results.
• Data collection, processing and submittal of the
  results strictly followed the general guidelines
  issued by the Secretariat.
• Kish Grid method was used to select the proxy
  respondents

3
Recruitment criteria of the Interviewers

• Selection of interviewers was strictly based on
• Ones educational qualification (at least with a
  minimum of graduate level university education),
  professional inclination and working experience
  in research related assignments
• Fluency (oral and written) in both English and
  their respective local language
• Availability though out the entire period of the
  survey

4

• Ethnic representation
• Good understanding of general principles of data
  collection
• Ability to translate the pretest from English
  (source language) to their local ethnic
  language)
• Ability to take interview notes (These two
  criteria were considered crucial in view of
  maintaining the original meaning of the English
  translation and debriefing session that was to
  follow after the interviews) and
• Quick grasp of the instruments (this was key for
  two reasons the limited time given for the
  exercise and two, respondents had to occasionally
  explain to respondents some of the
  hard-to-comprehend questions.

5

• All interviewers had
• Extensive training and experience in statistical
  surveys and research work.
• Good understanding of general principles of data
  collections, processing and analysis.
• Participated in diverse research assignments,
  mainly in CBS
• Acquired a minimum of graduate level education,
  with diverse educational specialization ranging
  from Economics, law, sociology to Applied
  Statistics.

6
2.3 Translation and duration of the Interviews

• Forward and reverse translation was performed
  from English (source language) to the four main
  languages that are mainly spoken. These are
  Kiswahili, Kikuyu, Kamba and Luhya).
• No major problems identified with language
  translation except that most of our local
  languages are not so rich in words that
  occasionally failed to get their equivalence of
  English words.

7

• On average, the interviews lasted between one
  and a half to 2 hours. This was by all standards
  a long duration and occasionally, interviewers
  had difficult time maintaining the respondents
  patience. This however depended on the
  respondents level of education, their
  relationship and the amount of interruptions
  during the interviews. Six interviewers, 2
  Editors/ data processing and 1supervisor were
  recruited for the exercise. They were drawn from
  the National Statistics Office, some of the
  reputable research institutions and the local
  universities.

8
RESULTS

• 2.1 Definition of Self Care Disability Been
• defined by the Facilitator
• 2.2 Key Disability and Functioning Self Care
• Questions Given by the facilitator
• 2.3 Explanation of the Patterns been
• discussed by the facilitator
• 2.4 Results of the Response Patterns for
• Kenya

9
TABLE 2 COMBINED RESPONSE PATTERNS FOR SELF AND
PROXY BY BROAD CATEGORIES

• 2.4.1 RESULTS BY RESPONSE PATTERNS
• The results are tabulated in three categories.
  The first set of results give 5 specific response
  categories (A,B,C, F, G) for both self and proxy
  as tabulated from the data, the second set shows
  the 3 broader pattern categories of disability
  (non problematic constituted by A and E,
  borderline by B,C, F, and G and Problematic -
  formed by D and H) while the third set presents
  cross tabulations of the broader response
  patterns by key demographic variables

10
(No Transcript)
11
CONT
12
CHART 2
13
2.4.2 Non-Problematic Pattern - Category A and E

• Pattern A and B are clearly non-problematic. A
  reports no disability and negative responses to
  all the follow-up functioning questions, which is
  perfectly consistent.
• E reports difficulty with self care and then at
  least three positive responses to the follow-up
  functioning questions. Again, perfectly
  consistent.

14
(No Transcript)
15
Borderline Category Pattern F and G and B and
C

• F and G are labeled as non-problematic but report
  at least one functioning problem (one functioning
  problem for G and two functioning problem under
  F). At first, it was a bit difficult to decide
  where to place it. If both of the categories had
  reported one functioning problem, then, it would
  have been easier to place them in the
  non-problematic domain. But F has two
  functioning problems which was somewhat mild and
  therefore placing it in the borderline seemed to
  me perfectly okay.

16
(No Transcript)
17

• Categories B and C reported no self care
  disability but at least one functioning problem,
  quite similar to categories F and G.
• The two classes of patterns show perfect
  agreement up that level.
• The only difference between the two is that
  category B and C did not report disability
  whereas F and G did.
• Category C had nil responses, while the other
  three categories had each registered only one
  response.

18
(No Transcript)
19
2.4.4 OTHER SUPPORTING FINDINGS

• Majority (85) of responses give perfect
  agreement i.e. fall within the non-problematic
  category, 15 are in the borderline and about
  less than 1 percent in the problematic category.
• That response patterns A, G and G for self and C,
  D, and H for proxy had nil responses.
• There were no missing cases in both classes of
  categories.
• Categories F and G had one case each under proxy.
  Such a skewed distribution limits ability to
  explain some of the hard-to-explain patterns as
  in the case of B and C category and F and G
  categories.

20

• Tabulations were run by key demographic variables
  (namely age, education, work status, marital
  status) for each response category. This was in
  an effort to get some more insights on their
  effects with a view to explaining some of the
  hard-to-explain patterns.
• The results of the tabulations are shown in the
  following slides.

21
(No Transcript)
22
2.5 Other Findings from further Tabulations by
Key Demographic Variables and by Category

• Category A
• No difficult was reported for self across all
  ages but the proxy data revealed that about 98
  had some difficulties using hands and fingers
  while about 3 of the respondents felt too tired
  to dress.
• For the education variable, about 3 of the proxy
  university level respondents reported feeling too
  tired to dress. Analysis by work status and
  marital status did not show any significant
  differences in percentage terms for category A.
• Category B
• A large proportion of proxy respondents (25 to
  50) reported difficulties ranging from using
  fingers to button shirts to tying of shoe laces.

23

• Category E
• There were no respondents for self care on this
  category, but about 7 for proxy. The 7 proxy
  respondents revealed difficulties ranging from
  mild to severe in virtually all the 8 parts of
  question 5, with severity difficulties levelling
  around 50(71 difficult putting on socks, 86
  combing hair and needing someone to help on daily
  activities).
• Category F
• 16 of the self care respondents aged 50 years
  and above had difficult using hands, while 33 of
  the university level graduates had difficult
  putting on shoes against 33 for post secondary.
  For the proxy, 100 reported some difficulty
  reaching over their heads, and same too tired
  to bathe-no direct relationship seen.

24

• Category G
• This category registered nil case for self and
  only 1 case for proxy.
• The proxy respondent was aged 25-29 years, had
  university education, worked for pay, was not
  married but needed someone to help in daily
  activities.

25
3. Explanations for some of the Observed
Response Patterns Based on field Experiences
and further Tabulations

• 3.1 Interpretation of WG Question and Extended
  Set
• The core and extended set of questions were
  interpreted most consistently among most
  respondents, whether through proxy or self. This
  point is collaborated by the overwhelming and
  perfect agreement of response patterns observed.
  From the interview observations and follow-up
  debriefing, It was quite evident that the
  extended WG and the extended set were well
  understood by the respondents. Almost all of the
  self-reporting respondents and at least 11 out of
  a sample of 12 proxy respondents sampled from the
  data
• Did not require any part of the question
  repeated

26

• Never sought for clarification nor demonstrated
  any difficulties in using the response options.
  However, on enquiring further why they answered
  that way, it emerged that majority of the
  respondents did not take this question kindly.
  Majority took offence. The impression created was
  that they were being doubted on.
• The following examples demonstrate this point
• She does not have to ask for help to dress
• She knows her well and

27

• If she has a problem, she would know.
• Part of these impressive results are attributed
  to elaborate training, close supervision during
  field work and engagement experience and
  qualified staff. Two, compared to other domains,
  the self care questions were relatively straight
  forward and easy to understand.

28

• 3.2 Question-response problems
• There was some evidence of question-response
  problems but these were quite minimal and mainly
  conditioned on the physical aspects such as
  shaking ones hands, putting on socks/stockings
  etc. and to a large extend, the question was
  interpreted as asking about relatively serious
  limitations. They could have contributed to the
  larger cases falling under borderline.
• Many of the problems were centred on the
  respondents ability to conceptualize the level
  of limitation and to decide whether or not to
  respond in the affirmative.
• Such difficulties apparently were more common
  among respondents who had secondary and lower
  level of education, reported inferior work status
  and the aged.

29

• There were relatively more question-response
  problems with the proxy than self reporting.
• 3.3 Inclusion of Any Difficulty
• The inclusion of the word any difficulty raised
  some interpretation problems about the scope of
  difficulties.
• Respondents with mild self care difficulties
  tended to ignore the question and only those
  respondents who considered their impairment to be
  serious responded affirmatively to the question.
• Further, respondents tended to concentrate only
  on the long term conditions that are serious but
  omitted the short term mild conditions.

30
5. CONCLUSION

• In conclusion, many of the difficulties were
  largely due to design limitations of the
  instrument but not content-based.
• Compared to other domains, the self care
  questions were relatively easy, straight
  forwarded, well understood and correctly
  interpreted by majority of respondents except
  those with low education, had inferior
  occupations, and aged (above 50 years of age).
• We see direct relationship between age and
  difficult in reporting but no direct
  relationship with the level of education.
• That the various response patterns are consist
  across key demographic variables except for age
  and to some extent work status.
• Overall, we saw significant difference in error
  reporting between self and proxy responses.
• THANK YOU