Knowledge, attitude, practice and belief (KAPB) surveys

This is a simple way of gathering standardised information from a large number of people. It is a questionnaire with, ideally, both closed and open-ended questions. It is essential to pre-test any KAPB questionnaire with a sample of about 20-30 to ensure that the questions are relevant and nothing important has been missed out (AIDSCAP, n.d.). The most effective pre-testers of such questionnaires are very often the children themselves. Closed questions have a set of possible answers, or request a simple response such as yes/no or a numerical answer An open-ended question does not give possible answers leaving respondents free to give any answer Follow-up questions can give explanations for an answer or a particular position.

Closed questions are:

How old are you?
Have you ever been to visit a traditional healer?
Have you ever seen a condom?

Open-ended questions are:

What problems are young people facing today?
What are the feelings of young people towards condom use?
Where could you buy a condom?
Would you feel confident in talking to your partner about HIV/AIDS? Why?

Questionnaires can be pre-coded, giving dosed answers a numerical code, e.g.:

How old are you?
1.10-12
2.13-15
3.16-18
4. 19-21
5. 22 or over

Or:

Where would a young woman go for help if she had an STD?

1. To the nearest clinic?
2. To a traditional healer?
3. To a private doctor?
4. To another clinic in another area?
5. She would get treatment from her mother
6. She would get treatment from her grandmother
7. She would not go anywhere and not tell anyone - just keep it to herself.
8. To a vendor or market stall.
9. Other - specify.

Pre-coding a questionnaire often makes the implementation faster and data input quicker Pre-testing can help ensure the questionnaire is not too long, and the questions are understood and appropriately worded. In pre-testing, the range of answers to a single question can be estimated which will assist with coding. Coding of answers in questionnaires is more difficult and time-consuming.

In outreach work such as peer education projects, the questionnaires can be asked by the educators themselves, after a short training. This training (to conduct face-to-face interviews) should ensure the following:

1. The educators understand the purpose of conducting the survey and the principles of being objective.

2. The importance of ensuring confidentiality of the respondent's identity and answers,

3. The educators fully understand all the questions.

4. The educators have the skills to probe when an answer is incomplete - without leading the respondent.

5. The educators fully understand the sampling and selection process of the respondents.

6. The educators can answer basic questions about the objectives and activities of the project after completion of the questionnaire.

Relevant local authorities need to be informed that the research is taking place. For out-of-school youth, social mapping exercises can assist in the identification of the sample frame and sample areas. These exercises can also be done during training, so each interviewer has their own sample site. For in-school youth, the assistance of teachers, or reference to the register can help in the sample design. As a general rule, the sample should, if possible, have no less than about 100 (certainly no less than 30). And, ideally, it should be at least 10 per cent of the sample frame (total number of potential respondents in the target area).

For a medium-sized peer education project, it is acceptable that five peer educators (in each site) be trained to conduct between ten to twenty interviews each. Their training should last at least half a working day, with about three days allowed to complete the allotted questionnaires. School students will often do a questionnaire more quickly than out-of-school youth, who may take twice as long. The input and analysis of about a thousand questionnaires into a database programme (such as SPSS) will take one person about ten working days.

There are several options in implementing the questionnaire:

1. Face-to-face, interviewer administered. The interviewer asks the questions and the respondent is not shown them (or the responses in a pre-coded questionnaire). This may be time-consuming but will allow for clarification and probing by the interviewer Most applicable in low literacy populations.

2. Face-to-face, partly administered. This is where the respondent is shown the questionnaire and the potential responses, but the interviewer still fills in the form. The disadvantage of this is that seeing the potential responses on the questionnaire may bias the answers.

3. Self-administered. Here, respondents complete the questionnaire themselves, either on their own with the interviewer present, or in a small group (focus group size), or large group - such as a school class. Questionnaires can be delivered or left with the respondent for later collection, as in a postal survey The advantage of the self-administered method is that a large sample can be reached in a short time. Disadvantages are that any confusion in the respondents may not be picked up by the facilitator (especially in larger groups), leading to higher levels of error in the responses. Another factor is that not all the respondents complete the questionnaires at the same speed, and some may feel rushed or may not finish at all. These questionnaires need to be simpler and carefully pre-tested, to ensure removal of all ambiguities. Self-administered questionnaires are most appropriate in higher literacy groups or when anonymity is essential.

Table 9.2 Data Collection Techniques

Techniques	Nature of procedures,	Specific use in behavioural development/change evaluation	Requirements
Demographic and Health Survey (DHS)	· National, randomly selected sample · Quantitative · 60-minute interview	· Provides national data which acts as a reference · Can assist with the selection of indicators	· Use of DHS data must be done with careful reference to the scale of the sample - how does the data relate to your sample?
Questionnaire (KAPB) survey	· Quantitative · Uses fixed and/or open standardised answers · Some limitations on reliability of answers · 10-15 minutes	· Provides information on self-reported behaviours · Replicated surveys can quantify the extent of change over time (i.e., baseline and follow-up) · Can demonstrate differences between groups	· List of specific indicators/questions · Interviewer's presence not essential · Pre-testing of questionnaire
In-depth interviews	· Can be both quantitative and qualitative · Allows detailed exploration of certain topics · Information more reliable · 60-90 minutes	· Can provide explanations for behaviour patterns uncovered in the KAPB survey · Elicits perceptions of gatekeepers and opinion makers · Can provide new information	· Question guide · Interviewers presence essential · Use of tape recorder (optional) · Need to match age/sex of interviewer and respondent
Focus group discussions	· Mainly qualitative data · Propositions, reactions, explanations, consensus building · Ranking exercises · 60-90 minutes per session	· Perceptions on nature of behaviours · Discussion of why positive behaviours are (not) occurring · Perceived link of reported behaviours to specific interventions) · Examination of perceived constraints on positive behaviours · Provides new information	· Selected, homogeneous group · Presence of moderator · 6-12 participants · Tape recorder necessary · Note-taker (for non-verbal reactions) optional · Need to match age/sex of interviewer and participants

KAPB questionnaires have proved extremely informative, and there is a certain amount of confidence in their value (UNAIDS/WHO 2000):

"Several studies have shown that trends in reported risk behaviour are usually reflected in STI trends, that trends in reported condom use are matched by trends in condom distribution, and that there is a remarkably high agreement between couples when questioned about their sexual behaviour.

"Although survey data on sexual behaviour and knowledge of preventive practices have serious limitations, the use of culturally sensitive population based surveys may still prove a meaningful evaluation method, provided they focus on qualitative conclusions combined with other; small-scale studies. At present, this certainly remains the most realistic option". (Konings and others, 1995)

But KAPB questionnaires and surveys do have various limitations, and it is worth being aware of them.

How long the interview lasts may affect the validity of the answers you receive

In relation to questionnaire type, significant differences appeared between short questionnaires (10-15 minutes' duration) and more in-depth, longer interviews (lasting about 90 minutes). Konings and others (1995), trying to validate population surveys for the measurement of HIV/STD indicators in Kampala, found that the proportion of young people reporting sex with a non-regular partner during the previous 12 months was considerably higher when using in-depth interviews than in short survey questionnaires. The difference between the 2 samples was a considerable 10 percentage points: 23 per cent of the respondents did not report sex with non-regular partners in the questionnaire, but did in the in-depth interview. In fact, only 55 per cent of respondents gave identical answers to the same question in both interviews. In explaining this disparity, most of the interviewers indicated a certain degree of discomfort on the part of the respondent during the short questionnaire. The choice, then, for the evaluation/research designer is between drawing qualitative conclusions from quantitative survey data, or in-depth (i.e., small sample) interviews giving more accurate, but less representative information.

Time-frames for recalling past behaviour should be kept as short as possible, and respondents should be constantly reminded that all answers are anonymous

These two ploys have been found effective in reducing inaccurate recall and avoiding biased answers, in a survey of young people in the United States (Schnell and others, 1996) and a review of the validity of indicators using self-reported behaviours (Catania and others, 1993). Confidentiality and anonymity of the respondent are crucial. In a research exercise with a project in Pakistan this was not possible, as, "although it was preferred that the adolescents reveal their responses in confidence, parents/guardians and sometimes friends were often present during the interview". Unavoidable as this may have been, the resulting bias to the responses is obvious. A significant degree of what is known as 'social desirability bias' would be operating here in terms of the interviewer being told what the respondent feels is the 'correct' answer using social norms as the frame of reference. It may have nothing to do with personal experience.

It is important to record all the answers of a respondent

Sometimes respondents may give more than one answer to a question. Recording these multi-responses is important, and the responses after the initial answer should not merely be categorised as 'other. A separate question can then ask: 'of the sources of information you have mentioned, which is the most useful?' It is also worth being careful in interpretation here - school children are far more likely to get information from a school teacher than a child who does not attend school. It is more helpful to have the percentage of respondents mentioning one particular answer e.g., 52 per cent of respondents mentioned menstruation as a sign of puberty'. The cumulative percentage will be more than 100, but at the bottom of the table put in the note that multi-answering was accepted, Ranking the responses when presenting data is also useful for analysis (Table 9.3).

Table 9.3 Multi-answering in Survey Questions. Which Physical Developments Indicate Maturity?

Physical development	% mentioning*
Menstruation	53
Height	23
Breasts	10
Hair	5
Voice	4
Mind	1
Don't know	23
Missing	4

* multi-answering accepted

Also valuable are 'check' questions. These ask the same question, in two ways, in different parts of the questionnaire. Analysis can compare and contrast these answers and look for contradictions. If contradictory answers are given, respondents either do not fully understand the questions, are deliberately giving false answers, or are genuinely confused about what they have done (or not). A typical example, in surveys of young people, is when they report having used a condom, but then state that they have never had sex. Strictly speaking, these questionnaires should be ignored when analysing these particular questions, and recorded 'void'.

Watch out for 'social desirability bias'

An often mentioned criticism of KAPB questions is that respondents will give the answers 'expected of them'. One way to minimise this is to ask respondents to show knowledge. Children are far more likely to answer 'yes' to the question 'do you know what AIDS is?' if they know that an HIV/AIDS project has been operating. It may be better to ask them to describe what HIV and AIDS are, and record the answers. A better question would be, can you name three ways in which HIV is transmitted?', then ask the respondent to name them. The 'don't knows' will be obvious. If the question is asked without providing option answers, the response is termed as 'spontaneous'. Questions phrased as, 'can having sex without a condom transmit HIV?', are known as 'prompted' responses.

The importance of triangulation is that data from surveys can be presented to focus group participants, of the same age group from the same area, to see if the group feels that the responses are 'accurate'.

Very often the questionnaire is repeated after a period of time: a pre-/post-questionnaire. It worth looking at an example of one of these, designed for school children, linked to a school-based, life-skills programme (Box 9.1).

The twenty-two statements cover a range of knowledge, and include both negative and positive statements. The HIV/AIDS-related questions concentrate solely on knowledge regarding HIV transmission and the latency period, but do not address the means of prevention, attitudes, stigma or intended practices in the future. The questions on self-confidence and assertive behaviour (13-18) are slightly ambiguous, as the words 'content', 'satisfied', 'harmful', 'needed to', and 'comfortable' are highly subjective. These need to be more specific, and each question needs to be linked to the objectives of the education. For example, it is difficult to see how question 13 relates to the education being provided. Question 18 could also be tightened - many people in many countries would use a medicine without consulting a doctor The wording in question 19 may also be problematic - are the phrases 'natural occurrence' and 'some mistake' mutually reinforcing or contradictory? This may confuse respondents -and the results.

Pre- and post-test samples should be directly comparable. To be so, they must be selected in exactly the same way, from the same sample group. There may be the possibility of asking the same questions of children in another area (or school), or perhaps part of the community directly involved in the current project (if the project is due to expand). This can provide essential comparison (control group) data as well as a valuable baseline.

Box 9.1 Example of pre-test/post-test questionnaire with school children

Please put Ö or X to indicate true or false (agree or disagree) Information on AIDS: 1) Good doctors can cure AIDS. 2) A young child can't have AIDS. 3) Touching a person infected with the AIDS virus is harmless. 4) The AIDS virus cannot spread from a husband to a wife and vice versa. 5) One can get the AIDS virus from a blood transfusion. 6) People in your community are at risk of getting the AIDS virus.: 7) AIDS is a fatal condition. 8) Anyone who has the AIDS virus immediately looks ill and weak and gets a I high fever. 9) Muslims are not at risk of getting the AIDS virus. Marriage: 10) A marriage is not valid unless both the bride and groom have given their consent. 11) According to Islam, only the husband has the right to divorce. 12) If her father agrees, it is not necessary for a woman to give her consent for marriage. I Self-Confidence and Assertive Behaviour: 13) Are you content/satisfied with your body? Yes/no/sometimes. 14) Is it all right to talk about your body? Yes/no/sometimes. 15) If you see your friends doing something harmful, do you tell them? Yes/no/sometimes.; 16) If you needed to, would you go to the doctor alone? Yes/no/sometimes. 17) Do you feel comfortable explaining health problems to the doctor? Yes/no/sometimes. 18) In the past month, have you used any medicine without consulting a doctor? Yes/no/maybe. Puberty: True/False 19) Discharge is a natural occurrence, which indicates some mistake on the part of: a boy: 20) Menstruation is a shameful occurrence and should not be discussed with anyone. 21) Menstruation is linked with the woman's ability to have children. 22) Discharge is linked with the male's ability to have children.