The 1993 Jamaica Contraceptive Prevalence Survey (CPS) is the fifth in a series of periodic enquiries conducted by the National Family Planning Board (NFPB).
The Survey seeks to update measures of fertility and contraceptive use among women aged 15 - 44 years and will for the first time include a special module for young adults (male and female) aged 15-24 years as well as males aged 15-54 years.
This CPS, coming as it does in the last decade of the century, is of significance to the NFPB in particular and the wider community in general, as it heralds the beginning of the twenty first century and the realization of the goals of Jamaica's National Population Policy. It also comes against the gradual phased withdrawal of contraceptive procurement by the major funding agency, the United States Agency for International Development (USAID), by a twenty percent (20 percent) annual decline over the period 1993-1998 under the Family Planning Initiatives Project (FPIP), as well as the phased diminution of funding from other donor agencies such as the United Nations Fund for Population Activities (UNFPA).
This CPS is in fact one of two surveys to be conducted during the life of the FPIP. Against this background, the NFPB has many challenges ahead which are, inter alia, not only to maintain but also to increase contraceptive prevalence and to achieve further milestones by the inception of the twenty-first century, such as a population of not more than 2.7 million and replacement level fertility of two children per woman. For contraceptive methods and family life services to impact on fertility and contribute to the processes of national development, it is vital that programme effectiveness be evaluated. The reliable and current data collected from the CPS will be of invaluable use in policy analysis and programme implementation for administrators and planners, not only in health but in those areas which impact on population issues at the broader national level.
The scope of the survey, as in earlier studies, is designed to gather information on a broad range of areas including knowledge, attitudes and practices in contraception; perceptions on the role of men and women, including views on sexuality, child bearing, child rearing and health care.
Producers and sponsors
National Family Planning Board
Statistical Institute of Jamaica (STATIN)
Ministry of Health
Division of Reproductive Health, National Centre for Chronic Disease Prevention and Health Promotion, Centres for Disease Control and Prevention
Technical assistance in the areas of survey design and sampling, questionnaire development and training, data processing and report preparation.
National Family Planning Board
United States Agency for International Development
The 1993 Jamaica Contraceptive Prevalence Survey utilized the design adopted for the Continuous Social and Demographic Surveys conducted by the Statistical Institute of Jamaica. This design is based on a two-stage stratified sample with the first stage being a selection of geographic areas and the second stage, of dwellings. For the selection of the first stage units, the country has been divided into enumeration districts (EDs) which are grouped into sampling regions consisting of a predetermined number of strata, approximately equal in size (where size is measured by the number of dwellings in each sampling region). Two EDs within each sampling region, selected with probability proportionate to size (determined by the number of dwellings), make up the sample at the first stage. At the second stage, a second predetermined number of dwellings are selected systematically from lists of dwellings arranged on a circular basis in each of the EDs designated in the first stage of selection.
The selection process for the 1993 Contraceptive Prevalence Survey involved a refinement of the general sample design of the household surveys conducted by STATIN. First stage sample selection is generally made by parishes, thus providing independent parish samples which allow for the calculation of major characteristics in the sample. In the 1993 CPS, the decision was taken that, based on the need to obtain an affordable sample size, the lowest level of disaggregation would be at the health region and not the parish level. Accordingly, the design provided for independent samples to be selected, with the lowest geographical level being the health region. Secondly, in the general sample design used by STATIN, the first stage sample is selected using identical sampling fractions in each parish; in the 1993 CPS, sampling fractions were varied to take account of the significant disparity in the population in the four health regions. This design was applied to ensure appropriate minimum levels of representation in the smaller health regions population-wise, while reducing oversampling in the more populated health regions. The modified sample design was adopted in an identical way for both the survey of males and that of females. In the effort to achieve cost-effectiveness, it was decided to use the EDs selected for the first stage sample of the Labour Force Survey carried out by STATIN as the basis for the first stage selection of the 1993 surveys, modified in accordance with the description stated above. This ensured that the creation of lists of dwellings prepared for the labour force surveys would be available at no extra cost for use in the 1993 CPS. The modification applied at this stage was to reduce scientifically by twenty five percent the number of EDs in the parishes comprising Health Region, which were selected for the labour force surveys. The numbers of EDs selected in the remaining three health regions were maintained in the 1993 CPS.
The determination of the numbers of dwellings to be chosen at the second stage of selection also took into account the size of the health regions to be surveyed. Thus, the numbers of dwellings chosen in each ED in the four regions were 15, 23, 19 and 15 in Health Regions 1, 2, 3 and 4, respectively. Based on these selections, and taking into consideration expected non-response rates, it was anticipated that a total number of 3,500 responses would be obtained in each of the two surveys.
It should be noted that, as described above, this is not a self-weighting sample design and smaller health regions have been oversampled. In addition, one respondent per household was selected with probability inverse to the number of eligible respondents in the household. Thus, results are based on weighted data.
The household and individual status of interviews for both female and male samples are shown in Table I-2.6.1 of the Administrative Report. The profile of household selection categories is similar for both males and females. In both samples, slightly more than 50 percent of households (females - 53.9 percent; males - 54.6 percent) had an eligible respondent while residents were hot at home in three percent of households. In households with eligible female respondents, the percentage with complete interviews was disappointingly lower than in previous surveys. The completion rate of 81.9 percent compares with a completion rate of 94.6 percent achieved in the 1989 CPS.
The refusal rate was higher than in past surveys - 7.0 percent compared with 2.2 percent in 1989. In male households, a similar proportion, 80.0 percent of eligible respondents, were interviewed, compared with 95.7 percent of males successfully interviewed in the 1987 Young Adult Reproductive Health Survey (YARHS). The refusal rate was comparable to the female sample - 6.7 percent.
At the national level, in spite of the lower completion rates for males (80.0 percent) and females (81.9 percent), the age distribution of respondents with complete interviews appears to be representative of the population. As seen in the following chapter, the age distribution of men and women with complete interviews closely approximates the end of year population estimates for 1992 produced by the Statistical Institute of Jamaica. For females, the only age group with more than a one percentage point difference between the CPS and the population estimates is the 35-39 year age group (11.7 percent versus 12.9 percent). However, since the 95 percent confidence interval for the 12.9 percent figure is 1.4 percent, the difference is not statistically significant. For males, the only age group statistically different from the end of year estimates for 1992 is the 15-19 year old age group. After taking into account sampling error, the percentage of teenage males in the sample is about 1.5 percentage points higher than expected. This was unexpected as sample surveys usually under-represent teenage males due to their mobility within the population.
Household completion rates and individual refusal rates appear to have been affected by two events that occurred during the survey. First soon after the survey field work began, a general election was called. As is customary in Jamaica, interviewing was suspended during the 30 day election period. Refusal rates were higher in the Kingston Metropolitan Area (part of Health Region 1) where the election campaign was more intense. Secondly, unseasonal early heavy rains in some parishes were devastating, resulting in landslides and flooding, especially in Health Regions 2 and 4. The rains led to an increase in demolished buildings and/or eligible respondents temporarily moving to other dwellings while their houses were being repaired.
The percentage of households with eligible respondents in the female sample ranges from 48 percent in Health Region 3 to 59 percent in Health Region 1 and in the male sample, the corresponding percentages range from 49 percent in Health Region 3 to 58 percent in Health Region 1. The pattern in each health region is similar for both samples. In both samples, there is an elevated proportion of households in Health Region 3 in the "other" category, which includes demolished households. Further analysis (not shown here) indicates the problem area to be in the parish of St. James. In this parish, 15 percent of female households and 21 percent of male households were in the "other" category compared with no more than 1 to 6 percent in the other three parishes in this health region. A sigificant factor in this finding is that there were a number of households not contacted during the visits, probably because of the nature of the employment is this parish, one of the main tourist areas in the country. Accordingly, it was not possible to identify if the households not located fell in the sample since the information to determine eligibility could not be obtained.
The higher refusal rates in Health Region 1 and eligible respondents not at home in Health Region 2 are noteworthy. In Health Region 1, the refusal rate in St. Andrew was 16 percent for both females and males. This parish, which forms part of the Kingston Metropolitan Area, includes a high proportion of both lower and upper socio-economic areas. In the lower socio-economic areas, the political climate is especially volatile. At the same time, in the upper socio-economic areas, there is a growing resistance to providing information to government interviewers in all of the social surveys being conducted.
Respondents not at home (after several revisits) in Health Region 2 was a particular problem in Portland, due to the heavy rains and in St. Ann, another tourist area with characteristics similar to those described for St. James. The problem was more severe for males in both parishes than for females; 28 percent versus 22 percent and 36 percent ersus 21 percent, respectively. The third parish in this region, St. Mary, did not have a serious revisit problem despite the heavy rains which fell in that parish but had higher refusal rates than the other two parishes (about 12 percent for both females and males).
All national results in this final report have been weighted to compensate for these differentials in non-response as well as the oversampling of smaller health regions and selection of one respondent per household.
Dates of Data Collection
Data Collection Mode
Data Collection Notes
RECRUITMENT AND TRAINING
The Statistical Institute of Jamaica had responsibility for the conduct of the field work and for the coding and editing of the questionnaires and also for data entry. Accordingly, they were in charge of recruitment of field staff, editor/coders and data entry staff. Due to the sensitive nature of the questions included in the survey, a decision was taken to use only female interviewers in the female survey; in the male survey, both male and female interviewers would be used. Thus, females selected for interview would be questioned by females only; males would be interviewed by males or females.
In the case of the interviews of males in the male survey, experience has shown that female interviewers are just as or even more effective than male interviewers. Coders, editors and data entry clerks are not genderspecific although the office clerks working with the supervisors who were included in the training were all female.
Training on field work in both the pretest and the main surveys was the responsibility of STATIN while that on the technical aspects including the pretest was undertaken by a team of three local trainers headed by the Survey Director, supported by a team from CDC. All trainers were professionals who have had extensive training and experience in the field of demography, epidemiology and survey taking.
Officers from the National Family Planning Board including the Medical Director assisted in the training on contraceptive methods. In the areas of coding and editing, the training was provided by the Survey Director while CDC had direct responsibility for training in data entry, using updated SURVEY software developed by them for data entry and editing on microcomputers. Training for the field work was undertaken in two phases. The first related to the pretest of the questionnaires and to procedures to be implemented during the survey programme and was carried out during the period January 12-22, 1993. The second training for the main surveys was carried out over the period February 15 to March 1, 1993. Three training classes were held in Kingston, Black River and May Pen during the first week of the training (February 15-19, 1993) while another was held in Kingston the following week. Due the postponement of field work in some parts of the country as a result of the holding of an election at the end of March, retraining of some twenty interviewers was undertaken by the Survey Director. In all cases, the training consisted of classroom lecturers, discussions, mock interviews and written tests. Field interviews were also carried out as part of the training with the completed questionnaires reviewed in the classroom. A total of 142 field personnel were trained. Of these, there were 23 supervisors (15 males and 8 females) and 107 interviewers (29 males and 78 females). Twelve office clerks, all females, were included in the training.
The organization of the field staff for administering the CPS surveys was similar to that used for all other household surveys conducted by STATIN. The country is divided into four contiguous, non-overlapping areas, each of which is managed by a senior supervisor. Within each area, there are four zones, each covering approximately 28 Primary Sampling Units. One supervisor and five interviewers are assigned to each zone.
Field work for the pretest was scheduled to be conducted in a few pre-selected areas and was scheduled to last three days. This was executed according to schedule. In the case of the main surveys, the original schedule identified three months for both surveys. There were, however, a number of interruptions to the schedule, starting with the calling of a general election for the end of March, 1993 which resulted in a postponement of the start date in some volatile areas and a suspension in others where enumeration had commenced. Full work resumed in all areas in April but within a few days, heavy rains began in some parishes, causing a further cessation. Parishes affected were St. Thomas, Portland, St. Mary, St. Ann, Trelawny, St. James, Clarendon and St. Catherine.
The highly volatile political climate, difficulties experienced in carrying out the field visits, reluctance of a relatively high proportion of persons to cooperate, not entirely unexpected in the context of the disruption to their normal routine caused by the floods, resulted in a longer than expected period of time needed to establish contact, both with householders and selected respondents as well as higher than expected refusal rates. Accordingly, in order to achieve the target for completed number questionnaires set, field work had to be extended beyond the scheduled time. Field work officially ended on August 31, 1993.
Four separate survey instruments were developed for use in the 1993 Contraceptive Prevalence Survey programme; one household questionnaire and one individual questionnaire for both male and female surveys.
The household questionnaires (Form CPS 1A for females and Form CPS 1B for males) were designed to record information on all members of the household and included the number of persons living in the household, name, sex, age and educational standard of all household members, identification of the head of household and relationship of all other members to the household head. Information on age provided the basis for identifying eligible residents, allowing for selection of respondents based on the predetermined age criteria, that is, females between the ages of 15 to 44 years in the female survey and males between the ages of 15 to 54 years in the male survey. Data on occupation of the head and number of rooms occupied by household members were also included.
The individual questionnaires, on the other hand, (Form CPS 2 for females and Form CPS 3 for males) were developed to capture the data on the selected respondents in each of the surveys. Each of the individual questionnaires were divided into eight sections with seven of the eight being identical except for the fact that each has been genderized.
The eight sections in both of the surveys are:
A. Respondent's background
B. Union status and partnership history
C. Fertility (female survey) or reproductive history (male survey)
D. Family planning
E. Attitudes toward contraception and sexuality
F. Young adult module
G. Current sexuality
H. Knowledge of AIDS and its transmission and prevention.
The individual questionnaires were developed to provide comparisons with earlier studies including the 1975/76 Jamaica Fertility Survey, the 1983 and 1989 Contraceptive Prevalence Surveys and the 1987 Young Adult Reproductive Health Survey. In addition, the evaluation of the 1989 Survey by staff members of the National Family Planning Board and the Ministry of Health contributed to the introduction of the new areas of coverage.
Consultations on the content of the survey were held between the Division of Reproductive Health, Centers for Disease Control and Prevention (whose core questionnaires for family planning, maternal-child health and young adult reproductive health surveys provided guidelines), the National Family Planning Board, the Ministry of Health, United States Agency for International Development and the Survey Director, and on the more technical aspects of the survey design, with the Statistical Institute of Jamaica.
Estimates of Sampling Error
The estimates for a sample survey are affected by two types of errors: (1) sampling error, and (2) non-sampling error.
Non-sampling error is the result of mistakes made in carrying out data collection and data processing, including the failure to interview the correct household, errors in the way questions are asked or understood, and data entry errors. Although quality control efforts were implemented during the design of the survey in order to minimize this type of error, non-sampling errors are impossible to avoid completely and are difficult to evaluate statistically.
Sampling error is defined as the difference between the expected value for any variable measured in a survey and the value estimated by the survey. Sampling error is a measure of the variability between all possible samples that could have been selected from the same population using the same sample design and size. For the entire population and for large subgroups, the sample size in the Jamaica survey is large enough that sampling error for most estimates is small. However, for small subgroups, sampling errors are larger and may affect the reliability of the estimates. Because the statistics presented in this report are based on a sample, they may differ by chance variations from the statistics that would result if all women 15-44 years of age or all men 15-54 years of age in Jamaica had been interviewed. The standard error of an estimate (or sampling error) is a measure of such differences. The standard error can be used to calculate confidence intervals for estimated statistics.
The estimated sampling errors for selected percents and sample sizes are shown in Table A-l of the Administrative Report. The chances are about 68 out of 100 (about two out of three) that sample estimate would fall within one standard error of a statistic based on a complete count of the population. The chances are about 95 in 100 that a sample estimate would fall within two standard errors of the same measure obtained if all people in the population were interviewed.
To obtain the sampling error for percents or sample sizes not shown in the table, one may interpolate. For example, for a sample size of 200 and an estimate of 25 percent, the sampling error would be 3.55 percent, halfway between the values of 3.3 and 3.8 for 20 and 30 percent, respectively; for an estimate of 40 percent and a sample size of 350, the sampling error would be 3.05 percent, halfway between the values for 300 and 400.
Statistical differences between percents discussed in this report were found to be statistically significant at the 5 percent level using a 2-tailed normal deviate test. This means that in repeated samples of the same type and size, a difference as large as the one observed would occur in only 5 percent of samples if there were, in fact, no difference between the percents in the population.
In the text, terms such as "greater," "less," "increase," or "decrease" indicate that the observed differences were statistically significant at the 0.05 level using a 2-tailed normal deviate test. Statements using the phrase "the data suggest" indicate that the difference was significant at the 0.10 (10 percent) level but not the 0.05 (5 percent) level. Lack of comment in the text about any two statistics does not mean that the difference was tested and found not to be significant.
The relative standard error (or coefficient of variation) of a statistic is the ratio of the standard error to the statistic and usually is expressed as a percent of the estimate. In this report statistics with a relative standard error of 30 percent or more are generally indicated with an asterisk (*). These estimates may be viewed as unreliable by themselves, but may be combined with other estimates to make comparisons of greater precision.
Disclaimer and copyrights
The user of the data acknowledges that the original collector of the data, the authorized distributor of the data, and the relevant funding agency bear no responsibility for use of the data or for interpretations or inferences based upon such uses.