of 0.05 determined the sample size. Power of 0.80, an effect size
of 0.25, and a significance level of 0.05 are generally viewed as
adequate and standard because many statisticians use diverse
figures based on type of studies.28–31 However, a decision to
oversample was reached based on information from the district’s
health information analyst regarding the death rate related
to HIV/AIDS among the target study population.
Therefore, a population sample of 700 was considered large
enough because it had a surplus of 300 participants. The VCWs
generated the names of all femaleswho met the study inclusion
criteria. The nameswere entered into the computer for random
selection after determining the proportionate samples.
Proportionate sample allocation
The population sampling fraction was derived from the
formula
f ¼ n=N;
which is equal to 700/3,052 = 0.23.30 Sample sizes within
strata were obtained by multiplying the size of each by the
sample fraction, yielding sample sizes of 121 farm, 50 mine,
350 traditional rural reserve, and 179 resettlement villages, for
strata 1 through 4, respectively. In proportionate allocation,
the use of a uniformsampling fraction in each stratumgenerally
attempts to produce a ‘‘population representative sample.’’30
Of the 700 subjects drawn for the study, 48 participated in
six focus groups of 8 females from all four sample strata and
were excluded from answering the questionnaire. The 48 females
were excluded from answering the final questionnaire
because they had already used the same instrument during the
focus group discussion to revise the vocabulary and terminology
used in the questionnaire for cultural effectiveness. Of
the remaining subjects, 514 were interviewed and completed
the questionnaire. Of those who completed the questionnaire,
91 were from farm, 34 from mine, 270 from traditional rural
reserve, and 119 from resettlement villages (Table 1). One
hundred of the rest were reported to have died, and 38 were
absent from their homes at the time of the study. Of those who
were reported dead or did not respond to the study invitation,
22 were from farm, 8 from mine, 56 from traditional rural
reserve, and 52 from resettlement villages. There were approximately
1–2 weeks between the time of invitation and
administration of the questionnaire. For those subjects who
had given consent to participate and were ill, mostly from
HIV-related conditions, 2 weeks was a long time to wait.
Oversampling was done in order to complete statistical
procedures with adequate power. The researchers had no idea
from day to day about how many invited subjects would have
died or were likely to report for the interviews because the
death rate from HIV/AIDS-related diseases was on the increase
in that district at the time of data collection.
Village types
Farm villages. Farm villages consist of a collection of mud
huts thatched with elephant grass arranged in compounds
where farm workers live. These village compounds usually
lack basic facilities, such as purified water and sanitation.
There is no electricity; villagers use firewood and kerosene gas
for energy. Commercial farms are owned by generations of
former European settlers and affluent black Zimbabweans.