Quasi-random sample selection was chosen as the fastest and most satisfactory sampling method, given the time available to undertake the survey as well as the fact that an adequate sampling frame was not available.
Data from the 1986 Population Census was initially collected for the villages to be surveyed. Estimates were then made of the number of households in each village by dividing the total de jure population with the estimated number of persons per household for lowland areas (Area 1) and Urban Other areas (Area 2) from estimates made by the Household Budget Survey 1986/87. A rough count of households in each village was also undertaken by the survey team to ensure that there was not too much variation between the estimates and the actual number of households.
Once the number of households had been estimated, a sampling fraction of 10% of the total number of households in each village was agreed upon as a realistic sample size given the time available. The first household at one end of the village was selected as the starting point and then every tenth household was interviewed. Where households were on both sides of a track, the tenth household was selected by counting left to right. The head, or the wife of the household head, was interviewed. Where there was nobody at home, it was agreed that a household on either side (i.e the ninth or eleventh) would be selected as a replacement.
Sample Selection by Area
Ha Ramapepe (Area 1)
Ha Ramapepe consists of 5 villages (excluding the Outward Bound Centre) with the following de facto and de jure population according to 1986 Population Census figures:
| De Facto | De Jure | |||||
|---|---|---|---|---|---|---|
| Male | Female | TOTAL | Male | Female | TOTAL | |
| Thaba-Bosiu | 22 | 21 | 43 | 29 | 28 | 57 |
| Liboping | 279 | 361 | 640 | 392 | 408 | 800 |
| Ha Khantei | 10 | 12 | 22 | 11 | 14 | 25 |
| Sentelina | 260 | 357 | 617 | 364 | 414 | 778 |
| Likiiling | 55 | 67 | 122 | 80 | 81 | 161 |
| TOTAL | 626 | 818 | 1 444 | 876 | 945 | 1 821 |
Only households in the villages of Thaba-Bosiu, Libobing and part of Sentelina were included in the survey because there was limited time available. The criteria for their selection was their relative proximity to the reservoir and because they were very near each other and could be regarded as different areas of the same village. The estimated number of households in these villages were:
| Denominator Av.Household Size:lowlands* | De Jure Population | Est.No. of Household | |
|---|---|---|---|
| Thaba-Bosiu | 5.27 | 57 | 11 |
| Liboping | 5.27 | 800 | 152 |
| Sentelina | 5.27 | 778 | 148 |
| TOTAL | 1635 | 311 |
* Household Budget Survey 1986/87
Only half of the village of Sentelina was included in the survey, so the total number of households from which the sample would be selected was 237. Thus, 24 interviews should have been undertaken, but only 18 were carried out.
Thaba-Patsoa (Area 1)
Thaba-Patsoa consists of 4 villages with the following de facto and de jure population according to 1986 Population Census figures:
| De Facto | De Jure | |||||
|---|---|---|---|---|---|---|
| Male | Female | TOTAL | Male | Female | TOTAL | |
| Jorodane | 63 | 73 | 136 | 88 | 80 | 168 |
| Ha Rantsane | 47 | 35 | 82 | 60 | 42 | 102 |
| Tarabane | 95 | 99 | 194 | 130 | 117 | 247 |
| Sehlajaneng | 43 | 55 | 98 | 65 | 63 | 128 |
| TOTAL | 48 | 262 | 510 | 343 | 302 | 645 |
Households in Thaba Patsoa are more scattered then in Ha Ramapepe. Households in the villages of Jorodane, Tarabane and Ha Rantsane were included in the survey as these were the villages closest to the reservoir. It was discovered, however, that households in Ha Rantsane do not use the main reservoir, but the smaller one nearby. The following number of households were estimated for the selected villages:
| Denominator Av.Household Size:lowlands* | De Jure Population | Est.No. of Households | |
|---|---|---|---|
| Jorodane | 5.27 | 168 | 32 |
| Ha Rantsane | 5.27 | 102 | 19 |
| Tarabane | 5.27 | 247 | 47 |
| TOTAL | 517 | 98 |
Although a sample size of 10% of the total number of households ( i.e. 10 households) was originally planned, the availability of time led the survey team to increase the sample size to 22, thus covering approximately 20% of all households.
Overall, therefore, in Area 1, 40 interviews were conducted, which represents 10% of the estimated number of households in the villages selected for survey.
Roma Valley Villages (Area 2)
Four villages were selected for participation in the survey in the Roma Valley for the following reasons:
all are close to 5 reservoirs of varying sizes
one village, Ha Mafefooane is part of Roma centre and can be considered as a “representative” village for Roma centre
the remaining three villages are close to Roma centre, but are only accessible by foot or by vehicles with good ground clearance.
The villages have the following estimated de facto and de jure populations according to the 1986 Population Census:
| De Facto | De Jure | |||||
|---|---|---|---|---|---|---|
| Male | Female | TOTAL | Male | Female | TOTAL | |
| Ha Mafefooane | 186 | 217 | 403 | 213 | 230 | 443 |
| Tleutle Ha Mpiti | 156 | 167 | 323 | 192 | 187 | 389 |
| Ha Shale | 192 | 231 | 423 | 254 | 254 | 508 |
| Pae-Lea-Ithatsoa | 121 | 145 | 266 | 152 | 152 | 304 |
| TOTAL | 655 | 760 | 1 415 | 811 | 823 | 1 634 |
Based on the above figures, the estimated number of households for the selected villages were:
| Denominator Av.Household Size:urban other* | De Jure Population | Est.No. of Households | |
|---|---|---|---|
| Ha Mafefooane | 4.48 | 443 | 99 |
| Tleutle Ha Mpiti | 4.48 | 389 | 87 |
| Ha Shale | 4.48 | 508 | 113 |
| Pae-Lea-Ithatsoa | 4.48 | 304 | 68 |
| TOTAL | 1634 | 367 |
However, the survey team felt that Ha Mafefooane was much larger village (in terms of number of households) than that suggested by the estimates. By undertaking a rough count of households, it was estimated that the number of households was almost double that of the estimate made using Population Census data. It was therefore decided to double the estimate of households in Ha Mafefooane to approximately 200 households.
A 10% sampling fraction was also used for Area 2, and the following number of interviews were conducted:
| No. of Interviews | |
|---|---|
| Ha Mafefooane | 21 |
| Tleutle Ha Mpiti | 10 |
| Ha Shale | 10 |
| Pae-Lea-Ithatsoa | 11 |
| TOTAL | 52 |
Sample Size for the Sesotho Questionnaire
The method of distribution of the Sesotho questionnaire made it impossible to use any form of sampling, as forms were distributed to those who attended the village pitso. In Ramapepe, because there were 24 persons attending, it was decided to give every second person a form, thus distributing 12 questionnaires (7 men, 5 women). Six questionnaires were returned.
In Thaba Patsoa, as participants were so few when questionnaires were distributed, it was decided to give all 15 attendants participate a form, but as two more joined the meeting, it was not possible to omit them. Thus 17 questionnaires were distributed (11 men, 6 women) and 15 returned.
