SAMPLE SELECTION
Present Practice
The rule up till now at the 3 major markets is that the data collectors must try to ensure that the sample is representative of the total population in terms of grade and sex.
Beyond that, approximately 6% of the population is sampled at Arusha, 8% at Moshi, and 3% at Dar. At Moshi and Arusha, actually 6% and 8% of each market-day is sampled, whereas for Dar (where there are up to 7 market-days per week), then usually only 2 market-days per week are actually sampled; the sample for those 2 market days is thus approximately 10%. For Dar, the rules are actually quite loose, and so long as approximately 100 animals are sampled per week, regardless of the numbers per day or the number of days, then the requirements are regarded as met. Note that these 6%, 8% and 3% samples referred to above are equivalent to 50, 30 and 40 animals respectively per market-day.
The sampling procedure at all 3 sites is that a group of cattle is identified, and then 20% of those cattle are taken for weighing etc etc.. At Dar, in theory at least, the group of cattle are physically driven through a crush and literally 1 in 5 is stopped and weighed etc.; the intention of this very work-intensive method is that a random 20% sample is obtained.
Note that the numbers above refer to NET sample - animals for which both initial and secondary data is available - i.e. weight AND price - for a percentage of animals weighed the price cannot for various reasons be obtained (more below).
The table below (9.2.95-1) shows average gross and net sample sizes for the 3 major markets for the period 1989 - 1994. It appears that 60% is a reasonable conversion factor in practice for gross to net sample. It appears also that 100 animals per market-day is a reasonable maximum gross sample for a once-per-week market, and 70 animals per market-day is a reasonable maximum gross sample size for an every-day-per-week market (based on there being only a single weighing scale at the market).
|
Weight > 0 |
Wt&price>0 |
Market-days |
Avg Net Sample |
%net/gross sample |
|
|
Arusha 94 |
2632 |
1668 |
25 |
67 |
63% |
|
Arusha 93 |
3731 |
1751 |
35 |
51 |
47% |
|
Arusha All |
9483 |
4863 |
97 |
50 |
51% |
|
Moshi 94 |
1795 |
1051 |
22 |
48 |
59% |
|
Moshi 93 |
2535 |
1260 |
32 |
39 |
50% |
|
Moshi All |
8001 |
3676 |
118 |
31 |
46% |
|
Dar 94 |
5361 |
3551 |
80 |
44 |
66% |
|
Dar 93 |
937 |
650 |
21 |
31 |
69% |
|
Dar All |
17376 |
10344 |
282 |
37 |
60% |
Table: Net Samples Sizes and Net/Gross Sample Size Conversion Ratios by Market and Time.