The Sampling Design and the Data Collection Design
Chapter -Nine
Syllabus topic 7, "Preparation of the Research Design"
Pages 499 to 501 of 543
In one line
The sampling design says who will be studied and how they will be chosen, and the data collection design says what will be collected from them, by what instrument, by whom and when.
In the wording a student can write in an exam: the sampling design determines the population, the sampling frame, the sampling unit, the method of selection and the size of the sample, together with the treatment of non-response; the data collection design determines the tools to be used, the items of information to be collected, the manner and sequence of administration, the persons who will collect the data and their training, and the record that will be kept, and both are settled before collection begins and stated in the research methodology.
The sampling design, as a written statement
Six lines, and a student who can write them has a sampling design.
The population, exactly, with place and period.
The sampling frame, and what it omits, chapter 1000.
The sampling unit.
The method, named and justified: simple random, systematic, stratified, cluster or multi-stage, chapter 1010; or one of the non probability methods with the consequences stated, chapter 1020; or a census, chapter 990.
The size, with the reason.
And the treatment of non-response: what will be recorded, and the rule about substitution, which should be that there is none.
Where it goes. Inside the Research Methodology component, chapter 1310.
The data collection design, as a written statement
Seven lines.
The items of information, listed. This is the design decision that determines everything, because an item not on the list will not be collected, chapter 1200's empty tables test.
The tool or tools, chapter 960 and chapter 970.
The instrument itself, drafted and pre-tested, appended.
Who collects, and how they are trained if more than one.
When and where, including times of day, which matters when the respondents work.
The consent procedure, chapter 1400.
And the record: what is kept, in what form, for how long, and where.
Choosing the items of information
The most important step and the least discussed.
Start from the empty tables, chapter 1200. Every cell requires an item.
Add the classification variables. Sex, age group, occupation, settlement, whichever the analysis will break down by. Omitting one of these is the commonest cause of a dissertation that cannot answer its own subsidiary questions.
Add a little redundancy. One or two items that are not strictly required but may prove informative. They are cheap at collection and impossible afterwards.
And add the open closing item. Is there anything I should have asked about, chapter 960.
Then stop. Every additional item costs respondent patience, and a schedule that runs to twenty minutes will be abandoned or answered carelessly.
The Sampling Design and the Data Collection Design
The two designs together, with an example of them failing
A study of clinic users, badly designed.
Sampling. Users interviewed on the days the researcher could attend. That is convenience sampling, chapter 1020, and the result describes those users and not the clinic's users.
Data collection. A schedule recording awareness and satisfaction, but not the distance travelled or the number of visits.
The consequence. The dissertation's most interesting possible finding, that users travel a long way and come repeatedly, cannot be reported, because those items were never on the list. And the proportions cannot be generalised to the clinic's users at all.
The same study, well designed.
Sampling. The clinic's register under regulation 20 of the 2011 Regulations is the frame, chapter 600; every fifth entry from a random start; ninety users; non-contacts recorded.
Data collection. Items: the classification variables; the problem brought; how they heard of the clinic; distance travelled; number of visits; whether a lawyer was assigned; what happened; and the open closing question.
The consequence. Every table in the design can be filled, the sample supports a statement about the clinic's users, and the unexpected findings have somewhere to appear.
The difference is two pages written in advance.
Sampling design for a doctrinal study
Students assume there is none, and there is.
The population is the body of material: which instruments, which courts, which period of decisions.
The frame is what is actually available: a database's coverage, which is never complete.
The selection is a decision: all decisions of the Supreme Court on a point, or all reported decisions, or all decisions in a stated period.
And the omission must be stated, because a database that does not carry unreported High Court decisions has omitted a category, and a study that does not say so implies a completeness it does not have.
A worked example
The remand study's two designs, written out.
Sampling design. Population: every person produced for remand in the named court between two stated dates. Frame: the court's own daily proceedings, observed directly, so the frame is the population and nothing is omitted except sittings the observer missed, which are recorded. Unit: the individual production. Method: a complete count, so no sampling. Size: whatever the period produced, being 217. Non-response: not applicable to observation, but two missed sitting days are recorded.
Data collection design. Items: date; sitting; number produced at that sitting; whether anything was said concerning free legal services; by whom; whether the accused was represented; time taken; and a free note. Tool: structured non-participant observation with a one page schedule, appended. Collector: the researcher alone, so no inter-observer check. When: every sitting day, from the start of the list. Consent: not required, the court being open, and no individual is identified. Record: the completed schedules, retained.
The Sampling Design and the Data Collection Design
Twelve lines, and the study can be executed by anybody.
Quick revision
Sampling design, six lines: population with place and period; frame and what it omits; unit; method, named and justified, or a census; size with reason; and non-response, with no substitution.
Data collection design, seven lines: the items of information, listed; the tool; the instrument, pre-tested and appended; who collects and their training; when and where, including time of day; the consent procedure; and the record kept.
Choosing items: start from the empty tables; add the classification variables, whose omission is the commonest cause of unanswerable subsidiary questions; add a little redundancy; add the open closing question; then stop.
A doctrinal study has a sampling design too: which instruments, which courts, which period, what the database omits.
Test yourself
1. Give the six lines of a sampling design. The population, stated exactly with place and period; the sampling frame and what it omits; the sampling unit; the method of selection, named and justified, or a statement that the study is a census; the size with the reason for it; and the treatment of non-response, including the rule that respondents are not substituted.
2. Give the seven lines of a data collection design. The items of information to be collected; the tool or tools; the instrument itself, pre-tested and appended; who will collect and how they are trained; when and where, including times of day; the consent procedure; and the record to be kept, in what form and where.
3. How should the items of information be chosen? By starting from the empty tables the analysis will need, so that every cell has an item; adding the classification variables the analysis will break down by; adding a little redundancy, since items are cheap to collect and impossible to add later; adding an open closing question; and then stopping, because every extra item costs respondent patience.
4. What is the sampling design of a doctrinal study? The definition of the body of material: which instruments, which courts and which period of decisions; the frame actually available, such as a database whose coverage is never complete; the basis of selection, whether all decisions of a court on a point or all reported decisions in a period; and a statement of what the source omits, such as unreported decisions.
The rest of this subject
These notes are cut from the University's printed syllabus. Open the syllabus itself, or the past papers, for the same subject.