Observation
Observation
Observation (^)
Observation is:
- focusing of attention on some entity
- eliciting of information from and about that entity
The aspects of Observation here examined concern:
- Witnessing
- Measuring
- Questioning
Witnessing (^)
Witnessing (^)
Definition
Witnessing is carefully noticing, precisely recalling and clearly recounting some Data or Facts concerning an entity.
A witness can be:
- - a detached observer
- - a participant observer
- - a composite of both
Witnessing : requirements (^)
A witness should possess a series of abilities in order to make the outcome of the witnessing highly reliable and effective.
They could be summarized as the abilities to:
- detect (sensation)
- interpret (perception)
- represent (conception)
- recall (recollection)
- recount (exposition)
Measuring (^)
Measuring (^)
Definition
Measuring is producing Measurements by assigning Numerals (i.e. numerical symbols) to variables, according to rules.
Classification
Measuring can result in:
- Coding (qualitative identification of variables)
- Ranking (positional level of variables)
- Counting (quantitative amount of variables)
Measurements (^)
Definition
Measurements are the result of Measuring.
Classification
Measurements can be classified into:
- Descriptive Measurements e.g. averages; percentages; dispersion and standard deviation; correlation amongst variables; etc.
-
Predictive Measurements e.g. extrapolations; probability distributions; etc.
Numerals (^)
Definition
A numeral is a numerical symbol. (e.g. Arabic system : 3 ; Roman system : III ; binary system : 11)
Classification
Numeral can be classified according to their function :
| - Nominal | e.g. 1 = female ; 2 = male (function : coding ; property : qualitative identity) |
|
|
|
| - Ordinal | e.g. first, second, third, etc. (function : ranking ; property : positional order) |
|
|
|
| - Cardinal | e.g. 1, 2, 3, etc. (function : counting ; property : quantitative operability) |
Scales (^)
Definition
A Scale is both a tool for measuring and a way of organizing what has been measured.
Classification
- Nominal Scale : assigning identity (e.g. labelling individuals)
- Ordinal Scale : assigning identity and generic ranking (e.g. labelling individuals and ordering them according to some parameter, for instance 'weight')
- Interval Scale : assigning identity and specific relative ranking (e.g. labelling and ordering individuals according to their relative differences)
- Ratio Scale : assigning identity and specific absolute ranking (e.g. labelling and ordering individuals according to their differences with respect to an absolute point). A ratio scale allows one to perform all arithmetic operations on variables (summation, subtraction, etc.)
- Multi Index Scale : compound of various measurements (e.g. miles per gallon)
Measuring : errors (^)
Errors in measuring can originate from :
- Suitability of measuring instrument : does the instrument measure what it is supposed to measure (e.g. is the question, concept, tool, etc. appropriate for the problem under investigation?).
Lack of suitability leads to systematic errors.
- Reliability of the measuring instrument : is the instrument reliable in all its measurements (e.g. will the same result be obtained by another experimenter in another experiment using the same instrument?).
Lack of reliability leads to random errors.
Questioning (^)
Questioning (^)
Definition
Questioning is the formulation and submission of questions in order to obtain Information.
Classification
Questioning can be performed through:
- Questionnaire(s)
- Interview(s)
Questionnaire (^)
- Method of administration
- Format of question
- Requirements (pre-administration)
- Requirements (post-administration)
Questionnaire (^)
Definition
A questionnaire is an observation-survey tool for eliciting Information relevant to the problem at large.
Classification
Questionnaires can be differentiated according to:
a) method of administration (direct - indirect)
b) format of question (open - closed)
Method of administration (^)
A questionnaire can be administered to respondents through:
a) Direct Contact (face-to-face, phone)
| Advantages | - full coverage of respondents - higher completeness and reliability of replies |
|
|
|
| Disadvantages | - costly in terms of people and time |
b) Indirect Contact (mail)
| Advantages | - cheaper in terms of people employed and time allocated |
|
|
|
| Disadvantages | - slow and partial return - incomplete replies - risk of possible general bias caused by non respondents |
Format of question (^)
A questionnaire can be:
a) Closed : each question accepts a specific answer selected from a pre-defined range of possibilities
| Advantages | - easier to administer - easier to tabulate and analyse |
|
|
|
| Disadvantages | - information has limited depth and shades of discrimination |
b) Open : each question accepts a variety of answers, without restriction as to length or depth
| Advantages | - more in-depth information (possible discovery of hidden aspects) |
|
|
|
| Disadvantages | - more difficult to tabulate and analyse - requires more time and thought on the part of the respondents - can lead to important omissions if compiled without sufficient care and attention |
Questionnaire : requirements (pre-administration) (^)
The construction of a questionnaire demands specific requirements concerning questions and answers, as for instance:
Questions
- brief questions
- honestly formulated (avoid emotionally charged words)
- easily understood (avoid words with difficult or ambiguous meaning)
- clearly presented (neat layout)
- logically interrelated and sequenced
Answers
- requesting simple answers
- relevant to the information sought
- easy to tabulate and analyse
Questionnaire : requirements (post-administration) (^)
The administered questionnaire needs to be checked for:
a) Internal consistency of replies : do the replies fit with each other ?
If not why ?
- questions badly formulated ?
- respondent unwilling or unable to clarify some point ?
- other ?
b) External congruency of replies : do the replies fit with other sources of information ?
If not why ?
- questions badly formulated ?
- respondent unwilling or unable to clarify some point ?
- external sources unreliable ? out-of-date ?
- other ?
Interview (^)
Interview (^)
Definition
An interview is a focused or semi-focused conversation aimed at obtaining some pertinent information from a respondent.
Classification
Interviews can be classified according to:
a) Degrees of structuring
- unstructured questions (e.g. in a preliminary stage of research)
- semi-structured questions
- fully structured questions (e.g. questionnaire)
b) Ways of administering
- face to face
- telephone interview
Questions (^)
The carrying out of an interview requires the formulation of:
- precise questions (to avoid vague answers)
- understandable questions (to avoid misunderstandings and so unreliable answers)
- non emotionally framed questions (to avoid emotive or stereotyped answers likely to be devoid of factual basis)
Interview : requirements (^)
The way of conducting the interview should be based on the following requirements:
- non authoritarian (presenting and not imposing)
- non patronizing (explaining and not persuading or giving advice)
- non interfering (facilitating and not correcting a reply other than trying to clarify the reply given)
Interview : pitfalls (^)
Possible pitfalls in an interview can derive from:
| Respondents | false data supplied by the respondents (Possible solution : cross-check the answers or check the data with other sources) |
|
|
|
| Questions | distortions caused by ill-formulated questions (Possible solution : make pilot tests) |
|
|
|
| Interviewers | errors caused by the interviewers in posing questions or in interpreting and reporting what the respondent has said (Possible solution : improve the training of interviewers) |
Observation : summing up (^)
Observation is the gathering of information through:
- Witnessing
- Measuring
- Questioning
Observation leads, in many cases, to a further, more controlled stage of information gathering, represented by Experimentation.