sampling design, measurement & scaling technique Flashcards
Steps in sampling design (7)
1) know type of universe
2) choose sampling unit
3) make a source list
4) sample size
5) parameter of interest
6) budgetary constraint
7) sampling procedure
1) know type of universe
- clearly define the set of objects, Universe, to be studied
- finite / infinite
- finite universe, number of items is certain 肯定
eg: population of city - infinite universe, is infinite (cannot have any idea about the total number of item)
eg: numbers of stars in the sky
2) choose sampling unit
-geographical unit
eg: state
-construction unit
eg: house
-social unit
eg: school
3) make a source list (sampling frame)
- sample to be drawn
- contain the names of all items of a universe (finite universe only)
4) sample size
- number of items to be selected from the universe
- not excessively large, not too small
(optimum - fulfills the requirements of efficiency, representativeness, reliability & flexibility)
5) parameter of interest
- consider specific population parameters which are interest
- may be interest in estimate proportion of person with some characteristics in the population / average or other measure of the population
-maybe important sub-groups in the population
6) budgetary constraint
- from practical point of view, major impact upon the decision relate size & type of sample
- this fact may lead to use of non-probability sample
7) sampling procedure
- finally, decide the type of sample, technique to be used in select for the sample
- must select the design which, for given sample size, cost, has a smaller sampling error
criteria of select a sampling procedure
1) cost of collect data
2) cost of two incorrect inference result from data
Two causes of incorrect inference
a) systematic bias
- error in sampling procedure
- cannot reduce / eliminate by increase the sample size
[ these errors can be detected & corrected ]
factors result of systematic bias (5)
1) inappropriate sampling frame
2) defective 缺陷 measuring
3) non-respondents
4) indeterminacy principle
5) natural bias in the reporting of data
1) inappropriate sampling frame
eg: a biased representation of the universe
2) defective measuring
- if measuring device is constantly in error
- if physical measuring device is defective, has systematic bias in data collected
( calibration of device / equipment to minimized the systematic bias) - In survey work, result if the questionnaire / interviewer is biased
3) non-respondents
- unable to get the proper / enough response
occur if
- refuse to answer due to embarrass info
- forgot return survey form
- survey didn’t reach all member in the sample
- invalidate无效 result
- result in higher variance for estimate, as sample size is smaller than expect
4) indeterminacy principle
- individual act differently when observe
5) natural bias in the reporting of data
eg:
1) understate the income if ask for the tax purpose, but overstate if ask for social status
2) in psychological survey, people give ‘correct’ answer rather than their true feeling
characteristics of a good sample design (5)
1) sample design must result in a truly representative sample
2) must result in a small sampling error
3) must viable in context of funds 资金范围内 available for the research study
4) systematic bias can be control in a better way
5) result of sample study can be applied with a reasonable level of confidence
different types of sample designs
1) probability sampling
- simple random
-complex random
eg: systematic sampling, stratified sampling
2) non-probability sampling
- convenience sampling
- purposive sampling
eg: quota sampling, judgment sampling
measurement scales (4)
1) nominal scale
2) ordinal scale
3) interval scale
4) ratio scale
1) nominal scale (qualitative)
- label / naming the variable
eg: gender - male, female - mutually exclusive (no overlap)
2) ordinal scale (qualitative)
- not really know the differences between each one
eg: satisfaction, happiness
1,2,3,4,5
3) interval scale (quantitative)
- know the exact difference between the values
- do not has a ‘true zero’
eg: 0 celsius ( no such thing as no temperature)
4) ratio scale (quantitative)
- have an absolute zero
eg: height, weight, duration
sources of error in measurement (4)
1) respondent
2) situation
3) measurer
4) instrument
Techniques of develop measurement tools (4)
1) concept development
- researcher should understand the major concepts of the study
2) specification of concept dimensions
3) selection of indicators
4) formation of index
scale classification bases (5)
1) subject orientation
2) response form
3) degree of subjectivity
4) scale properties
5) number of dimensions
important scaling technique
1) rating 评分 scale
-graphic & itemized rating scale
2) ranking 排行 scale
-method of paired comparison
