Multivariate statistical analysis Introductions and basic data analysis

Preview:

Citation preview

Multivariate statistical analysis

Introductions and basic data analysis

Multivariate Variate ( 變量 ) vs. variable ( 變數 )

The attributes that the researcher concerned and observed performance

The attributes that the researcher could operate for the expected performance

Uni-variate ( 單變量 ) vs. multi-variate ( 多變量 ) Single concerned performance Multiple concerned performance vector

Measurement scale

Nominal Ordinal Interval Ratio ref. p.10 表 1.2-1 四種衡量尺度之比較

Four types of measuring scale

Measuring Variables Measuring variables: used to describe

the attitudes of specific concerned attributes

Analytical variables: internal scale, ratio scale

Categorical variables: nominal scale, ordinal scale

ref. p.11, 表 1.2-2,-3,-4

Example

Cost of measurement

Error cost: the impact resulted from the deviation to the true attitude

Measuring cost: the difficulty of accurate measuring

Reliability

Retest reliability Verify the stability of the responses

Split half reliability Designing the contrast questions

Cronbach’s α (>0.7)

Cronbach’s α

Validity

Effectiveness to reflect the concerned issues

Content validity Criteria-related validity Construct validity

Problems of validity

Likert scale Quasi-interval scale 5-scale, 7-scale, (in the form of 2/3

negative scale and 2/3 positive scale around the original)

Data format

Cases: the observant, the experimental subjects/objects

Variables: the set of concerned attributes

Observations: the collected data Observation vector: the set of all

attributes retained from a specific case

Data format

Classification of multivariate models Functional relation model

Responsive variates=f (independent variables) Interdependence relation model

Variables interdependence Cases interdependence

Systemic relation model Path analysis LISREL model

ref. p.33, 表 1.7-1 多變量統計模式之歸類 ; p.40, 表 1.7-2; p.41, 表 1.7-3

Multivariate analysis models

Multivariate analysis models

Multivariate analysis models

SAS/SPSS introductions