control variables and model 2 includes the Relation variable. Only difference between two model

is existence of relation variable. Result for **lrtest** which compare two model shows Relation

variable has effect on Attitude variable and it is statistically significant (Prob > Chi2= 0.0031).

**Table 3: Ordinary Logistic Regression Result **

Model 1

Model 2

Coefficient

Std. Error

Coefficient

Std. Error

**Relation **

** **

** **

**-0.401*** (0.12) **

Age 0.0155**

(0.0061)

0.0117*

(0.0062)

Size 0.220**

(0.11)

0.206*

(0.11)

Wage level

-0.106

(0.13)

-0.0672

(0.13)

Manufacture

Industry 0.115 (0.51)

0.174

(0.51)

Construction Industry

0.404 (0.76) 0.638

(0.78)

Service

Industry

(Reference: The others)

-0.181 (0.5) -0.154

(0.51)

Irregular Member

Yes

-1.159*** (0.41) -1.186***

(0.41)

Irregular Member

No

(Reference: no irregular

workers in workplace)

-0.518** (0.24) -0.588**

(0.24)

Observations 318

318

R-squared 0.0276

0.0394

*Note*: Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1

This table shows coefficient value in logistic regression method, the coefficient given by all

logistic method describes independent variable’s effect on the log odds that attitude for irregular

workers. Interpretation is that for a unit increase in relation variable (more bad relation with

employer), the probability to be friendly to irregular workers increases 1.5(e

β=-0.401

)

times, holding

other variables constant. It means if the more union thinks *‘employer supports union and treats *

8