Skip to main content

Table 3 Plausible-values adjustments for log wage regressions

From: The use of cognitive ability measures as explanatory variables in regression analysis

 

Men

Never married women

Model

MESE

PVs

Unadj.

MESE

PVs

Unadj.

Skill control

Literacy

All

Median

Literacy

All

Median

 

skill

PVs

PV

skill

PVs

PV

 

(a)

(b)

(c)

(a)

(b)

(c)

Lit. Skills ( β ̂ 1 ):

      

 MESE

0.191

  

0.185

  
 

(0.010)

  

(0.025)

  

 All PVs

 

0.221

  

0.220

 
  

(0.015)

  

(0.033)

 

 Median PV

  

0.276

  

0.276

   

(0.012)

  

(0.031)

Effect of a one SD

0.218

0.221

0.251

0.210

0.220

0.259

Change in Skills

      

Race ( β ̂ 2 )

−0.094

−0.121

−0.065

−0.012

−0.031

0.022

 

(0.033)

(0.041)

(0.033)

(0.061)

(0.062)

(0.059)

DIC

103839

5492

5462

20905

1127

1114

N

3267

3267

3267

640

640

640

  1. Notes: Data are from the 1992 NALS, restricted to individuals aged 25–55 who work fulltime, reported wages, and who answered at least one literacy item. MESE model estimates (column a) are from Table 2. “All PV’s” estimates (column b) employ the recommended procedure (Mislevy et al. [1992]) for combining regression results for multiple imputations. “Unadjusted Median PV” estimates (column c) employ the median PV in the wage equation (1), with no adjustment for measurement error. All regressions also control for potential experience entered as a quartic, census region (entered as dummy variables), and urban setting (entered as a dummy variable).