Skip to main content
. Author manuscript; available in PMC: 2011 Dec 17.
Published in final edited form as: J Hum Resour. 2011 Fall;46(4):875–906. doi: 10.1353/jhr.2011.0005

Table 4.

Effects on Latent Cognition and Response Scales Estimated from Combined Vignettes and Proxy Indicator Model (Model 3)

Latent cognition as function of test scores
Latent cognition of vignettes
Orientation 0.371 (0.262)
Immediate memory 0.515** (0.259) Vignette 1 0.616** (0.253)
Short-term memory 1.028*** (0.223)
Prospective memory 0.067 (0.094)
Word-finding & verbal fluency 0.781** (0.327) Vignette 2 −0.615** (0.253)
Processing speed 0.252 (0.335)
Search accuracy 0.138 (0.275)
Numeracy 0.133 (0.184) Vignette 3 −1.351*** (0.255)
Literacy 0.067 (0.145)
Constant −1.717*** (0.607)
ση 1.132 (fixed) σξ 1.000 (fixed)
Response scales identified from test scores
Response scales from vignettes
Cut-point – moderate/severe/extreme
Age 55 to 64 −0.092 (0.223) 0.000 (0.087)
Age 65 to 74 0.208 (0.218) 0.072 (0.089)
Age 75+ 0.567** (0.222) 0.132 (0.092)
Female −0.127 (0.090) −0.056 (0.041)
White −0.189 (0.377) 0.663*** (0.178)
Log Wealth −0.084*** (0.025) −0.050*** (0.014)
No wealth −0.664* (0.351) −0.440** (0.192)
Qualifications 2 −0.147 (0.112) −0.156*** (0.052)
Qualifications 1 −0.134 (0.108) −0.061 (0.052)
Not working <65 0.398*** (0.149) −0.059 (0.062)
Cut-point – mild
Age 55 to 64 −0.105 (0.152) 0.153 (0.118)
Age 65 to 74 0.090 (0.155) 0.206* (0.120)
Age 75+ 0.422** (0.167) 0.361*** (0.128)
Female 0.017 (0.077) 0.006 (0.059)
White −0.223 (0.341) 1.057*** (0.191)
Log Wealth −0.013 (0.024) −0.021 (0.020)
No wealth 0.208 (0.343) −0.580** (0.276)
A-level or above 0.127 (0.094) 0.000 (0.074)
Qualification < A-level 0.081 (0.093) 0.051 (0.076)
Not working <65 0.320*** (0.107) 0.095 (0.088)
Constant 0.529 (0.527) 0.437 (0.334)
Log likelihood −5178.03
Number of observations 1782

Notes: Standard errors in parenthesis. *, ** and *** indicate significance at 10 percent, 5 percent and 1 percent respectively. Coefficients are identified up to the scale parameters ση and σξ. Estimates presented are with normalization σξ = 1 and ση equal to the estimate under the null hypothesis of RC1. Inference does not depend on the chosen normalization.