Can we use CFA (confirmatory factor analysis) to ascertain validity coefficient? Validity coefficient is an important measure for test development especially in psychology and sociology wherein we have too indirect measurements. Perhaps we have different ways to know its size.
 A: Yes, CFA is useful for validity coefficient estimation. Here's a quote from Evaluation of convergent and discriminant validity with multitrait-multimethod correlations (Raykov, 2011):

In order to point and interval estimate these three types of validity coefficients, a
  special confirmatory factor analysis (CFA) model can be used in a first step (Kühnel,
  1988). This model differs from popular CFA approaches to examining convergent and
  discriminant validity, which have been available for more than two decades (e.g.,
  Bagozzi & Yi, 1993; Marsh & Hocevar, 1983; Marsh, 1989; Eid, 2000), in one main
  feature. This is the fact that in the cited procedures use is made of ‘proper’ latent
  variables that are evaluated by multiple indicators. The latter are the collected multiple
  measures of several traits using several methods of evaluation, whereas the latent
  variables represent the traits themselves and somewhat more difficult to interpret latent
  constructs. These constructs have been included in part to explain residual variation
  among measures of the same trait with different modes of assessment, and have been
  frequently referred to as ‘method factors’ (e.g., Marsh & Hocevar, 1983).

There's plenty more relevant info in this article after this section, so I won't quote it here, but do check it out!
Reference
Raykov, T. (2011). Evaluation of convergent and discriminant validity with multitrait–multimethod correlations. British Journal of Mathematical and Statistical Psychology, 64(1), 38–52.
