This is a type of experimental validity that pertains to the interpretation or basis of the effect that was demonstrated in an experiment. (i.e., Given that the intervention was responsible for change, what specific aspects of the intervention or arrangement was the causal agent -- what is the conceptual basis (construct) underlying the effect?) 

Before Construct Validity can be evaluated, threats to Internal and External Validity must be ruled out.

Threats to Construct Validity

• Inadequate Preoperational Explication of Constructs: Constructs should be precisely explicated through detailed operational definitions.  The reader should have an unambiguous understanding of the construct.

• Mono-Operation (or Narrow Operation) Bias: Construct is operationalized too narrowly.

Example A: Two treatment groups are compared.  Say we use one set of therapists who are experts in Treatment A to administer Treatment A, and another set who are experts in Treatment B to administer Treatment B.  Suppose Treatment A appears more effective upon outcome.  Because therapists were different for the two treatments, we cannot separate the impact of therapists from the treatment.  Treatment is confounded with therapists.

Example B: Two treatment groups are compared. Say we use the same therapist for both treatments to avoid the above noted confound.  However,  now the conclusions are confounded by the therapist by treatment interaction -- could the therapist have been more competent in one treatment vs the other?

A better approach is to sample across a wider range of stimuli (i.e., Two or more therapists could be included, each of whom would administer both treatments -- then the interaction can be statistically examined.)

• Mono-Method Bias: Methodology for collecting data may be too narrow (and not generalizable).  One cannot generalize beyond the method used.  Example: measuring memory through only one modality.

• Hypothesis-guessing Within Experimental Condition: An example is the Hawthorne effect.

• Evaluation Apprehension:When subjects attempt to present themselves in a favorable light. 

• Experimenter Expectancies:When expectancies, beliefs, and desires about the results on the part of the experimenter unintentionally influence how the subjects perform. Replication by other investigators is necessary.

• Confounding Constructs and Levels of Constructs: Really, another example of too narrow stimulus sampling.  Example: Repeated measures designs (over time).

• Interaction of Different Treatments

• Interaction of Testing and Treatment

• Restricted Generalizability Across Constructs: Solution: Use multivariate designs.

• Attention and Contact With the Subjects: An example is the Placebo Effect.  In treatment outcome studies -- comparing treatment to no treatment -- we must question whether there are plausible features associated with the intervention that ought to be ruled out.  Generally, there is a threat to construct validity when attention, contact with subjects, and their expectations might plausibly account for the findings and have not been controlled or evalulated in the design.