Abstract: A Review of Campbell's Causal Inference Model Using Research on Motivational Interviewing (Society for Social Work and Research 22nd Annual Conference - Achieving Equal Opportunity, Equity, and Justice)

A Review of Campbell's Causal Inference Model Using Research on Motivational Interviewing

Schedule:
Friday, January 12, 2018: 5:45 PM
Marquis BR Salon 7 (ML 2) (Marriott Marquis Washington DC)
* noted as presenting author
Alexis Kuerbis, PhD, Associate Professor, Hunter College, New York, NY
Significance. Donald T. Campbell was a highly influential psychologist and social scientist, for whom the Campbell Collaboration is named. Within the social sciences, Campbell is known for a broad range of important work in research methodology, yet his work on causal inference is most prominent and now the dominant paradigm for quantitative and experimental social scientists.

 

Content. In this presentation, the primary components of the Campbell Causal Model (CCM) will be reviewed as a primer and base for comparison for other models of causal inference. Specifically, this presentation will describe the four types of validity within the CCM that provide scientists the ability to infer causation: internal validity, construct validity, statistical conclusion validity and external validity.

To illustrate and critique the CCM, both seminal and recent research on one of the most promulgated evidenced based practices in social work, Motivational Interviewing (MI), will be used. Meta-analytic reviews of experimental studies of MI and data from recent randomized controlled trials examining the mechanisms of behavior change within MI will be used to provide examples of the four types of validity that shape how we infer causation.

Internal validity is the extent to which causation can be inferred based on differences across conditions (e.g., experimental manipulation). For example, when studies state they used a "motivational intervention" in the context of SBIRT[RR1] , to what extent does that reflect MI or an adapted intervention? What are the methods for assessing treatment fidelity and discriminability across conditions--signaling successful experimental manipulation?

Construct validity is the extent to which the theoretical framework has been operationalized successfully. In the case of MI, have we captured outcome variables accurately? Are its hypothesized mechanisms of action, such as reducing ambivalence, captured in a way that is true to the theory and reliable?

Statistical conclusion validity refers to the ability to make inferences about the relationship between the experimental manipulation, e.g., treatment, and the outcome of interest. Are there other variables for which we need to control in order to increase the level of validity? In the case of substance abuse treatment trials, for example, one must always account for baseline levels of drug use to isolate the change in the outcome variable to those that occurred during, and therefore assumed to result from, the treatment.

External validity describes to whom results of a study of causal inference can be generalized. It refers to how representative a sample is of the population of interest. In the context of MI, who has been included in study samples and who have been excluded? According to Campbell, what can be reasonably inferred about generalizability by these inclusions and exclusions?  Finally, strengths and limitations of Campbell's model will be reviewed in the context of social work research.

 

Implications. The CCM remains the standard in social work research, owing to its richness and precision. While Campbell's model can be precise and inspire confidence for certain types of conclusions, adhering only to this model prevents study of important social problems and injustices.