Thursday, December 20, 2007

The Ten Commandments of Services Research

Taking a cue from my friend Sarah Moffett, a lawyer who follows a similar credo...

I. I am the Lord thy God. My name is Validity — Internal and External. Worship both. Thou shall not have any gods before Me. Much of science tends to be preoccupied with internal validity, the strength of the claim that the experimental manipulation causes the outcome. This leads to the desire to control all potentially confounding variables but may reduce external validity. External validity, or the extent to which the study findings generalize to the world beyond the research project, is very important in services research.

II. Do not worship idols. Although you must have comparison conditions, they need not be exact images. Remember that you worship Internal Validity not Perfection. In medication trials we are used to seeing "placebo," which resembles the experimental condition to the fullest extent possible. In services research, although it is important to have control conditions, it is often not possible or desirable for the comparison condition to be a twin of the experimental condition.

III. Do not take the name of "effectiveness" in vain. Remember that you worship External Validity. Many researchers think that they are doing an effectiveness study if they measure outcomes beyond symptoms. That is not what effectiveness means. Effectiveness refers to the impact of a program in the real world, beyond the tightly controlled world of clinical trials.

IV. Remember the Sabbath day and keep it holy. Research assessments should not be scheduled on important religious holidays and on anyone's Sabbath. Services researchers should pay close attention to issues of culture.

V. Honor thy fathers and mothers — and grandparents, foster parents, and families of choice. Services researchers should pay close attention to issues of culture.

VI. Do not murder your data analysis section - or your biostatistician. The quasi-experimental and group cluster designs of services research require complex statistics. The statistician should be a part of the study from the very beginning.

VII. Be faithful to intervention design, and use measures of program fidelity at all times. Given the complexity of many services research studies, which often test psychosocial and organizational interventions, it is essential that strategies are used to ensure that the interventions being tested are true to their descriptions.

VIII. Although thou shalt not steal, thou shalt borrow frequently. To the extent that services research often involves working in unique cultural and system contexts and adapting standard approaches, it is tempting for investigators to assume that they have to reinvent the wheel. It is important for services researchers to borrow heavily from the work of others.

IX. Sins of omission can get you into as much trouble as lying. Don't stick with the psychiatry literature. Remember sociology, anthropology, psychology, economics, marketing, education, political science... Services research draws heavily from a broad range of disciplines for conceptual models and approaches.

X. Do not covet the grants of your psychopharmacology clinical trials friends.

Lisa B. Dixon, M.D., M.P.H.
Dr. Dixon is affiliated with the Veterans Affairs Capitol Health Care Network and with the Department of Psychiatry, University of Maryland School of Medicine, Baltimore.

Originally appeared in Psychiatric Services (December 2007)

1 comment: