To get the most out of medical surveys, one must understand the demographics of the respondents, beyond straightforward inclusion in the target lists of doctors or facilities. Designing for, gathering, and using the modestly more detailed information will allow better understanding of survey answers, collection of sufficient replies by category, and inclusion of the right participants.
Starting with the correct overall pool of respondents is the goal of the first demographic questions. In target lists procured through medical societies or third party vendors, one will most often get not just US-based physicians, but residents or interns, as well as doctors outside the country. These would be excluded from most surveys. Additionally, even if the clinical specialty is correct, some participants may not be familiar enough with the topic to render valid opinions, so should also be screened out.
Refining as you go
When starting the first wave of a survey, it is often good to ask for more detail, rather than less. However, keep in mind that the demographic questions add to the actual and perceived length of the survey. So if analysis of those early surveys show that either the heterogeneity of the participants is low, or that no substantive difference exist between groups in a category, then the demographic questions should be simplified or cut back. This will allow for more topic-oriented questions to replace demographic questions.
In addition to the previously mentioned screening questions, the demographic questions can be used to more specifically screen, as well as can be used to send various participants to alternative branches of questions.
There are a handful of key demographic categories for which to gather information. Some are obvious and often used, such as location, size, and (sub)specialty. However there are others to consider, for example geographic environment, practice setting, and a participant’s role.
Location, environment, and practice setting
Much more can be useful beyond geographic location of city or state (should be used instead of region to avoid confusion–can be aggregated in analysis). The geographic environment is also important, urban, suburban, and rural. Moreover, facility type should be gathered, e.g., for hospitals: academic, government such as VA (Veterans Administration), or community. One should also ask subjective size, as many respondents may not know well the # of beds for example. For physician practices, knowing whether it’s hospital-based, private, staff-model HMO (e.g., Kaiser), or academia, is worthwhile. Also for physicians, is it a small, single-specialty practice with only a few doctors, or a large, multi-specialty practice?
Knowing even more how the physician practices can be gleaned from demographics. Does the MD have a subspecialty. How does the doctor spend time proportionally among routine patient care, teaching in a clinical setting, teaching in a class room, conducting clinical studies, or other research. Without specifically asking these questions, it is easy to misinterpret responses as typically from routine patient care. Another good check is asking for a title, if the participant may have one.
Size is a prototypical demographic question. Usually, things like bed size for a hospital, or number of patients for a physician, get asked. However, in many cases, respondents won’t really know, so it is often better to start in the first wave with not only a quantitative question, but a qualitative question as well, e.g., as mentioned above — small, medium or large hospital. As surveys are completed, one can see if participants have a good handle on the quantitative values, and if they match with the qualitative assessment. At this point, it is likely best to pick one or the other — the most consistent and valid, as it will lead to more comparable results, as well as free up space in the survey (mentioned above), to add more topic-oriented questions.
For physicians, in addition to number of patients individually, it is often useful to ask how many doctors are in the practice (by specialty if relevant), and total number of patients in the practice. When asking for numbers of patients, it is important to be a clear as possible whether the total number by practice, or by individual physician, is asked for. As in the case of hospitals, it is good after the first wave of surveys to look for consistency, validity, and whether one of the aspects of the questions may be dropped.
To gain the best insight into a survey, properly detailed and worded demographic questions are necessary. These should be checked for consistency, validity and usefulness in the first wave of surveys. Finally, demographic questions should be revised and pruned for later waves of surveys, in order to make room for those questions that are topic-oriented.
© 2013 Winton Gibbons