There is a long tradition of using photographs in population research.1 That tradition, however, appears to have been almost completely lost, perhaps because population specialists have avoided photographs, as they have avoided narratives, to appear more scientific.2
Photographs were more widely used research tools in the early decades of the last century than they are today—even though the costs of taking and printing photographs have declined and the quality of cameras and reproduction techniques have improved. This is surprising, given the increased use of qualitative data by family planning and reproductive health researchers. Examining the work of social scientists who used photography to study population-related issues in the 1930s gives a sense of what we have lost, as well as what we might gain from greater use of photographs as research tools.
In 1939, Margaret Hagood, who would become president of the Population Association of America in 1954, pioneered the use of photography in a study of rural North Carolina. Her book, Mothers of the South, recently reissued 57 years after its initial publication, includes photographs taken by Dorothea Lange and Marion Post for a study that Hagood and Harriet Herring carried out in 13 counties. What is striking about Hagood's effort was her eagerness to explore "some of the potentialities of photography as a tool for social research"3 at the same time that she was also completing a landmark textbook aimed at increasing the use of sound statistical methods.4 As a result, she is remembered not only for being "a pioneer in statistics and demography" but also as "an early advocate of combining qualitative and quantitative methods."5
Photographs have also been used in international population research. In 1930, the Thai government undertook the first nationwide household survey of economic, health and social conditions. Respondents were interviewed about a variety of topics, including agricultural practices, economic circumstances, health and fertility. The final report was illustrated with photographs of respondents and of their possessions and work routines;6 these illustrations help us understand life in rural Thailand in the 1920s and 1930s.
It is time to reclaim the lost art of using photographs to conduct research and to disseminate results. Photographs should be used as data in family planning and reproductive health research, and as illustrations in publications and on Web sites. Reviving the use of photographs would help researchers better understand and more easily explain numerous aspects of health and family planning services.
Photographs as Data
Population and family planning researchers regularly combine quantitative and qualitative data,* and photographs should be added to this mix. Doing so would help explain important social processes. Because it is difficult—perhaps impossible—to use photographs as an exclusive source of data, they should be part of a multiple-method research strategy. Images need to be interpreted in their social and historical context by analysts who are able to use other data or their personal knowledge to describe that context.
Photographic evidence could be used to improve our understanding in several areas, particularly in research on family planning and reproductive health facilities. Studies of the physical settings in which services are provided (including waiting lines, other circumstances of clients waiting to be served, the availability and condition of restrooms, the types of transportation available and changes in these aspects over time) could clearly be enhanced through the use of photographs. More challenging would be studying issues such as client-provider interaction. In this case, photographs might be employed to examine how body language reinforces social status and hierarchy.
A good place to begin collecting photographic data would be in conjunction with the family planning facility surveys now being undertaken around the world as part of the MEASURE program,† funded by the U.S. Agency for International Development. Photographs could be used to clarify findings related to respondents' attitudes and perceptions of a facility, since one respondent may say that a clinic (or a portion of it) was clean or dirty, while another may provide a different assessment of the same space. Photographs can also be used to provide data on "unobtrusive measures" of activity—that is, less-obvious indicators of quality of care, such as the amount of refuse in clinic waste receptacles.7 One or two illustrations could be incorporated in the published report of the facility survey, and additional images could be made available on the Web site of the researchers' institutions.
Using photographs in this way would help reduce the ambiguity of what the data actually convey. The authors of a recent U.S. study on the effects of cleanliness, for example, concluded that "some 25 years after its assessment, the cleanliness of a respondent's home is predictive of higher respondent earning;"8 they also suggested that growing up in a clean home may produce long-term positive health effects. Their measure of cleanliness was based on the interviewers' evaluation of how clean the dwelling unit was. Photographs of different levels of cleanliness would have improved the training of the interviewers, and photographs taken at the time of data collection would have increased the reliability of the findings. The same is true for research in clinic settings, where the effects of cleanliness and of other organizational characteristics are probably at least as great as they are in private homes.
Photo Collection and Analysis
There are useful suggestions for collecting photographic data. Jon Wagner, for example, suggests using "shooting scripts," which outline standardized routines for the photographer-researcher.9Such scripts can be used to set both physical and temporal limits on data collection (i.e., for the former—"photograph the counseling cubicle" or even "photograph the cubicle from a particular spot;" and for the latter—"photograph morning rounds for the third, ninth, and 17th day of the month"). It is also possible to collect a random sample of images by dividing the subject into numbered parts and selecting which parts to photograph through a random selection process. For example, different areas of a facility can be numbered (entrance-1, reception area-2, waiting room-3, kitchen-4, storage area-5, director's office-6, etc.), and these areas can then be randomly selected for photographing in conjunction with a facility survey. The same approach can be used to randomize the timing of photographs.
For researchers interested in using photographs but not in taking their own pictures, several sources of photographs can be explored. For research on service delivery outlets, useful images may be found in staff photograph albums, institutional reports (especially annual reports), public information office files, the files of technical assistance groups or donors that have worked with particular clinics or service centers, and the reports and files of any supervising ministries or other oversight bodies. In some cases, local newspapers or news agency files may also have useful photographs.
George W. Dowdall and Janet Golden recommend a "layered analysis" strategy for the interpretation of photographic images.10 Such an analysis begins with an appraisal that views the images in their social and historical context, and then compares the visual data with administrative data and with other written sources and interview data. The issues here are mutual verification and correspondence among different sources of data.
A second level of analysis, termed inquiry, concentrates on collecting images from a particular locale, and examines the prevalence of certain themes and patterns. For example, photographs from a rural service delivery site might show marked differences in "gendered space" and in how the clinic allocates facilities to men's and women's activities.11 Photographic evidence likewise could be used to document the levels and effects of privacy in health clinics.
The third stage of the layered analysis is what Dowdall and Golden call interpretation, which is similar to Clifford Geertz's "thick description,"12 in which the researcher focuses attention on individual images and makes an effort to explain how the people involved perceive the context and meaning of the scenes that are depicted.13
How difficult would it be for researchers to take and use photographs in their work? It will take time and experience before family planning researchers learn how to use photography to maximum advantage. But a great deal can be gained at a low cost and with little time and effort. The first step for many researchers is to become more systematic about the snapshots they routinely take when conducting field studies. One strategy would be to focus on a particular issue or a given service over time at several locales.
Potential Problems
Photographs as data have the same problems of sampling and measurement error that bedevil other types of data. Moreover, because the use of photographs as data is still rare, methods for handling these problems have not been well developed. Because of the visual nature of photography, some problems will be worse than those resulting from more anonymous sources of data.
Designing confidentiality and informed consent procedures that take into account photography's loss of privacy will be especially troublesome. For those working internationally, these problems could be mitigated somewhat by the fact that photographs published outside of one's country might not be considered as much of a violation of privacy as would photos published locally. On the other hand, putting images on the Internet may create additional problems that can be difficult to explain to many developing-country subjects who are unfamiliar with the medium's reach.
Ensuring confidentiality or biographical anonymity may be particularly difficult when using photographs from personal files or institutional archives, even when permission can be obtained from the people and institutions that hold these photographs. (Some U.S. institutions, especially mental health institutions that deal with stigmatized behavior, grant permission to publish only "those images depicting patients who could safely be assumed to have died."14)
The experience of photojournalists provides a helpful guide for some of the issues related to bias, privacy, and the staging or composing of photographs.15 But because photojournalists enjoy First Amendment protections, their work does not offer much help in the area of obtaining informed consent.
The question of sampling will also be difficult, because the circumstances of both people and places vary so much throughout the day. At one time during the day, a facility might have long lines, while a few hours later, the same space might be empty. Photographs obtained from institutional archives will have the same problems as other archival data—that is, documentation about the source, purpose, subject and intended audience of the photographs typically will be lacking, which makes interpretation difficult.
Using photographs as a research tool is not easy. They portray reality as it is perceived and arranged by the photographer (and sometimes by the subjects), just as a regression equation portrays reality as it is perceived by the modeler. Some of the uses I describe are straightforward, but in many circumstances it will be difficult to stop people from creating a new reality for a photograph.
The quandaries posed by using photographs is illustrated by the work of the photographers, hired during the Great Depression by the U.S. Farm Security Administration, who worked with Margaret Hagood during the 1930s. Consider the famous and beautiful Dorothea Lange photograph of a black-haired woman, known as "Migrant Mother," who is obviously at the end of her rope, staring down, indifferent to the baby in her lap and to the two children leaning against her (Figure 1, below).‡ Lange took several photographs of this woman, apparently focusing less and less on her children (of whom there were seven), perhaps because she wanted to avoid the implication that the woman had brought poverty on herself by hapless high fertility. Lange might also have focused on the mother to convey the impression that the woman could manage with only a bit of help because she had only three children.16
Figure 1. "Migrant Mother," by Dorothea Lange, 1936, USA. Library of Congress, Prints & Photographs Division, FSA-OWI Collection [reproduction number, LC-USZ62-95653]. 
"Migrant Mother" is the best known of a series of six photographs that, taken together, provide a fair picture of the woman's circumstances. Because all of the photographs in the series are available on the Internet, we can dispute the selection of one image to characterize the woman's circumstances and wonder about the motives of those who selected it.
Many scientific journals, especially those published in developing countries, probably cannot afford the costs of printing photographs of adequate quality. The Internet provides an opportunity to transcend that limitation, however. Just as authors are increasingly likely to post data sets or a full table of coefficients on the Internet, so too could they be encouraged to post a photo-database supporting their conclusions.
Photos as Illustrations
Photographs should also be used more frequently as illustrations in scientific publications. Much of what appears in our journals would be easier to understand and would be more effectively communicated, especially to those without firsthand experience of a particular situation, if articles were illustrated. Given the wide diversity of local and national circumstances covered in the reproductive health and family planning literature, photographic illustrations would help readers understand what is being described.
Several examples of the potential yield of the greater use of photographs as illustrations come to mind.§First, given the importance of quality of care in studies of reproductive health and family planning services, it would be useful for many readers if different levels of quality were illustrated. For example, different waiting conditions, levels of cleanliness or types of operating theaters could be shown. Carefully selected photographs would help readers understand the practice of female genital cutting.** Another example could be before-and-after photographs of "simulated" or mystery clients, so readers understand the extent of the camouflage involved in research efforts that ask people to pose as fictional clients.17
There are numerous contemporary examples of photographs serving as illustrations and as data. Many readers will doubtless think of a specific series of pictures or the work of a particular photographer as examples of my point. Peter Menzel's Material World: A Global Family Portrait,18 which was funded in part by the United Nations Population Fund, persuasively demonstrates the ability of photographs to illuminate what other data cannot. J. Mayone Stycos, in Children of the Barriada19 and Cornell Capa and J. Mayone Stycos, in Margin of Life20 also tried to use photography to illustrate population issues. In addition, some of John Knodel's work provides good examples of photographs as illustrations.21
Conclusions
I am not advocating the use of photographs to help market or promote research or specific policies, or to evoke sympathy or aversion for a particular type of service or population. Such uses of photography are already widely understood and practiced, and indeed may have contributed to the debasement of photography as a research tool. For example, in an assessment of the role of visual images in reproductive politics, Rosalind Petchesky noted that in the abortion debate, "a picture of a dead fetus is worth a thousand words."22 Because photographs are currently used to advance arguments but not to investigate issues, visual images are not taken as seriously as they should be.
Given the availability of a range of new technologies, especially the moderately priced digital cameras and the Internet, it is past time for researchers to collect photographic evidence and for journals such as Family Planning Perspectives to publish their findings. Photographs should not be compared to some hypothetical absolute truth, but only to other data collection methods. In that company, they look pretty good.