Objectivity

"Objectivity is the regulative ideal that guides all inquiry [which is] largely a measure directed at how researchers undertake and carry out their research in that it requires them to be precise, unbiased, open, honest, receptive to criticism, and so on" (Smith 1990, p 171, also Phillips 1990, Schwandt 1990). In a similar vein, Lather (1990, p 319) states that "objectivity means being aware and honest about how one's own beliefs, values, and biases affect the research process."

Objectivity in the research process

At the first step objectivity will depend on characteristics of the subject of study. The quantitative recording technique - counting - will allow the researcher to be objective if the subject is discrete and unambiguous (Figure 1). For example, counting people is more straightforward than counting "adults." The possible ambiguity involved in defining the latter introduces a subjective twist.

"Counting" continuous subjects (masses of people or water flow) will also require a subjective judgment by the researcher which can be ameliorated by clearly describing the unit of measure. This explains the emphasis placed on accurate documentation of methods and also illustrates the inescapable and non-symmetric linkage between quantitative and qualitative data. Establishing the quantity to be recorded requires a qualitative description. The corresponding trade-off between subjectivity and objectivity and the two characteristics is illustrated by the shading in Figure 2. Though connected, this representational objectivity should not be confused with objectivity at other levels of research.

Analysis, Interpretation, and Objectivity

For the purpose of this discussion, I make a distinction between the next two steps, though they are intricately connected. Analysis involves the search for pattern or significance in data within the context of the research situation. Interpretation involves explaining this pattern or significance within a wider context by applying relevant theory. Analysis questions what the data "says," interpretation, what it "means." In both cases, the potential for objectivity depends on the subject-researcher interaction. Although similar factors are involved at the two levels, the difference is critical. As illustrated in Figures 2 and 3, the potential for objectivity is reversed.

[I do not claim these diagrams to be either comprehensive or representative. Other factors such as personal skill and awareness of the researcher, are also involved (e.g. Glassner and Moreno 1989).]

Researchers can reduce personal bias at the level of analysis by administering strict control, as in a typical experiment. Responses can be attributed to the subject. This objectivity can be increased or decreased according to the latitude of response allowed. A restrictive response (e.g. yes/no answer) limits the researcher's opportunity to introduce bias as the data is analyzed. Figure 2a illustrates the relative positions of a few typical social science methods and the corresponding potential for objectivity.

Those factors which provide some objective potential at the analytical step, however, introduce bias at the interpretive step. Control and restrictions imposed on the subject through use of a particular experimental design involve presuppositions on the part of the researcher (Morgan 1983, Popkewitz 1990). In contrast, working closely with the subject allows continual adjustment of the theoretical constructs to ensure an appropriate match with reality (LeCompt and Goetz in Thompson 1989). Morgan notes further difficulties:

I found that I could use my data to support and refute in some degree most of the rival theories and hypotheses that I cared to generate... It occurred to me that the coefficients I was examining were both affirmational and negational. While supporting a hypothesis they simultaneously negated it by identifying the basis for a counterexplanation (sic) in terms of what was unexplained - the affirmational 0.45 identified a problematic 0.55!... The conclusions to be drawn from a particular piece of data depended very much on the frameworks through which it was being interpreted... As a researcher I had the power to realize these different meanings in the way that I presented my results. (Morgan 1983, p 11, 12, 15)

Figure 3 illustrates that objectivity of the lower levels and data manipulation (e.g. statistical analysis) also plays a role in the degree of objectivity possible. For example, once removed from the subject and research situation, it is difficult to determine if statistical "error" actually represents a real behavioural difference (Thompson 1989).

At the representational step there is a fairly straightforward link between degree of objectivity and use of the quantitative recording technique. This direct correlation decreases at the analytical step where the degree of data manipulation is more important. Although quantitative representation may increase objectivity (since the data can be manipulated more easily), the difficulties inherent in mathematical language noted above come into play. It is at this step that the application of craft skills emphasized by Funtowicz and Ravetz becomes crucial. At the next step, interpretation, the direct correlation between objectivity and quantitative representation disappears. The biases introduced by integrating results and theory will occur irrespective of the representation style used to express the results.

Revisiting Objectivity

Earlier, when considering the potential for objectivity, I noted confusion among different levels of the research process. I quoted Smith (1990) who states that objectivity "requires [researchers] to be precise, unbiased, open, honest, receptive to criticism, and so on." I believe that the importance of these requirements should be weighted according to level: precision is more important at the recording level (e.g. precision of definition or measure), openness and honesty at the methodological level (e.g. recognition of theoretical bias). To recognize one's personal bias - and preferably to remove it - is a critical concern at all levels. The potential to do so varies. Although removal of researcher bias may be possible at the level of recording technique, it will be impossible at a methodological level. To interpret the former as more important or more relevant than the latter, or as the definitive issue related to objectivity, is to ignore historical change in knowledge. This relates to the danger of unquestioned paradigms noted above and leads to consideration of one more level of objectivity.

Objectivity and the Scientific Community

Popper states: "scientific objectivity is based solely upon a critical tradition which, despite resistance, often makes it possible to criticize a dominant dogma... [It is] not a matter of the individual scientists but rather the social result of their mutual criticism" (quoted in Phillips 1990, p 44). This refers to the potential for the scientific community itself to be objective - to recognize their paradigmatic bias, to be open and honest regarding the fit of their theories with their ontological, epistemological and methodological positions.

Considering objectivity at the 'lower' and 'higher' levels, I believe there is an important nuance recognizable in use of the term that parallels use in natural and social sciences, respectively. In the former cases, where the focus is on representational and analytical objectivity, the term refers to detachment. In the latter cases, where the focus is on interpretive and paradigmatic objectivity, it refers to being critical. I believe that scientists, especially "hard" scientists, relying on the scientific method as a method legitimately applaud its ability to improve objectivity at the representational and (to some extent) the analytical steps. The scientific 'method,' however, is also viewed as an epistemological (Harding 1987), or even paradigmatic, position. It represents a particular way of knowing. At this level it has a more limited potential to be objective. Considered from a historical perspective, reason and the scientific method provided a relatively objective contrast to the dogma of regal and religious authority prevalent prior to the scientific revolution (Smith 1983, Popkewicz 1990, Morowitz 1996). It must now be recognized, however, that its adoption was a replacement of the dominant dogma - a change in bias, not a reduction or removal of bias. To quote Saul (1992, p 4): "with time and power [reason] has become a dogma, devoid of direction and disguised as disinterested inquiry. Like most religions, reason presents itself as the solution to the problems it has created." Perhaps it is now time for another change - a shaking up of the dominant dogma to generate new approaches to understanding.

The post-modern and post normal critiques arise, in part, due to a concern that science has lost the ability to be objective at a paradigmatic level (Figure 4).