Inductive and Deductive Approaches to
Research: Article Review
The objective of this paper is to discuss the inductive and deductive approaches to research. The paper will later narrow its focus to an article that has done a research on current affairs. In this article, the research method will be identified and described. Specifically the paper seeks to determine whether the approach used is the deductive or inductive approach, as well as provide an appraisal of the approach. The rest of the paper is organised as follows: section 2 provides an overview of the deductive and inductive approaches to research; section 3 provides a critique of the article “Gender Differences in Self-Evaluations of Information About Current Affairs” by Slevin and Aday (1993) and published in Sex Roles, vol. 29, Nos. 11/12; Section 4 provides some conclusions and recommendations.
- Deductive and Inductive Approaches to Research.
There are two main types of scientific approaches to tackle a research problem. They include the deductive and inductive approaches.
- The Deductive Approach.
Under the deductive method, a theory and hypothesis (or hypotheses) are developed and a research strategy is developed to test the hypothesis. (Saunders et al., 2003). It involves formulating a tentative theory of the phenomena, determining the consequences of the implied empirical evidence, as well as controlling situational events so as to observe the validity of the empirical deductions. (Bonoma, 1985: 199). The deductive method is often regarded as the scientific method. It is used to test theoretical knowledge by employing “logical principles of deduction”. The deductive begins by reviewing the theory, developing testable hypotheses and forecasts and then using empirical data to test the theory with the objective of supporting or refuting the theory. (Dictionary of Social Sciences, 2002). The deductive approach in effect works from the general to the specific. It is informally called the top-down approach because it begins by thinking of the general theory and then narrows the theory down to a more specific hypothesis that can be tested. The narrowing down process is even further when data is collected to test the hypothesis. Once the hypothesis is tested, the theory is either confirmed or discarded. The process is illustrated in figure 1 below:
Figure 1: Deductive Approach.
Source: http://www.socialresearchmethods.net/kb/dedind.php
As shown in figure 1 above, the deductive approach follows a top-down approach by starting with the theory and ending with a confirmation as to whether the theory is correct or not. The decision to refute or accept the theory depends solely on the underlying population that the sample is drawn from as well as on the sample size. It may also depend on different categories of the members of the population. For example, age, sex, race, religion, income level, occupation, etc. These factors may influence the results of the study significantly.
The inductive approach is often associated with quantitative methods, which focus on large samples, standardised measures, and highly structured interview instruments as a means of collecting data to test hypothesis. (Marlow, 1993). Categories that can easily be quantified are generated typically prior to the study and statistical techniques used to analyse the data collected. (McRoy, ND). These approaches are commonly used finance, economics, and other social sciences which directly lend themselves to quantitative methods.
Bonoma (1985) observes that a number of studies have made extensive use of the deductive approach. Since the deductive approach is exemplified by “logical syllogism” whereby a hypothesis is formulated and tested to draw conclusions, the validity of the findings depend solely on the quality of the logic employed because the conclusions cannot provide any information above that contained in the hypotheses. (Bonoma, 1985).
2.2 The Inductive Approach
The inductive approach on the other hand enables the researcher to collect data and develop theory depending on the results obtained from analysing the data. (Saunders et al., 2003). The inductive approach is less frequently used. It focuses on individual and naturally occurring observations towards the development of a theory. The observations are usually less controllable. (Bonoma, 1985). The is a bottom-up approach. It begins by observing data, determing specific partterns and irregularities in the observations, developing tentative hypothesis and then developing theory. (http://www.socialresearchmethods.net/kb/dedind.php). It is called a “bottom-up” approach because it begins with the specific to the general rather than from the general to the specific. (http://www.socialresearchmethods.net/kb/dedind.php).
Figure 2: The Inductive Approach.
Source: http://www.socialresearchmethods.net/kb/dedind.php
The inductive approach is more often associated with qualitative research, which is concerned with non-statistical non-quantitative methods of inquiry and analysis of social phenomena. (McRoy, ND). Themes and categories are drawn through analysis of data collected by such techniques as interviews, observations, videotapes, and case studies. The samples are usually small and are often “purposively” selected. To examine specific issues, as well as problems under study, qualitative research employs makes use of detailed descriptions from the perspective of the research participants themselves. Another commonly phrase associated with the inductive approach often used interchangeably with qualitative research is “field research”, which describes the systematic observations of social behaviour with no preconceived hypothesis to be tested. (Rubin and Babbie, 1993; McRoy, ND). Hypothesis are developed based on the interpretation of human behaviour. (McRoy, ND). Taylor (1977) describes qualitative research as “naturalistic research” or “inquiry” into everyday living. In this case, human behaviour is studied by making direct observations in the everyday life of human beings. (Taylor, 1977; McRoy, ND). Based on symbolic interaction theory postulated by Blumer (1969), it is a common believe among naturalistic researchers that knowledge gained through sources that have intimate familiarity with an issue is far better than the “objective distancing approach that is employed in quantitative methods. (Haworth, 1984). This approach was employed by Zurcher (1983) to study the behaviour of passengers in an airplane and fans in a football game. Another commonly used term associated with anthropology which is directly linked to qualitative research and thus the inductive approach is “ethnography”. (McRoy, ND). This term is used to describe a field study of a particular site or population undertaken with the objective of better understanding the culture from the perspective of that population. (McRoy, ND). Teams of researchers in ethnographic studies collect data from subjects by observing and interviewing the subjects over a period of time. (McRoy, ND).
The inductive approach and thus qualitative research is said to provide a lot of advantages. For example, qualitative methods can be employed in situations where problems and issues cannot be treated using quantitative examination. (Sherman & Reid, 1994). In addition, collecting data using descriptive, inductive, and “unobtrusive” techniques are considered compatible with the knowledge and values of the social work profession. (Epstein, 1988 cited in McRoy, ND). According to the “social-psychological bases of qualitative research” qualitative research is compatible with the “person-in-environment paradigm of social work practice”. (Epstein, 1988; Taylor, 1977). Moreover, qualitative approaches are widely employed in clinical social work assessments, in which case the clinicians tend to rely on the interviews and data collected on a client’s issues in the context of the environment. (Gilgun, 1994). In clinical studies, each case is treated individually and without imposing preconceived notions or attempting to generalise a particular problem to all clients. (McRoy, ND). Furthermore, Reid (1987) suggests that qualitative research methods may be the most appropriate research approach in studies of social processes of complex human systems such as families, organisations and communities. Finally, Babbie (1986) argues that qualitative research approaches have a high degree of flexibility, in-depth analysis, and potential to observe a variety of aspects of a social situation.
The designs, procedures and analysis under qualitative approaches are based on “grounded theory” because the theory developed through qualitative approaches is “discovered, developed, and provisionally verified through systematic data collection and analysis of data”. (Strauss and Corbin, 1990: 23). Two types of grounded theory have been identified: substantive grounded theory and formal grounded theory. (Straus and Corbin, 1967). In substantive grounded theory, the development of theory is based only on one area of inquiry whereas in formal grounded theory, theory, theory is developed based on applying hypothesis across several areas of research inquiry with a variety of sample populations. (Gilgun, 1992). The grounded theory approach enables the researcher to select cases using a sampling procedure that enables her to identify new cases that are similar to previous ones. (McRoy, ND). Observations are made for each case and if it is observed that no new information is provided by the cases, the process is repeated by making a selection of new cases that can yield different results until new information is obtained. (McRoy, ND).
- Appraisal of Slevin and Aday
Slevin and Aday (1993) sort to understand how gender shapes the responses to information on a variety of topics ranging from “Strategic Defence Initiatives (SDI) to funding for the arts”. The study was aimed at providing answers to a series of questions. For example, the study was interested in understanding whether there is “male hegemony” in some conversational domains; it also sort to determine whether “women’s self-perceived lower levels of achievement-related competence” as suggested by Frieze et al. (1978) and Maccoby & Jacklin (1974) reduce women’s confidence to engage in certain conversations as well as whether a generalised sense of inferiority influences women’s self-perceptions of competence to discuss certain issues. (Slevin and Aday, 1993). The study also investigated whether women and men feel equally informed on issues of foreign affairs and domestic public policy. (Slevin and Aday, 1993). Moreover, tests were also conducted to determine whether the above evaluations have any link to age, generation, and life cycle experiences. (Slevin and Aday, 1993). The study depended on the deductive approach to achieve its objectives.
The sample was selected from “alumni of four cohorts (classes of 1965, 1975, 1980, 1985) of graduates of a mid-Atlantic public university”. (Slevin and Aday, 1993). The questions asked in the questionnaires include: “during a discussion with friends, how informed would you feel in conversations on each of the following: SDI, the greenhouse effect, arms control, corruption in government, political change in Eastern Europe, government funding abortion, pollution, federal funding of the arts, race relations, the drug problem, health care, and the budget deficit?”. (Slevin and Aday, 1993). Respondents were asked to rate their levels of information on a 5-point scale ranging from 1 to 5 where 1 represented very competent and 5 represented not competent. (Slevin and Aday, 1993).
The first issue identified here, which remains a common problem in most inductive studies is the rating scale that is used. The Likert scale used here induces a bias in the responses. A rating scale of this nature is analogous to what we call in statistics, ordinal scales, which sort data into categories that are ordered with respect to some characteristics. (Defusco et al., 2009a). Although the scale sorts data into categories with respect to characteristics, it is difficult to determine whether the distance between a response of 1 and a response of 5. In the above case, a response of 1 means very competent while a response of 2 means a lower level of competence but the exact difference between 1 and 2 cannot be quantified. All we know is that someone who gives a response of 1 is feels more competent than someone who gives a response of 2. We are not aware of the magnitude of the difference in their competence. Thus, the first problem with the study is the measurement scale used in measuring how people feel informed in conversations. Moreover, the term “competent” has just been used arbitrarily. Any other term could have been used. For example, the researchers could have used “well informed” and “not informed” or better still “very knowledgeable” and “not knowledgeable”. These different means of retrieving the same information induces a bias in the manner in which the respondents respond to the questionnaire. It will not be surprising to observe different responses from the same respondent if the question is framed in another way. For example a different response can be observed from the same respondent using “well informed” against “not informed” as opposed to using “very competent” against “not competent”. Another problem with the study is the fact that respondents are considered to comprise of a group of similarly educated men and women in distinguishably different points in their life cycle. (Slevin and Aday, 1993). This may be considered another source of bias because although the respondents may be similarly, educated they may have different intelligent quotients (IQs) which may affect their level of reasoning and thus their responses to the questionnaire. The responses are then used as dependent variables against gender which is considered the independent variable. However, gender may not be the only variable determining the responses. Thus, gender may not accurately account for the variability in responses. A limitation of the study and thus of the deductive approach is that, it was difficult to address important dynamics of actual conversations such as the composition of the conversational unit or the effects of gender and power on conversational privileges and duties. (Slevin and Aday, 1993). The study was also interested in observing time series variations in the responses. (Slevin and Aday, 1993). However, the study employed cross-sectional data. This is another shortcoming of the study. They could have used panel data, which allows for both time-series and cross-sectional analysis grouped together in one data set.
- Conclusions and Recommendations.
This paper has observed the deductive and inductive approaches to research. As we have seen from above, one can conclude that the deductive approach is a top-down approach, which leaves from the general to the specific, while the inductive approach is a bottom-up approach, which leaves from the specific to the general. (http://www.socialresearchmethods.net/kb/dedind.php). The study by Slevin and Aday, (1993) reveal a number of shortcomings in the deductive approach. The approach depends a lot on sample size to be able to yield acceptable results and thus draw appropriate conclusions. however, large sample size does not necessarily mean that all biases have been eliminated as the central limit theorem may suggest. It may be argued that large sample size helps to reduce the standard errors of sample statistics. (Defusco et al., 2009b). However, a large sample may also mean that one is sampling from more than one population. That is we may go beyond the population of interest. Moreover, using a large sample increases the cost of collecting data. This makes the deductive approach a less desirable approach to research. On the contrary, the inductive approach employs grounded theory and focuses on understanding the behaviour of a small group, developing hypotheses based on patterns of behaviour and making generalisations. (http://www.socialresearchmethods.net/kb/dedind.php). Such an approach is not only cost-effective; it also avoids the risk of generalising observations from one case to other cases as is the case with the deductive approach. In a nutshell, both the inductive and deductive approaches are designed to build knowledge and it is common to see studies that employ both approaches as a means of arriving at accurate conclusions. (McRoy, ND). This paper therefore recommends the use of the inductive approach in most circumstances although the deductive approach may be appropriate in certain cases such as cases in which data points can be measured reliably using a ratio scale. Where necessary both approaches should be employed so as to provide better explanations of the phenomena exhibited by the populations under study.
REFERENCES
Babbie, E. (1986). Observing ourselves. Belmont, CA: Wadsworth.
Blumer, H. (1969). Symbolic interactionism: Perspective and method. Englewood Cliffs, NJ: Prentice Hall.
Bonoma, T.V. (1985), “Case research in marketing: Opportunities, problems, and a process”, Journal of Marketing Research, Vol. 22 pp.199 – 208.
Defusco, R. A., McLeavey, D. W., Pinto, J. E., Runkle D. E. (2009a). Statistical Concepts and Market Returns. In Ethical and Professional Standards and Quantitative Methods; CFA Program Curriculum, vol. 1, pp. 239-318; Pearson Custom Publishing.
Defusco, R. A., McLeavey, D. W., Pinto, J. E., Runkle D. E. (2009b). Sampling and Estimation. In Ethical and Professional Standards and Quantitative Methods; CFA Program Curriculum, vol. 1, pp. 421-453; Pearson Custom Publishing.
Epstein, I. (1988). Quantitative and qualitative methods. In R. Grinnell, Jr., (Ed.), Social work research and evaluation (pp. 185–198). Itasca, IL: F. E. Peacock.
Gilgun, J. (1992). Definitions, methodologies, and methods in qualitative family research. In J. Gilgun, K. Daly, & G. Handel (Eds.), Qualitative methods in family research (pp. 22–40). Newbury Park, CA: Sage Publications.
Gilgun, J. F. (1994). Hand into glove: The grounded theory approach and social work practice research. In E. Sherman & W. J. Reid (Eds.), Qualitative research in social work (pp. 115–125). New York: Columbia University Press.
Marlow, C. (1993). Research methods. Pacific Grove, CA: Brooks/Cole.
McRoy, R. G. (ND). Qualitative Research. Available online at: http://www.uncp.edu/home/marson/qualitative_research.html
Reid, W. (1987). Research in social work. In A. Minahan (Ed.-in-Chief), Encyclopedia of social work (18th ed., Vol. 2, pp. 474–487). Silver Spring, MD: National Association of Social Workers.
Rubin, A., & Babbie, E. (1993). Research methods for social work. Pacific Grove, CA: Brooks/Cole.
Research Methods Knowledge Base (2009). Deduction and Induction: Deductive and Inductive Thinking. Available online at: http://www.socialresearchmethods.net/kb/dedind.php
Sherman, E., & Reid, W. J. (1994). Qualitative research in social work. New York: Columbia University Press.
Slevin, K. F., Aday, D. P Jr. (1993). Gender Differences in Self-Evaluations of Information About Current Affairs. Sex Roles, vol. 29, Nos. 11/12, pp. 817-828.
Strauss, A., & Corbin, J. (1990). Basics of qualitative research. Newbury Park, CA: Sage Publications.
Taylor, J. (1977). Toward alternative forms of social work research: The case for naturalistic methods. Journal of Social Welfare, 4(2–3), 119–126.
Zurcher, L. A. (1983). Social roles: Conformity, conflict, and creativity. Beverly Hills, CA: Sage Publications.