Many, perhaps most, psychology researchers apply the quantitative method, which involves subjecting numbers to statistical analyses. Nevertheless, an alternative approach, qualitative research, is becoming increasingly popular (also see Constructionism). This perspective often shuns the use of numbers, and instead collects, analyses, and interprets words.
Qualitative research has filtered into many areas of psychology, especially in applied domains. Researchers have used this approach to understand motivations of criminals, experiences of patients, and responses to catastrophes for example. Nevertheless, virtually all studies can potentially be buttressed and elaborated with a qualitative component. Furthermore, many studies generate qualitative information in the form of short answers or anecdotes, but do not utilise this material effectively. This document provides a cursory overview of qualitative research.
Qualitative research involves the collection, analysis, and interpretation of words rather than numbers. Data are collected via interviews, documents, or observations. The objective of this form of research is to explore, explain, describe, and appreciate the experiences, feelings, and motivations of individuals in particular settings. In contrast to the positivist nature of quantitative research, exponents of the qualitative approach often regard science as subjective and phenomenological.
The process of qualitative research is also very different. The design, procedures, objectives, theoretical framework, and hypotheses are not often specified in advance. Instead, all of these elements gradually evolve throughout the study.
The use of numbers varies somewhat across schools of thought. For instance, many qualitative researchers eschew numbers altogether. They believe that counting the frequency of some object, event, or quality is artificial. Counting involves classifying an entity into a category. This classification will disregard the subtle and unique properties of that event or object. These unique attributes should be underscored rather than dismissed.
In contrast, some qualitative researchers undertake some counting. For instance, they may report the frequency of some theme or issue to bolster their arguments. Nonetheless, these numbers are always regarded as peripheral to the key focus of any study.
Qualitative research has evolved from numerous domains and philosophies. Nevertheless, most qualitative studies comprise four phases: data collection, data reduction, data display, and generating conclusions. These phases are often undertaken concurrently and iteratively, rather than sequentially (see, for example, Grounded theory).
Briefly, data collection involves gathering information from interviews, memos, documents, personal journals, and emails. Alternatively, the researcher may observe a group, organisation, or individual from afar.
To some extent, this process is guided by a theoretical framework. However, the researcher must ensure they are not blinded by this framework. That is, they must ensure the framework does not bias their interpretations of the data. Indeed, some researchers do not rely on a conceptual framework at all. One complication, however, is that individuals who do not operate with a conceptual framework might not be able to decide which data is relevant.
In addition, researchers must decide which individuals they will interview or observe. Quantitative researchers argue that samples should be random, otherwise conclusions cannot be generalised. In contrast, qualitative researchers usually select informative and credible individuals only, called purposive sampling. Their rationale is that informative individuals can facilitate the development of a comprehensive and valid theory. Once a complete understanding has been obtained, the applicability of this theory to other situations can be determined in advance. In other words, internal validity, that is, a deep appreciation of one situation, can lead to external validity, that is, an understanding of other situations.
Furthermore, reseatchers must decide whether or not interviews will be recorded verbatim? Selective recording of answers provides several shortfalls. First, this tactic may provide demand characteristics& recording certain answers provides tacit approval of their applicability. Second, disregarding certain information is hazardous, because material that may not seem relevant at one point may become vital later. Nevertheless, verbatim recording can be cumbersome and can inhibit the interviewee.
Data reduction primarily involves coding the raw data. The researcher often scans the material and immerses themselves in the data, while applying category names and phrases to segments of text as required. Gradually the codes are refined and systematised until all segments of text have been categorised. Coding ensures the data is manageable, assists the retrieval of particular segments later, and highlights common patterns or themes.
Three approaches have been devised to create codes. First, some researchers begin with an a priori list of codes--such as "academic problem", "personal problem", and "social problem". Second, some researchers begin with an a priori list of broader categories, such as "personality triats" and "external events". Using these categories, the researcher then reads the data and attempts to devise more specific codes. Finally, some researchers immerse themselves in the material without any preconception about codes. Codes and theme emerge over time, often spontaneously.
After creating an initial set of codes and then applying these codes to the data, the researcher refines this output. For instance, some codes may share some abstract quality. Hence, these codes can be converted into a broader category, called a pattern code. These pattern or abstract codes are analogous to theoretical constructs and a highly informative. For instance, after reading the transcript of a University student, three codes may emerge: "labelling lecturer as arrogant", "despising students who study diligently", and "feeling cheated". These codes can be amalgamated to the pattern code "resentment of high achievers".
Furthermore, during this process, several revelations may ermege. Any thoughts, concerns, ideas, uncertainties should be recorded. Often, a thought that initially appears to be extraneous may ultimately become crucial.
During the data display phase, the codes and words are represented in tabulated matrices, cause-effect networks, and sometimes scatterplots. Alternatively, vignettes or quotes are collected. These displays and descriptions both guide the development of theoretical development and also provide a means to communicate the results. A matrix is merely a cross-tabulation of rows and columns,.
In this instance, the columns reflect different roles, and the rows reflect different issues. Rows and columns can also represent different times, causes, effects, events, and so forth. The entries within each cell may be short blocks of text or summaries, quotes, symbols, ratings, etc. Objective criteria are needed to determine whether or not a particular segment of text should be incorporated into the table.
Sometimes the researcher creates a host of related matrices, such as one matrix for each year. These matrices are then concatenated. The order in which the matrices appear is consistent with some variable, such as average temperature for each year. As the researcher scans the matrices in sequence, a pattern may emerge. This pattern forms the basis of theoretical conclusions.
A causal network that specifies the relationship between constructs is presented below. Most of this information is derived from codes, such as "low study" codes normally precede the "low understanding codes". Nonetheless, some of this information may arise from direct quotes, such as "I started studying because I was uncertain about the future".
The researcher then refines and develops these displays in an iterative manner. Redundant categories are combined, and potential moderators are considered.
The researchers must be circumspect when using vignettes: unless stated otherwise, readers will tend to erroneously regard the vignette as representative. In other words, when communicating information, the researcher must understand the biases, assumptions, and limitations of the audience, otherwise their claims may be misconstrued.
Finally, conclusions are then derived from the codes and displays. Typically, the theories derived from this form of research resemble the theories derived from quantitative studies. Nonetheless, a few special issues need to be considered.
As a consequence of the subjective nature of this endeavour, researchers often consider plausibility as a criterion to assess theories. That is, theories are evaluated according to their intuitive appeal. Two problems emerge from this strategy. First, interpretations contaminated by biases will often seem plausible. In other words, plausibility may simply reflect the biases that spawned this theory from the outset. Second, many great, novel theories seemed implausible at first, such as quantum mechanics.
Furthermore, the depth and breadth of data that emerge from this approach can be utilised to assess rival explanations. In other words, the researcher should continually attempt to create alternative theories and assess these concepts with new data. Apart from assessing the empirical validity of each theory, the researcher ascertains the explanation that is less likely to be the culmination of biases and pitfalls. For example, the theory that is less likely to simply reflect representative, holistic, elite or confirmatory biases is preferred.
Qualitative research is often used to complement quantitative techniques. For instance, the qualitative component can be undertaken to uncover issues, attributes, problems, and other variables that can then be subjected to a subsequent quantitative study. In other words, the qualitative procedures provide the groundwork for a more rigorous study.
Alternatively, qualitative research can be undertaken to explore the relationships that were derived from a quantitative study. In particular, the qualitative component can attempt to identify the sources, mediators, reasons, motivations and limitations of some relationship.
Finally, the qualitative study can incorporate the setting, context, and culture. That is, quantitative studies tend to treat everyone as equivalent. A score of 80 on some IQ test is considered to be equivalent irrespective of the circumstances. In contrast, qualitative studies can attempt to understand the impact of these contextual variables. In other words, qualitative research can explore the subtleties of a situation.
Despite the benefits of qualitative research, several shortcomings need to be recognised. Many of these concerns, however, are misguided or resolvable. For instance, some researchers complain about the excessive workload. Admittedly, qualitative research is a demanding and laborious process. Nevertheless, each study can yield a wealth of information. The ratio of output to input can thus be higher in qualitative research relative to quantitative research.
Furthermore, some investigators question the external validity of qualitative research. That is, the conclusions do not seem to be applicable to other situations. Proponents of qualitative research challenge this claim. First, they do not regard external validity as crucial. In particular, each situation and context is unique, and hence no conclusions can be truly generalisable. Second, the profound understanding that is sought can actually assist external validity. That is, this understanding enables researchers to appreciate whether or not their conclusions will also apply to another situation.
In addition, another concern relates to researcher effects. Specifically, the presence of a researcher can influence a situation. This problem is exacerbated by the common practice of forming relationships with the individuals involved in the investigation to generate trust. Nonetheless, these researcher effects can be mitigated. For instance, researchers may employ an informant to monitor the effect of their presence. Alternatively, the researcher may utilise unobstrusive measures as well, and so forth.
The final problem, however, is the most consequential. In particular, the results can be contaminated by various biases. For instance, researchers may also be sensitive to confirmatory bias, where they only notice events that conform to their expectations. In addition, the holistic fallacy may undermine conclusions. In particular, events are regarded as more structured than perhaps they are in reality. This bias may be responsible for the perceived relationship between a full moon and lunacy& two salient events are presumed to be connected. Finally, researchers may be susceptible to elite bias and thus tend to trust articulate and senior informants only.
Thematic analysis is a technique that is often used to analyze qualitative data, such as responses from interviews and focus groups (Boyatzis, 1998& Braun & Clarke, 2006& Roulston, 2001). In essence, to apply this technique, researchers first peruse and read the data in detail several times, while initial ideas or impressions are recorded. Next, chunks of data that are interesting or insightful are assigned an initial set of codes. Then, to uncover themes, similarities or relationships between codes are identified. Furthermore, to ensure these themes are comprehensive and representative of the data, the data are scrutinized again, as the themes are modified and delineated. Finally, these themes are assimilated into a meaningful and unified narrative.
Unlike some other qualitative techniques, thematic analysis permits significant discretion on the part of researchers. Researchers do not have to follow a specific set of procedures or uphold a particular philosophy--such as phenomenological epistemology. Indeed, thematic analysis entails several other paradigms. Yet, this technique has been defined clearly enough to offer some clarity and guidelines.
Braun and Clarke (2006) differentiate six key phases of the analysis: becoming familiar with the data, generating initial codes, searching for themes, reviewing themes, defining themes, and producing the report. The codes are really summaries of the data, whereas themes are patterns or sets of related codes that correspond to a concept that could be interesting.
Braun and Clarke (2006) have emphasized some of the key decisions that researchers need to reach as they analyze data. For example, researchers need to decide whether to uncover all the key themes of a dataset or to characterize one or two key facets or themes in great detail. Second, they need to decide whether they would prefer to infer the themes from the data or, at least initially, to apply some established themes, perhaps derived from past theory& that is, they need to ascertain whether the analysis will be primarily inductive or deductive. Third, they need to decide whether to code the surface meaning of comments or to consider the underlying assumptions, motivations, and perspectives of participants& that is, they need to ascertain whether they will be more descriptive or interpretative.
Finally, Braun and Clarke (2006) enumerates some of the criteria that can be applied to evaluate thematic analysis. These criteria also provide some insights into how to undertake thematic analysis. Specifically, they argue:
Many researchers combine quantitative and qualitative data. For example, when researchers undertake the Sequential Explanatory Strategy (Creswell, 2011), they first collect and analyze quantitative data. Then, qualitative research, often comprising interviews, is conducted to understand the mechanisms that underpin the relationships that were uncovered during the quantitative phase.
Boyatzis, R. E. (1998). Transforming qualitative information: Thematic analysis and code development. Sage.
Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3, 77-101.
Creswell, J. W. (2011). Research Design (3rd [South East Asian Ed.] ed.). New Delhi: Sage.
Roulston, K. (2001). Data analysis and 'theorizing as ideology'. Qualitative Research, 1, 279-302.
Last Update: 6/1/2016