The Parekh-Berger Hierarchical Interpretive Phenomenological Analysis Model

Interpretive Phenomenological Analysis (IPA) is a qualitative approach that allows researchers to understand and examine the ‘lived experiences’ of participants [4]. Smith et al developed IPA to capture the personal meaning and narratives that often are not adequately addressed with more quantitative forms of analysis [5]. IPA gives depth to understanding the human condition by allowing participants to explain their experiences in their own voice. It has been routinely found that IPA is an excellent method to examine constructs found within clinical psychology [5]. The domain of clinical psychology consists of addressing the emotional, psychological, behavioral, and intellectual adjustments of populations across the lifespan [6]. Much of clinical psychology research is focused on aggregating large amounts of quantitative data to understand people’s growth and development. In contrast, IPA is distinctive because it emphasizes the importance of finding common meaning across fewer participants by examining their rich personal expressions [7].
While there are a number of qualitative approaches available, the current authors have found that the IPA approach developed by Smith has produced very valuable results when analyzing the narratives of participants with unique ‘lived experiences [2]. The ability to aggregate and develop thematic content that shows trends in participants’ experiences makes it a particularly applicable methodological approach to those students conducting clinical psychology dissertations. As Smith highlighted, it is extremely important for the researcher to provide the conditions for the participants to feel safe and comfortable to share their stories [5]. Clinical psychology graduate students are ideal researchers to implement IPA approaches since they are taught methods for establishing rapport with others using verbal and nonverbal techniques.
Because of the synchrony between IPA and clinical psychology, the current authors have routinely used IPA with clinical psychology dissertations, which has aided our development of a hierarchical approach to IPA. The current authors have made several modifications to the original IPA methodology that will be discussed in this paper. The authors believe that these modifications are particularly helpful when applied to understanding clinical psychology phenomenon, as they lend themselves to providing a more holistic and inclusive narration of participants’ experiences. Allowing participants to share their stories on their own terms through a semi-structured interview can reveal important nuanced information that a quantitative study simply cannot replicate.

Smith developed Interpretative Phenomenological Analysis (IPA) to understand the participant responses in a single case design or involving several participants [2]. Traditionally IPA can be quite labor intensive, as it involves multiple steps of analysis [5]. According to Smith JA et al. [8] their methodology is not strict or inflexible, but rather it can be adjusted to suit the researcher’s empirical style and end goals [8]. It should be duly noted that all semi-structured interviews need to be transcribed for ease of analysis. The following is a description of the stages involved in conducting a traditional IPA.
The first stage is comprised of reading the transcripts several times in order to become familiar with the transcript content. Additionally, as the researcher becomes more familiar with the content, they can make comments in the left-side margins [8]. There are no explicit rules or requirements for making these types of comments and are based solely on the researcher’s impressions and observations [9]. In the second stage of IPA, the researcher rereads the transcripts and now begins to examine the information in terms of emergent themes, which are denoted on the right-hand margin [8]. This is perhaps the most challenging and crucial stage of analysis. In some instances, it may be easy to provide names and identifiers for themes (Specific Themes), yet in other cases it may require considerable interactions with the transcripts for Specific Themes to emerge from the text [9]. The key to this phase of analysis is developing themes that allow for connections across all participants but also maintains the uniqueness of each participant’s expression [8].
In the third stage of the IPA, the researcher begins to cluster the individual Specific Themes to capture theoretical or experiential similarities. These thematic clusters are then assembled into Higher Order Themes, which should comprehensively represent content across all participants. 8,9 In the final stage, the Higher Order Themes are written up highlighting the narratives of participants [9]. By identifying the overlaps between general themes to create a Higher Order Theme, IPA allows for theoretical unification, while preserving the unique ‘lived experience [8].

The first modification in the Parekh-Berger model is that the researcher does not read and reread every transcript. In some instances, rereading part or even the entire transcript might be done. In addition, the laborious process of note taking is not deemed necessary. Instead, the researcher relies upon their first impression-their own interpretive perspective as to possible Specific Themes and notes those. This simplified process has turned out to be sufficient for the generation of an initial list of possible Specific Themes.
Another distinctive adaptation that the current authors have made to traditional IPA is the utilization of interrater reliability. It is imperative that IPA research has tools to evaluate and denote the authenticity of its results [4]. Often, authentication involves having participants review the actual transcripts for accuracy or triangulating data among different methods or various stakeholders [4]. The current authors have applied the statistical concept of interrater reliability as a method to solidify the genuineness of themes found across transcripts [10]. Interrater reliability essentially examines if data collectors who examine the same situation or phenomenon will record the same observations [10]. Thus, data collectors with higher degree of agreement will have a higher interrater reliability. The current authors are proposing two methods of using interrater reliability to examine consistency in terms of themes: Blind Review and Independent Check Review. In Blind Review, researchers give another reviewer the participant transcripts and ask them to go through each transcript and denote individual Specific Themes. In this method, it is important to examine if the second evaluator, who is blind to the proposed Specific Themes of the first reader, will find similar Specific Themes in the transcript. The level of agreement between reviewer 1 and reviewer 2 is calculated using an interrater reliability analysis. Unfortunately, this method can be very time intensive and requires a significant commitment on the part of the second reviewer.
Another method that is not as cumbersome to implement is the Independent Check Review. In this method, the second reviewer receives a list of proposed Specific Themes found in the transcripts by the first reader. The second reviewer is to denote the frequencies or number of times they find a specific theme across the transcripts. The first reviewer has also conducted a frequency count of their proposed Specific Themes; however, the second reviewer is not privy to that information and must evaluate the frequencies blindly/independently. Finally, the second reviewer’s frequency count is compared to the first reviewer’s frequency count to determine if there is agreement across both reviewers. This method is less difficult to implement and still provides a solid assessment of interrater reliability. The current authors believe that these two methods are essential to bolstering the consistency of IPA designs. It is readily apparent that researcher’s individual training, experiences, and possibly biases may impact the review of transcripts. The current authors believe that reliability checks are a method by which to enhance the empirical robustness of IPA designs.

In the traditional IPA, the task after the Specific Themes have been delineated is to group them together into clusters. The clusters are thought of as Higher Order Themes. An analogy can be made to Factor Analysis. In a Factor Analysis, participants receive scores on a number of measures. Then, through correlational techniques, various of the measures (or scales) are found to be related to each other and not to other scales (or groupings of scales). Thus, these correlated measures are grouped together under the label: Factor. Because these are all quantitative relationships, it is possible to determine the amount of variance among the scores is accounted for by each Factor. It is our perspective that IPA can be thought of as a non-quantitative factor analysis. The Specific Themes can be thought of as non-quantitative measures. Through logical and conceptual analysis, the researcher groups Specific Themes together and gives a name to that grouping that captures/reflects the Specific Themes that the researcher determines have a common relationship with each other. These groupings or clusters of Specific Themes are considered to be Higher Order Themes. An example of a similar concept can be found with the Wechsler Intelligence Scales. There are individual measures (e.g., Coding and Symbol Search). These are found to correlate with each other and are grouped together under a Higher Order concept called: Processing Speed. In IPA, the goal is to determine if there are Specific Themes that can be grouped together in Higher Order Themes that “tell the story” of what emerges across all of the interviews. Finally, in traditional IPA, the final task with the Higher Order Themes is to see if they can be arranged in a numbered order such that the names of the Higher Order Themes is a short-hand expression of the story that the participants have shared about their phenomenological experience has been of the life events being investigated. An example of this is provided at the end of the next section of this report.

Figure 1: Pictorial Representation and Narrative of Universal Themes.

Another of the major modifications to Smith’s IPA approach is that we examine for Specific Themes that are expressed by every participant. We term these as Universal Specific Themes. It turns out that it is not a coincidence when every participant expresses the same Specific Theme in their separate and independent interviews. It is not that the interviewer probes for such Themes. They get expressed spontaneously during the course of each and every interview. We have found that when these Universal Specific Themes occur, they tell the story that emerges across the interviews of the experience that the participants have had in regard to the matter being investigated. Below, we provide a Figure from the dissertation of Ortega C et al. [11]. Psychologists who conduct child custody evaluations on referral from the Family Law Court were asked to share their experience of conducting such evaluations. Figure 1 clearly shows how Universal Specific Themes provide a short-hand version of the story that the Higher Order Themes reveal. As can be seen, across the psychologists, the story that emerged was that one has to start from a legal, ethical approach and then progress to addressing parent variables, then child variables, and then the psychologist her or himself is a variable in the mix with psychological testing factors having to be taken into account to complete the evaluation. As the bottom of the Figure show, these are not discrete steps, but each factor influences the others in the sensitive matter of child custody evaluations.

In a recent study, we had the experience of discovering that it is even possible to have Superordinate Higher Themes that subsume the Higher Order Themes that are extracted from the interviews [12]. In this study of women and men who had been sexually abused, and who reported that they had forgiven their abuser. In this study of the forgiveness process, a number of Higher Order Themes were extracted. Upon further analysis, it was determined that the Higher Order Themes themselves could be further grouped in a Hierarchy that contained the Higher Order Themes and their included Specific Themes. What also stood out was that the Superordinate Higher Order Themes turned out to reflect the model that had been proposed for how such forgiveness could occur psychologically. Table 1 presents the Superordinate Themes and the Higher Order Themes subsumed within them.

The Parekh-Berger HIPA Model begins in the traditional manner. A determination is made of a psychological phenomenon of interest. The researcher defines a population for whom that phenomenon is relevant. Then, through sampling processes, participants are sought who agree to participate in a semi-structured interview. Using predetermined, mostly open-ended questions, the interviewer seeks to facilitate the participant expressing in their own words what their life experience has been like and what it has meant to them.

When the interviews are completed, the researcher reads and reviews the transcripts and attempts to get a sense from each of the transcripts what are the unique, individual themes that seem to be expressed by the participant. The reviewer is attempting to identify the individual pieces (Specific Themes) that are being expressed by each participant. The reviewer is attempting to specify the elements of the participant’s internal representations of what they experienced-the phenomenology of the participant’s experience.
Once the first reader (reviewer) has extracted all of the possible Specific Themes that they can identify (whether it is from one reading of each transcript or several readings of any or all of the transcripts), a second reader/reviewer is now involved. A decision is made as to whether this reviewer will use the Blind Review method of attaining an interrater reliability or the Independent Check Review method. Accordingly, a final list of Specific Themes is created that are not just the product of one person’s perceptions of the interviews, but that there is some degree of independent assessment and concordance regarding the Specific Themes that appear to have been expressed by at least two participants and recognized by two independent reviewers. Each Specific Theme is now given a final official name. The Specific Themes are numbered in the accordance with the sequence in which they were identified in the transcripts.
Now comes the task of creating Higher Order Themes. Again, the Parekh-Berger HIPA Model involves at least two researchers. In the case of dissertations, the Chair of the Dissertation Committee and the student then make an effort to group the Specific Themes into Clusters that are called Higher Order Themes and given names that are to reflect the concept that the Specific Themes have in common. In other types of studies, the lead researcher enlists the involvement of another researcher in the determination of Higher Order Themes. It should be understood that the names of the Higher Order Themes are an abstraction representing, in some sense, more of the phenomenological experience of the participants that reflects a conceptualization of the individual themes that cluster together.
The next step in the Parekh-Berger HIPA Model is to determine an ordering of the Higher Order Themes. We have found that almost invariably, the Higher Order Themes can be placed in a numbered order such that their names basically tell the “story” that emerges across the interviews that reflects the common, phenomenological experiences of the participants, in a sort of short-hand way. Finally, an examination if made of the Specific Themes of those that were expressed by every participant. These Specific Themes are named: Universal Specific Themes. Again, we have found that these Universal Specific Themes also tell the story, in greater specificity than the reading of the Higher Order Themes, yet less specificity than the entirety of all of the Specific Themes. This process can be seen in Figure 1.
Finally, it may be possible that even the Higher Order Themes can be grouped into clusters that tell the story expressed by the participants. These types of themes are called Superordinate Higher Order Themes. We have not found this to be common in the several dozen HIPA studies of which we are aware. However, one recent example emerged and can be seen in Table 1.
Our method for reporting the results of the Parekh-Berger HIPA Model is to first present the list of Specific
Themes that has emerged retaining their sequential number as part of their identification. Next, we present a Table showing the Higher Order Themes in their numbered order with their constituent Specific Themes listed under them. If Universal Specific Themes are found that help delineate the story, then a chart or Figure is constructed and presented at that point such as is seen in Figure 1. Next, if Superordinate Higher Order Themes are created, those are then presented in a Table or Chart such as is seen in Table 1. Finally, the Results section of a report will present specific quotes from the transcripts that illustrate the Specific Themes and therefore the bases for which the various Higher Order Themes were created. Thus these Tables or Charts are organized and presented for each Higher Order Theme, one at a time, with quotes for each of the Specific Themes that constitute that Higher Order Theme. Now comes the task of creating a Discussion section to explain and expand upon the perspectives gained from this hierarchical ordering of Specific Themes, Higher Order Themes, possible Universal Specific Themes and Superordinate Themes.

The most traditional method for reporting the results and conclusions from a study is to create a journal article with introduction, methods, results and discussion section. Often, that model suffices for reporting statistical tests and presenting interpretations and discussing the results. Often, the traditional method also works well for reporting IPA results for traditional IPAs as well as for the Parekh-Berger HIPA Model. However, we have found that the traditional journal format does not always do justice to the wealth of personal experience that is revealed through IPA methodology.
As an example that we can offer is the dissertation of our former student, Tucker M et al. [13]. She interviewed therapists who conducted psychotherapy using sign language with hearing impaired patients. It was determined that the experiences of these therapists would not be properly shared by a traditional journal article. It was agreed that much more detail and information needed to be reported. Consequently, a e-book was created. Presented below is the Table from her dissertation (her Table 1, our Table 2) that depicts the Higher Order Themes that were created with their constituent Specific Themes.

Table 1: Superordinate themes and concomitant higher order themes that reflect the forgiveness process experienced by women and men who were sexually abused and report having forgiven their abuser.

Table 2: Higher order themes and component universal specific themes.

Note: The number of the universal specific themes is presented in brackets.

We have presented a model for conducting Interpretive Phenomenological Analysis that we have been using for a number of studies. The model from which we have worked is based on the methods developed by Smith JA [2]. We have termed our approach the Parekh-Berger Hierarchical Interpretive Phenomenological Analysis Model (HIPA). We and our colleagues have utilized the HIPA model in dozens of studies. Our Model has been applied in a nomothetic method. Consequently, we have relied upon having interrater reliabilities so that the resultant themes are not just a reflection of one person’s perspective of the “story” that emerges from the analyses of the interviews. In this article, we have provided the rationale for the HIPA model, delineated the steps in extracting themes from the interviews of participants and provided a guide as to how to report the results and interpretations of the resultant themes. We have also provided examples of what such results are best presented. It should be noted that by providing the list of the Specific Themes and the resultant Higher Order Themes along with specific quotes from the transcripts, readers are well able to create their own phenomenological interpretations of the presented results.

1. Willig C (2001) Qualitative research design. Introducing qualitative research in psychology: Adventures in theory and method. Open University Press, Buckingham, UK.
2. Smith JA (1996) Beyond the divide between cognition and discourse: Using interpretative phenomenological analysis in health psychology. Psychology and Health 11(2): 261-271.
3. Smith JA (2004) Reflecting on the development of interpretative phenomenological analysis and its contribution to qualitative research in psychology. Qualitative Research in Psychology 1(1): 39-54.
4. Alase A (2017) The Interpretative Phenomenological Analysis (IPA): A guide to a good qualitative research approach. International Journal of Education & Literacy Studies 5(2): 9-19.
5. Smith JA, Flowers P, Larkin M (2009) Interpretative phenomenological analysis: Theory, method and research. (2^nd edn), Sage Publications, California, USA.
6. American Psychological Association (2008) Clinical Psychology. American Psychological Association, Washington DC, USA.
7. Creswell JW (2003) Research design: Qualitative, quantitative, and mixed approaches, Sage Publications, California, USA.
8. Smith JA, Osborn M (2003) Interpretative phenomenological analysis. In: Smith JA (Eds.), Qualitative Psychology: A Practical Guide to Methods, Sage Publications, London, UK, pp. 51-80.
9. Fade S (2004) Using interpretative phenomenological analysis for public health nutrition dietetic research: A practical guide. Proc Nutr Soc 63(4): 647-653.
10. McHugh ML (2012) Interrater reliability: The kappa statistic. Biochemia Medica (Zagreb) 22(3): 276-282.
11. Ortega C, Berger SE (2016) Qualitative analysis of child custody evaluation practices. International Journal of Psychological and Behavioral Sciences 10(6): 2126-2137.
12. You M, MacMillin M, Berger SE (2022) Intrinsic motivation, The process of forgiveness, and the effect of forgiveness on healing for sexual abuse survivors.
13. Tucker M, Berger SE, Parekh B (2020) A phenomenological examination of the hearing therapist-deaf patient dyad: Barriers, language, culture, and training, e-book. Journal of Complementary and Alternative Medicine, Iris Publishers, California, USA, pp. 1-38.