Paradigm Parley: A Framework for the Dialectic Stance

Characteristic 1: Sustain a Data Dialogue

The first characteristic is that the researcher seeks and sustains a dialogue among data collected within differing paradigmatic perspectives (Greene & Hall, 2010). This conversational process occurs aloud and in written documents when conducted by a team of researchers able to dialogue together. However, when conducted by an independent researcher, the conversational process is an internal dialogue documented in a journal or in research memos (Greene & Hall, 2010). This conversational process requires a researcher adept at switching between two different paradigmatic perspectives and/or one competent in seeing the sum of disparate parts. The ability to make these transitions between perspectives is identified by Johnson (2015) as fundamental to the dialectic stance:

Making Gestalt switches is at the heart of mixed methods research, when a single researcher hopes to conduct MMR at the paradigm level. If a single researcher has not obtained competence at this, it is readily obtained with a heterogeneous research team composed of reasonable, open-minded researchers whose primary paradigms are different. On a heterogeneous research team, the mixed researcher needs to facilitate, and emphasize, good communication and translation, listening, empathy, process, and openness to larger and more complex wholes than what is obtained via a single paradigm. (Johnson, 2015, p. 690)

Characteristic 2: Give Equal Voice

The respectful dialogue of Characteristic 1 leads directly to the second key characteristic: all paradigmatic perspectives used in the study are given equal voice in the dialogue. While the mixed methods literature uses the terms equal weight or equal status in identifying a mixed methods study where the quantitative and qualitative portions of the study are equal (QUANT/QUAL), the idea of equal voice discussed here furthers this idea of equality into the mixed methods analysis or integration phase. As a researcher seeks to integrate the quantitative and qualitative data, he or she must constantly monitor the source of the data being analyzed and be certain that cognitively he or she is interpreting these data within their original paradigmatic perspective. Finally, although a statistically significant result from a large sample might feel more “powerful” to some researchers than the voice of one participant, the researcher must allow both pieces of evidence an equal voice as he or she integrates the two phases of data. All forms of data must be contemplated for what they offer to an increased understanding of the researched phenomenon. An independent researcher cannot ignore the paradigmatic perspective with which he or she feels less competent, anymore than a team of researchers can negate the perspective of one member of the team. The equality of voice in the dialogue between collected data necessitates openness to complexity and to the possibility that deeper insights will occur as a result.

Characteristic 3: Preserve Data Integrity

By giving equal voice, no one paradigmatic perspective subsumes the other. Within this equality of voice is the third characteristic: preserve data integrity. In order to give all involved paradigmatic perspectives equal voice in the dialogue, the researcher cannot transform qualitative data into quantitative or vice versa. To transform data is to give more power, and thus, more voice to one perspective over the other. This is not to negate the work of mixed methods scholars who choose data transformation as a legitimate form of mixed methods analysis; rather, I suggest that transformation runs counter to the underlying democratic principles of Greene’s (2007) dialectic stance. Insights generated through the preservation of data integrity are typically complex, detailed, and potentially confounding, but data integrity is essential when all paradigms are given equal voice.

Characteristic 4: Value Consonance and Dissonance

The complexity generated by preserving data integrity leads to the fourth characteristic of the dialectic stance: Results that are dissonant must be valued as much as results that are consonant (see, Greene & Hall, 2010). A primary tenet of the dialectic stance is that difference resulting from the use of multiple paradigmatic perspectives “must be honored in ways that maintain the integrity of the disparate paradigms” (Greene & Caracelli, 1997, p. 8). The dialectic stance encourages the complexity and richness of understanding that occurs when multiple perspectives are brought together. In this way, a researcher adopting the dialectic stance seeks both convergence and divergence when mixing data from multiple methods. The researcher does not strive only for convergence or data triangulation but also seeks divergent data within the mix of methods. This divergence must be discussed openly and equally with the data that converges. According to Teddlie and Tashakkori (2010), “this emphasis on divergent results often provides greater insight into complex aspects of a phenomenon, which can then lead to more in-depth investigation of previously unexplored aspects of that phenomenon” (p. 9). The dialectic stance thrives on both the convergence and divergence that emerge when multiple views of the collected research are mixed.

The four characteristics of the dialectic stance return us to the cartoon description that opened this article. In dialectic analysis, the goal is a parley between the paradigms used in the research study. The paradigmatic perspectives used in data collection metaphorically sit on the shoulders of the researcher(s), each expressing his or her view of the phenomenon investigated. While the researcher may feel torn between these conflicting views, it is the researcher’s responsibility to allow these perspectives to dialogue, then equally account for each perspective in the data presentation. The dialogue generated by this form of mixed methods analysis is not straightforward and is often circular or iterative in nature. The subsequent discussion of the insights that emerge from the dialogue is a net gain in understanding, one that is much greater than if the two perspectives had not dialogued.

While Greene and Hall (2010) provide a hypothetical example in which the dialectic stance is used by a single researcher, I have found only a few mixed methods studies explicitly identifying the dialectic stance as the conceptual stance used. Mirchandani et al. (2016) and Leal et al. (2016) both use the dialectic stance with a team of researchers. Headley (2016) and Howes (2017) use this stance as independent researchers and provide the most practical discussions regarding the application of the dialectic stance. However, all four studies were published recently, long after the dissertation study upon which the framework proposed here was conceived. Additionally, none of these studies explicitly take into consideration all four of the characteristics identified above.

The purpose of this article is to put forward one possible framework for conducting mixed methods analysis using Greene’s dialectic stance. Given that the dialectic stance values multiple viewpoints, I put this framework forward as one possible approach to dialectic mixed methods analysis, not the framework. My goal is to provide a practical tool for individual researchers seeking to conduct independent mixed methods analysis using Greene’s dialectic stance (although research teams might also find this framework practical). Applying the four characteristics described above, I first outline the framework and then provide a brief example of how the framework was applied in the analysis I conducted during my dissertation. I conclude with recommendations drawn from lessons learned, potential applications, and a call for ongoing dialogue.

Analysis Techniques Used

The proposed framework is based on two analysis techniques. The first is a mixed methods approach described by Smith (1997). In her study, Smith and colleagues combined case study/focus group data and survey data in a mixed methods analysis she developed from Erikson’s (1986) analytic induction model. In this analysis, Smith returned to the original interview transcripts and field notes from her case study/focus group phases as well as the descriptive statistics from her survey phase in order to work with the least-processed level of data. Smith reread all pieces of data and tracked her thematic thinking using memos. After reading the data several times and attempting to view all forms of data as equally important, Smith developed assertions she believed that could be supported by the data. Finally, she organized the data into warrants that either confirmed or contradicted the assertions. When data from both the case studies/focus groups and the survey supported an assertion, this strengthened her mixed methods analysis. When the disconfirming evidence outweighed the confirming evidence, Smith rejected the assertion. It is important to note that Smith sought only convergence, not divergence, when she analyzed her data across the phases of her study, which violates one of the four characteristics of the dialectic stance as described above. Although Smith’s proposed technique involves mixed methods analysis and provides a partial template for the proposed method, Smith does not explicitly claim that she is mixing at the paradigmatic level, perhaps because Greene’s dialectic stance had not yet gained currency in the literature.

The second analysis technique used in the proposed framework is the cyclical reading, coding, rereading, and writing technique of analysis often used by narrative researchers and described by Clandinin and Connelly (2000). In narrative inquiry, analysis and text construction are intertwined in a cycle of writing and revision that requires researchers to continually return to the original data. According to Clandinin and Connelly (2000), “a narrative inquirer spends many hours reading and rereading field texts” along with other data sources (p. 131) in order to “search for patterns, narrative threads, tensions, and themes” (p. 133). This technique was modified for the proposed framework such that as the researcher moves through the framework he or she must continually return to the unprocessed data to reread, recode, and reprocess. There are two notable differences between narrative inquiry and what has been adapted for the approach to dialectic mixed methods analysis suggested here. First, Clandinin and Connelly do not suggest that narrative analysis is a series of steps, but rather an organic and lived process. The framework below is described in steps, though these are somewhat fluid and a researcher can move both forward and backward in the process as needed. The second difference is that writing and analysis are not intertwined as is common in narrative analysis. Although a researcher will document his or her analysis process, writing of complete sections or drafts is not required (nor prohibited) in the proposed dialectic analysis. Although the iterative process of returning to the data described here is drawn from Clandinin and Connelly’s (2000) narrative analysis, the focus is on the cyclical process of rereading, recoding, and reprocessing the data, not the coding technique used nor the crafting of a researched story (other aspects of narrative inquiry). Within this framework, a researcher might use qualitative coding techniques drawn from a variety of qualitative methods to code for themes or conduct other textual analysis. This decision is best left to individual researchers guided by their research questions and theoretical frameworks. Returning to the data in the ongoing manner described above enables the researcher using the dialectic stance to be continually reminded of the original forms of data collected and the paradigmatic perspectives used for data collection.

Applying the Four Characteristics of the Dialectic Stance

The proposed dialectic analysis framework is detailed in Figure 1. The two analysis techniques discussed above form the backbone of the proposed framework. Although both analysis techniques come from an interpretivist perspective, “whether data happen to be in the form of words or in the form of numbers should not materially affect the process of constructing meaning” across the two forms of data because “analysis of both quantitative and qualitative data involves construction and reduction at appropriate times” (Smith, 1997, p. 80).

OPEN IN VIEWER

Figure 1.

The proposed framework.

Figure 1 provides a visual representation of how the four characteristics of Greene’s dialectic stance are used in this framework. First, the diagram has a solid background that is the foundation on which all the analysis occurs. The foundation of this analysis is the dialogue between different forms of data and requires that the researcher using the framework maintain an equal and ongoing dialogue among the paradigmatic perspectives used in the study. This foundational premise (denoted in the diagram by the gray background circle and the encircled text) combines Characteristics 1 and 2 of the dialectic stance into one overarching principle for the analytic framework. Second, the inner circle at the heart of the framework specifies that the integrity of the qualitative and quantitative data is maintained; words are not transformed into numbers nor are numbers converted into words at any point in the analysis. This circle represents the third characteristic of the dialectic stance as outlined above. These two components of Figure 1 represent the two guiding analytic principles that the researcher must constantly keep in mind as he or she moves through the analytic process. The researcher does not move into or out of these components of the diagram; rather, these two components are ever present throughout the analytic process.

The detailed middle arrow of Figure 1 represents Characteristic 4 of the dialectic stance. The researcher begins the analytic process by using the least processed form of all data collected and then begins a search for both consonant and dissonant findings across the data. After rereading the data, the researcher moves clockwise along the middle arrow of Figure 1 through four main analytic steps: first, generating assertions that he or she believes that are supported by the collected data; second, searching for data across the phases of the study to support/contradict the assertions; third, rejecting any assertions that are neither supported nor contradicted by both qualitative and quantitative data; and fourth, generating dialogic writing about each assertion supported by the findings. At each step along the analytic process, the researcher returns to the data to reread, recode, and reprocess the data before proceeding to the next step. Although this process is described here in four clockwise steps, a researcher may choose to move counterclockwise around the middle arrow as necessary during the process of analysis.

As the researcher moves through this analytic process, a reflective journal can be useful for noting any challenges or conflicted understandings that occur, particularly given that an individual researcher may often feel as though he or she is battling conflicting internal voices representing the two paradigms. Despite this battle, it is essential that the researcher maintain an ongoing dialogue between the two paradigmatic perspectives used, ensuring that each has an equal voice. Critical to the entire process is that the researcher must be careful to interpret each data form only within the paradigmatic perspective within which it was originally collected, thus, the inner circle of Figure 1. This is perhaps the most difficult part of the dialectic stance—the paradigmatic perspectives sitting on each shoulder must each be acknowledged, valued, and given equal weight in the analysis process—in essence, the paradigmatic perspectives must parley.

Step 1: Generate Assertions After Reviewing the Least Processed Form of All Data

When I combined all the data together for the first time, I began with the original data as collected, in my case, descriptive statistics, observation notes, and interview transcripts. Following Smith’s (1997) approach, I read through all pieces of data and thought about the statements or assertions I might be able to make about the data. In my notes, I generated nine assertions that I believed might be supported by the collected data.

For illustrative purposes, I will use only one assertion here to demonstrate the process. This assertion is “a music teacher’s knowledge of and confidence in young adolescent development comes from his/her teaching experience, not from undergraduate teacher education (preservice).” At this point in the process, it is not important whether or not this assertion is actually supportable by data as this occurs in subsequent steps. During this first step, it is only important for the researcher to write down any assertions he or she believes that might be supportable by the collected data. In the next two steps of the process, if no data to support or contradict the assertion exists within either the quantitative or qualitative data, the researcher is required to stop analysis and reject the assertion. It is important to note that because the dialectic stance encourages consonance and dissonance as stated in Characteristic 4 (above), dialectic researchers only stop the analytic process if no data exist within one phase of the study, not if the data from the phases contradict one another.

Step 2: Search the Data for Support/Contradiction of the Assertions

Next, the researcher moves clockwise along the arrow and returns to the data to reread, recode, and reprocess, always maintaining the data within their paradigmatic perspectives. It is at this stage of the process that the researcher needs to contemplate what new codes or statistical tests are needed in order to discover if there are data available to support or contradict the assertions put forward. In doing so, the researcher never transforms the data, but rather reanalyzes using techniques appropriate for the respective paradigm. For the assertion above, I needed to code the qualitative data for statements focused on knowledge regarding young adolescent development, statements related to confidence on this topic, as well as any identification by each participant regarding where he or she learned about young adolescent development. I also needed to run statistical tests on the quantitative data in order to compare the survey respondents’ reported influence of preservice to the reported influence of teaching experience specifically for the survey items focused on knowledge of young adolescent development. I identified items on the original survey instrument that would allow me to do so (see Table 2). Had I not been able to develop codes or identify appropriate items on the survey instrument, my dialectic analysis would cease for this particular assertion. In recoding and running additional statistical tests, I needed to maintain the paradigmatic perspectives of interpretivism and realism under which the narrative (QUAL) and survey (QUANT) data were originally collected. I could not, for example, count the number of times a narrative participant talked about his or her preservice and compare these textual codes with the number of times he or she referred to teaching experience. This violates two of the four characteristics established previously: First, it transforms qualitative data into quantitative; and second, the data transformation negates the meaning of the qualitative data within the interpretivist paradigm. Likewise, I could not pull out one individual survey respondent and discuss that data subset as it would violate fundamental principles of realism. While these examples are perhaps rudimentary, they are provided for illustrative purposes.

OPEN IN VIEWER

Table 2.

Influence of Teaching and Preservice Experiences on Confidence With Young Adolescent Development.

	χ2 results				Likert-type scale scores (totals)
Variable (characteristic of young adolescent development)	N	df	χ2	p	Not at all	Small extent	Moderate extent	Great extent
Teaching experience
Physical transitions of puberty	690	9	70.04	.000	0.29	1.16	18.70	79.86
Cognitive transitions of puberty	689	9	62.84	.000	0.29	1.16	18.70	79.86
Development of abstract thinking	689	9	70.57	.000	0.29	1.16	18.87	79.68
Search for competence	687	9	57.59	.000	0.29	1.16	18.78	79.77
Development of personal identity	688	9	58.18	.000	0.29	1.16	18.90	79.65
Desire for personal autonomy	688	9	57.06	.000	0.29	1.16	18.75	79.80
Desire for belonging and peer acceptance	687	9	91.34	.000	0.29	1.16	18.49	80.06
Need for support and desire for independence	685	9	80.68	.000	0.29	1.17	18.54	80.00
Uneven and unequal development	687	9	81.53	.000	0.29	1.16	18.78	79.77
Composite variable a	666	9	84.03	.000	0.30	1.20	17.87	80.63
Preservice experience
Physical transitions of puberty	688	9	27.75	.001	16.42	28.34	34.16	21.08
Cognitive transitions of puberty	688	9	28.08	.001	16.42	28.34	34.16	21.08
Development of abstract thinking	687	9	24.05	.004	16.45	28.38	34.06	21.11
Search for competence	687	9	22.06	.009	16.50	28.18	34.31	21.02
Development of personal identity	686	9	21.46	.011	16.47	28.43	34.26	20.85
Desire for personal autonomy	686	9	25.83	.002	16.47	28.28	34.11	21.14
Desire for belonging and peer acceptance	686	9	27.64	.001	16.47	28.13	34.26	21.14
Need for support and desire for independence	683	9	24.57	.003	16.40	28.26	34.11	21.23
Uneven and unequal development	686	9	29.50	.001	16.47	28.28	34.11	21.14
Composite variable a	665	9	20.33	.016	16.84	28.12	33.83	21.20

a

Composite additive variable created using data from nine previous variables.

Step 3: Reject Any Assertions Without Data Support/Contradiction From Both Phases

Moving along the arrow in Figure 2, I then returned to the data to reread, recode, and reprocess, in this step, specifically focusing on the pieces of data that were coded and the results of the statistical tests run (as described above). The researcher might compile all the appropriate codes and statistics in one place and examine them as a whole. From this new analysis, I discovered support for the assertion within both forms of data; however, there was a contradiction within the data that presented an interpretive puzzle for me to consider. Initial review of the composite variable comprising all nine dimensions of young adolescent development indicated that an overwhelming 81% of respondents’ confidence in young adolescent development was influenced to a great extent by their teaching experience while only 21% were influenced to a great extent by their preservice. After breaking down the composite variable into the original nine variables, the results were similar (see Table 2). In addition, all the cross-tabulations were statistically significant at the 99.99% confidence interval suggesting a potentially critical finding within a realist paradigmatic perspective that values statistical significance. Similarly, three of the narrative participants indicated that they gained most of their confidence working with young adolescents during teaching experience, not during preservice. For example, one participant, Michael said, “In undergrad . . . I think they kind of watered down what you can expect with teenagers.” He also said, “I try to follow-up with doing my own PD [professional development] and reading stuff online about what you can expect [of young adolescents] and why” (see also Table 3). However, one narrative participant disagreed. The fourth participant, Beth, completed an undergraduate degree in music therapy with a minor in psychology, rather than a bachelor’s degree in music education. She said,

I did a heavy concentration in children and adolescents [and . . . ] it was helpful for me to have that book background. And then when you get out [in the classroom] and you start seeing things [ . . . ] you can identify the commonalities and figure out what’s going on and determine what’s appropriate for that person’s development stage.

OPEN IN VIEWER

Table 3.

Narrative Participants’ Sample Responses Regarding Preservice Preparation for Teaching Young Adolescents.

Sarah	Rachel	Michael	Beth
I remember some [YA information] from basic education classes when we had to read about middle school—I think we had a special middle school class during the first six weeks; we had six weeks prior to our student teaching. I think there was a class in there about that where it was just middle school or a big section on it, and that was one of the classes that actually might have been helpful. But as far as what do middle schoolers need beyond that textbook learning is on-the-job-training!	Question: You’ve said before that you just basically had no teacher prep for middle school, right?	I think one of the big things that if you’re teaching middle school is you have to be thinking about brain development, thinking about hormones and thinking about which parts of the brain are reacting and stuff, and I try to follow-up with doing my own PD and reading stuff online about what you can expect and why, and all sorts of prefrontal cortex, getting involved the way that students at that age might perceive things differently than an adult might. They might see a statement as being really aggressive when in fact it’s really not, and so there is a lot of patience and a lot of communication that has to go into it.	Yeah, as part of my undergrad when I did the music therapy I picked up a minor in psychology. I did a heavy concentration in children and adolescents so I had Psych of the Exceptional Child; I had Abnormal Psychology, I had Behavioral Psychology
	Answer: Not for middle school general music. I mean I think now like looking back I mean undergrad was a really long time ago so I mean I remember I had a classroom instruments class that I had to take and I had a second methods class.	In undergrad—I think that my undergrad kind of—how do I put this? I think they kind of watered down what you can expect with teenagers a little bit.	I have used those classes in just about everything, every job I’ve ever held, teaching or not; those were really beneficial in being able to—it was helpful for me to have that book background and then when you get out there and you start seeing things you’re like oh so that’s how this looks and it looks different for each person or each situation, but you can identify the commonalities and figure out what’s going on and determine is that appropriate for that person’s development stage.

From an interpretivist paradigmatic perspective, Beth’s viewpoint, while potentially the minority voice in music education, is nonetheless valuable for its perspective. Within the interpretivist paradigm, a single voice is valuable and helps illuminate the nuances of a complex research phenomenon.

The researcher must interpret these new findings within the individual paradigms and connect the findings to the original assertion. There are two specific questions that the researcher should ask himself or herself at this point in the analytic process. First, the researcher ought to ask himself or herself: How should I interpret these findings within their respective paradigms? For the analysis described above, I wonder, is Beth an individual who only represents her own experiences, but does not speak for the field as a whole? This is certainly one interpretation of this finding and one appropriate to the interpretivist paradigm within which the narrative data were collected. Yet her voice is still valued within this paradigm. However, from the perspective of the realist paradigm, the 99.9% confidence interval suggests a significant finding regarding the insignificance of preservice on the preparation of music teachers for working with young adolescents. Analyzing all the existing data within their own paradigm suggests that preservice is far less influential than teaching experience on a teacher’s reported confidence in young adolescent development, but that there are some exceptions which are valuable for the individual perspective they provide. Here, the paradigmatic perspectives are dialoging with one another and set an empirical puzzle with which the researcher must contend.

Second, the researcher must also ask himself or herself: Is there additional data to support or contradict this assertion? Beth’s experience and the resulting contradictory evidence between data collected in the two phases of the study led me to wonder if there was a difference between those survey respondents who reported holding a bachelor’s degree in music education as compared with the 12% of survey respondents who reported not holding this degree. Given that these data were available in the survey (though the degrees held are unknown), independent group t-tests were run on the variables included in the analysis above (see Table 4). All tests were insignificant at the 99.5% confidence interval. Within the realist paradigmatic perspective, as I used it, these insignificant findings mean that no statistical difference exists between these two groups of respondents and thus no relationship can be claimed. A scholar working exclusively within a realist paradigmatic perspective might conclude that Beth’s experience is an anomaly or outlier within the data set, yet the data about Beth’s experience were collected within an interpretivist perspective and thus her individual perspective remains a valid piece of data to support the assertion. Having now exhausted the analytic possibilities, I moved along the arrow in Figure 2 to interpretation and dialogic writing.

OPEN IN VIEWER

Table 4.

Independent Group t-Tests for Bachelors’ Degree in Music Education and Other Variables Under Consideration. ^a

Item	Bachelors’ degree	n	M	SD	95% CI		t	df	Pr(|T| > |t|)	Pr(T < t)
Influence of teaching experience	N	85	3.79	0.44	3.70	3.89	0.1538	111.748	.8781	.5610
	Y	601	3.79	0.46	3.74	3.82
Influence of preservice experience	N	84	2.40	1.11	2.16	2.65	−1.7507	101.721	.0830	.0415
	Y	602	2.63	0.98	2.55	2.71
Physical transitions of puberty	N	85	3.40	0.62	3.27	2.53	0.3553	109.974	.7230	.6385
	Y	601	3.37	0.63	3.32	3.42
Cognitive transitions of puberty	N	85	3.15	0.63	3.02	3.29	−0.6785	115.875	.4988	.2494
	Y	601	3.20	0.70	3.15	3.26
Development of abstract thinking	N	86	3.06	0.73	2.90	3.21	−0.3607	111.242	.7190	.3595
	Y	600	3.09	0.73	3.03	3.15
Search for competence	N	85	3.20	0.72	3.04	3.36	0.1932	107.206	.8472	.5764
	Y	598	3.18	0.69	3.13	3.24
Development of personal identity	N	85	3.30	0.74	3.14	3.45	0.3772	106.37	.7068	.6466
	Y	599	3.26	0.70	3.21	3.32
Desire for personal autonomy	N	85	3.34	0.59	3.21	3.47	0.3108	121.207	.7565	.6217
	Y	598	3.32	0.71	3.26	3.38
Desire for belonging and peer acceptance	N	83	3.66	0.48	3.56	3.77	0.6632	114.105	.5086	.7457
	Y	600	3.625	0.55	3.58	3.67
Need for support and desire for independence	N	85	3.47	0.57	3.35	3.59	−0.8982	112.133	.3710	.1855
	Y	596	3.53	0.60	3.48	3.58
Uneven and unequal development	N	85	3.41	0.68	3.27	3.56	−0.1139	108.766	.9095	.4548
	Y	599	3.42	0.67	2.37	3.47

Note. SD = standard deviation; CI = confidence interval; df = degree of freedom.

a

Unequal variances, Satterthwaite’s approximation calculated.

Step 4: Generate Dialogic Writing

Interpreting both confirmatory and contradictory evidence for an assertion is one of the major challenges of the dialectic stance. The scholar must strive to write up a dialogic discussion of the findings specifically referencing the original assertion and presenting the consonant and dissonant evidence for the reader to consider. At this point the dialogue moves from within the researcher’s head to a dialogue between the researcher and the reader. The dialectic scholar must be sure to give equivalent weight to the findings from both portions of the study in the final written report.

Although sometimes the qualitative and quantitative data will corroborate one another, contradictory evidence is equally valid evidence within the dialectic stance (Greene, 2007, 2015). Assertions supported with both convergent and divergent data often raise tensions that are foundational for the particular discipline within which the study was conducted. These tensions can lead to further questions rather than conclusions. For dialectic researchers committed to remaining open to the possibility of confirmatory and contradictory findings, these questions should be exalted rather than ignored, but researchers should also be prepared for these questions to evoke debate within the “home” discipline in which the study was conducted.

For my example analysis, it seemed that the most important thing to put forward from this particular assertion was that preservice teacher education provides most middle school music teachers with very little confidence in working with young adolescents whereas the hands-on experience of holding an actual middle school teaching position provides much greater confidence. This suggests that perhaps hands-on experience working with middle school students during preservice might improve this situation. However, Beth’s experience also suggests that learning about child and adolescent development in depth, as was done in her music therapy undergraduate, might provide an additional foundation on which to base one’s teaching practice. Perhaps music education would benefit from drawing upon some of the content present in music therapy programs. The tension between preservice education and on-the-job experience as an educator is a long-standing challenge in teacher education, not just within the discipline of music. An undergraduate education designed to prepare a student for teacher licensure is limited to 4 years of curricular work and by the maturity of undergraduate students. In any preservice program, topics, skills, and competencies must be selected, often at the neglect of others. Most university programs in music education prepare students for a certification that enables them to teach music to students from Kindergarten through 12th grade (Henry, 2005) and as such the developmental characteristics of young adolescents (or middle school students) is only one tiny component of the total undergraduate education. As a scholar, I can make some preliminary suggestions, carefully couched in the limited findings related to this assertion, for improvements to music educators’ undergraduate preparation as I write up my dialogic discussion, but I must be continually aware of each perspective provided by the phases of the study as well as the limits each paradigmatic perspective affords my claims.