Qualitative Data

Thematizing and three coding types (open, axial, and selective)

• Thematizing refers to the analysis of interview responses.

• To analyze the interview responses, data is analyzed for the presence of themes.  Interview responses can be examined multiple times on an overall basis OR  on responses per interview question to identify the themes.

• Typically, if agreement, or a repeated comment(s), is found among three responses or more, it is considered a theme. 

• Three types of coding are open, axial, and selective.

• Open coding refers to revealing X set of themes and what those themes are related to and/or can be considered by.  Open coding responses can be presented in a table by theme, by the phrases/descriptive words that help illustrate the theme, and by the participant id that the phrases/descriptive words come from.

• For example: An interview was conducted on 12 middle school teachers, regarding new school grading policies.  Open coding revealed two sets of themes, negative reaction and uncertainty, that related to interview question 4 “How are your students reacting towards the implementation of the new grading policies?”  Three participants felt that the implementation of the new school grading policies causes a negative reaction from students, such as tantrums and violence, and five participants felt that the implementation of the new school grading policies causes uncertainty, such as crying and screaming.  See the table below for details.

Interview Question 4 Open Coding Responses

Participant	Theme	Phrases/descriptive words
	Negative reaction
3		He [the student] started hitting his desk and threatening to just walk out of the classroom.
4		They [the students] just started yelling at me and telling me that I better not or else.
12		Papers and pencils were thrown everywhere.
	Uncertainty
1		I saw the look on their faces…as if I was punishing them and they didn’t understand why.
4		One student was quiet at first, then started to sob uncontrollably.
7		…just kept asking me why over and over again.
11		Received a call from a parent telling me that their child couldn’t describe how school was that day.
12		I had to repeat the instruction multiple times throughout the day.  The kids just didn’t get it.  One kid began screaming, which got a few others at it too.

• Axial coding refers to revealing the thematic relationships in order from strongest to weak associations, and also presenting examples.

• For example: (using the same example stated previously)  Axial coding for responses on interview question 4 revealed the following thematic relationship in order from strongest to weak associations:

1.      Five of the 12 participants (42%) specified lack of comprehension and confused reactions (e.g., just didn’t get it, couldn’t describe how school was, kept asking me why, didn’t understand why, quiet at first then started to sob).

2.      Three of the 12 participants (25%) specified threats and physical reactions (e.g., tell me I better not, started hitting, papers and pencils thrown).

•  Selective coding refers to the topic of the research question and an excerpt from a participant that best mentions/illustrates the themes mentioned. 

• For example: (using the same example stated previously) Selective coding of interview question 4, student reaction toward policy implementation, generated two themes: negative reaction and uncertainty.  The researcher selected one participant’s statement that reflected the axial coding with the most congruence.  Participant 4 stated, “I really thought it wouldn’t be too difficult.  I feel like my students are fast learners, but I was wrong.  I guess they are just too young to understand why and what was happening.  On the first week, I had several mishaps.  One student was quiet at first, then started to sob uncontrollably.  My kids weren’t taking it too well.  They just started yelling at me and telling me that I better not or else.  After a couple more days, I just didn’t know what to do.”

Types of Qualitative research designs

Grounded theory, ethnographic, narrative research, historical, case studies, and phenomenology are several types of qualitative research designs.  The proceeding paragraphs give a brief over view several of these qualitative methods.  

Grounded theory is a systematic procedure of data analysis, typically associated with qualitative research, that allows researchers to develop a theory that explains a specific phenomenon.  Grounded theory was developed by Glaser and Strauss and is used to conceptualize phenomenon using research; grounded theory is not seen as a descriptive method and originates from sociology.  The unit of analysis in grounded theory is a specific phenomenon or incident, not individual behaviors.   The primary data collection method is through interviews of approximately 20 - 30 participants or until data achieves saturation.  

Ethnographic studies are qualitative procedures utilized to describe, analyze and interpret a culture’s characteristics.  Ethnography was developed in the 19^th and 20^th centuries and used by anthropologists to explore primitive cultures different from their own; it originated from Anthropology.  Ethnography is used when a researcher wants to study a group of people to gain a larger understanding of their lives or specific aspects of their lives.  The primary data collection method is through observation over an extended period of time.  It would also be appropriate to interview others who have studied the same cultures.

Phenomenology is used to identify phenomena and focus on subjective experiences and understanding the structure of those lived experiences.  It was founded in the early 20^th century by Edmund Husserl and Martin Heideggar and originated from philosophy.  Phenomenology is used to describe, in depth, the common characteristics of the phenomena that has occurred.   The primary data collection method is through in-depth interviews.  

Case studies are believed to have originated in 1829 by Frederic Le Play.  Case studies are rooted in several disciplines, including science, education, medicine, and law.  Case studies are to be used when (1) the researcher wants to focus on how and why, (2) the behavior is to be observed, not manipulated, (3) to further understand a given phenomenon, and (4) if the boundaries between the context and phenomena are not clear.  Multiple methods can be used to gather data, including interviews, observation, and historical documentation. 

Attribute

An attribute is generally a quality or characteristic that an individual possess. Some of the common attributes that an individual possesses are smoking, drinking, blindness, intelligence, etc. In or to determine attributes in statistics, the theory of attributes is used. Here, quality is emphasized over quantity. Hence, attribute cannot be measured or calculated, but it requires a completely different mode and method. A distinctive statistical treatment is required to compute attribute.

Statistics Solutions is the country's leader in qualitative analysis and dissertation consulting. Contact Statistics Solutions today for a free 30-minute consultation.

Attributes are detected by their presence or absence in a particular individual. Statistics is therefore widely used to determine attributes through the various numbers of techniques available. These statistical methods are used in the study of variables and are used in a wide scope in the theory of attribute. The theory of attribute has certain representations. A population is divided into two classes (i.e. negative and positive). This is done in accordance to the presence or the absence of attribute. The positive class indicating the absence of the attribute is represented by the roman letters A, B, C etc., while the negative class is indicated by Greek letters - α, β etc.

Any combination between two attributes is denoted by assembling the two attributes. For example, if the combination is A and B, then the letters would be denoted as AB. If the population is divided into two subclasses, then it is called dichotomous classification. Observations that have been allocated to the attributes are termed as class frequencies, which are denoted by bracketing the attribute symbols. For instance, (A) stands for the frequency of the attribute A. A class characterized by ‘n’ attribute is called a class of nth order and the matching frequency of that attribute is the frequency of the nth order. An example, (A) is a class frequency of the first order. These class symbols of the attribute also work as an operator. For example, A.N=(A) implies that the process of dichotomizing N in accordance to the attribute A gives the class frequency equal to (A). Attributes A and B are said to be independent only if there is no connection between the two. When independent, the same proportion of A attribute occurs amongst B attribute and amongst the β attribute, and the proportion of B attribute amongst A attribute is same as that amongst the α attribute. The attributes A and B may be considered to be linked if they are not independent but are related in some way. For positive association between two attributes, (AB) > (A) (B) / N and are negatively connected if (AB) < (A) (B) / N. If attribute A cannot occur with attribute B, but attribute B can occur with attribute, A (or vice versa), then the two attributes are actually connected. Again, two attributes are said to be connected if the two occur together in a number of cases. The reliability between the two attributes (A)=20 and (AB)=25 is not present as the attribute (AB) cannot be greater than the attribute (A) if they have been studied from the same population.

The consistency between the two attributes (A)=20 and (AB)=25 is not present as the attribute (AB) cannot be greater than the attribute (A) if they have been observed from the same population.