-July 11,2009-



Relatedness Evaluation Qualitative Analysis

Toshiharu Kasai
(Until 2020, March:
The Department of Clinical Psychology, The Faculty of Psychology,
Sapporo Gakuin University,Sapporo, Japan)

[Japanese site]

WHY KH-METHOD (Kanrensei Hyoutei:"Relatedness Evaluation") ?

WHY KJ (Kawakita Jiro) Method ?

In 1960s, Dr. Jiro Kawakita, a Japanese anthropologist, started a very much renowned KJ-Method, a method for integrating a variety of descriptive data obtained by observations, conversations with the local people, or by reading reports or documents. The KJ-Method yields a spatial arrangement consisting of all the descriptive data cards that include all materials available. He wrote that he started to create a spatial representation of the available data in his tent in order to find out any hidden patterns of the culture he was studying.
When I first encountered several charts made by Grounded Theory Analysis (GTA) or Interpretative Phenomenological Analysis, I simply thought they must have been obtained by KJ-Method with some modifications because they were very similar to the spatial arrangement by KJ-Method. (I spent nearly one month to study KJ-Method while lodging in a tent at a research venue when I was an undergraduate student.)
After studying GTA and IPA to some extent, I found them rather unsatisfactory to me because they tend to disregard some of their data cards, or, in other words, they tend to skim only some of their "important, essential" cards in order to confirm, for example, a researcher's presupposed hypothesis. In the anthropological study, Kawakita wrote, our way of thinking or understanding can be a major obstacle when trying to grasp the real meaning of what the people of the targeted culture do in their seemingly odd way.
It reminds me a example of folk taxonomy of a culture in the southeast Asian island: They use "humidity" as the color concept. Usually, humidity can not be a primary concept to talk about the color, but because they use the concept of humidity to define the color, we have to imagine that it must be very important for them to tell the color differences, say, between two kinds of grasses in terms of humidity: One has "humid color" and it is edible, but the other is poisonous because it is "non humid color"...
My question is : While both GTA and IPA are trying to catch precisely how the targeted people see the world, why can you drop or disregards some of their case-specific descriptive cards by skimming only some cards with general contents ? Although I could understand its meaning from a theoretical point of view, I wanted to avoid it because of some other theoretical reasons.

* The theory of chaos shows that seemingly trifling things or insignificant differences determine what follows as in the butterfly effect: How a butterfly flutters and flies in China dictates how a hurricane attacks North American Continent. Also, in a small sample situation where the law of great numbers is not necessarily suitable, seemingly unessential factors or exceptional conditions (descriptions) in a small sample case could become significant when understanding the very case and its peculiarities in detail. In the qualitative research, we basically deal with small samples, and have to keep attention to isolated and outwardly inessential descriptions.

In KJ-Method, lonely cards that don't belong to any of the category groups are left alone, and never dropped. They are called "a lone wolf" or "a free ape", and each one makes its own category.
I think we need to appreciate a surprisal that sometimes visits us when our pet theory or shallow hypothesis about the people with different backgrounds turned out to be wrong or not suitable. Technically speaking, this is concerned with how you avoid an invisible trap called "projective understanding" that projects your way of thinking onto the target and then recursively finds that it should be the truth. I like the KJ-Method because it keeps all the descriptions intact in the spatial arrangement, and tries to find out any patterns in those descriptive cards. Although it would not be easy to separate a) understanding the targeted people or event precisely from b) interpreting the target favorably, I would like to stick to the former attitude.

*Dr. Jiro KAWAKITA passed away on July 8th, 2009 in Tokyo, 89 years old. One of his famous studies as an eminent anthropologist was about the bird using funeral in Nepal (letting wild birds eat the corpse for lifting its soul to the heaven).

WHY Hayashi's Quantification Theory Type 3 ?

In 1950s, Dr. Chikio Hayashi, the ex-director of the Japan National Institute of Statistical Mathematics, pioneered a new analyzing method called "Quantification Theory", more than 10 years before the Correspondence Analysis by a French researcher. (The Correspondence Analysis is theoretically similar to Hayashi's Quantification Theory Type 3. Hayashi's Type 1 is the qualitative Multiple Regression analysis. Type 2 is the qualitative Discriminant Analysis. Type 4 is the qualitative Multi Dimensional Scaling.)
His theory was proposed, in a sense, too early to become popular in psychology or other social sciences: There was no need for such a qualitative approach because the psychology at those days, for example, was solely interested in statistical approach and statistical testing. Qualitative researches were not considered scientific, and not accepted in major academic journals. Hayashi's Quantification Theory interested only statistical mathematicians…
However, the research trend started changing in 1980s in psychology and other social sciences, and the qualitative research once again revived long after the curse of behaviorism.
* The software for Hayashi's Quantification Theory has been developed: Add-in software for SPSS and Microsoft EXCEL are available.

Quantification Theory Type 3, hereafter QT-Type3, gives you mathematical results similar to those given by Factor Analysis, that is, several spatial axes and mathematical spatial representation of data. When you have (descriptive) data cards and obtain several categories that include some data cards in each category or other categories that include some categories in each of them, you can make a data table of cards and categories. The table in which all cards are lined in row and all categories are lined in column is the input data for QT-Type3. As you can find in this website, the table consists of {1 or 0}: {1} means that the card belongs to the category in the column, and {0} means that the card is not included in the category of the column. This table is called "Corresponding Table of cards and categories". QT-Type3 analyzes this table and yields multidimensional axes for both categories and cards.
The task that you create categories is an interpretational and qualitative work, and the resultant correspondent table of cards and categories gives rise to mathematical axial structure among the categories and, at the same time, among the data cards. QT-Type3 can be said to be a connector or catalyzer that combines the qualitative data and the mathematical axial structure.


WHY KH-METHOD NOW ?

Relatedness Evaluation Qualitative Analysis, "KH-Method" in short from Japanese "Kanrensei Hyoutei (Relatedness Evaluation)", is a methodology that makes this qualitative - quantitative connection possible by introducing a mathematical concept "lattice" (semi-order structure) onto the spatial arrangement of cards and categories.
In Japan, as in United States of America, because the trend for Evidence Based Medicine has been very strong, and so has been in the psychology, purely qualitative approach as GTA and IPA is not fully acknowledged as scientific approach even nowadays by major academic journals. We had been in strong need for any method that can combine qualitative and quantitative approach, and Relatedness Evaluation Qualitative Analysis is one of the most effective solutions that satisfies this need.

(added. Sep.10,2008)

Journal of the Society of Humanities - Sapporo Gakuin University, No.83,61-100, 2008

Relatedness Evaluation Qualitative Analysis for Transcript Study- Its basic ideas and practice(JAPANESE)
[English abstract]


Researches with KH Method in psychology and the science of nursing
(KH Method stands for Relatedness Evaluation in Japanese as Kanrensei Hyoutei)
    Papers for conventions/ journals
  • "A study of regional differences of Baum Test and S-HTP method: Comparison of Hokkaido and Okinawa students and some influences of length of living" T.Sano and A.Urata
    The 37th Annual Convention of The Japanese Society of Psychopathology of Expression and Arts Therapy, 2007
    [KH Method and Quantification Theory Type 3]

  • "A case study of a female undergoing dialysis: by Utilizing KH Method" R.Nihonyanagi
    Japanese Society of Nursing Research Hokkaido Convention, 2008
    [KH Method and Quantification Theory Type 3 with 1 interviewee]

  • "The suicidal mentality of husbands caring for their wives and husbands' attitudes towards their caregiving life" E.Uehira, K.Saeki and Y.Kimura
    The 18th Annual Convention of The Japan Academy of Psychiatric and Mental Health Nursing, 2008
    [KH Method and Quantification Theory Type 3 with 4 interviewees]

  • "A study of impression of collaged pieces and production reports" T.Sano
    The 27th Annual Convention ofThe Japanese Association of Clinical Psychology, 2008
    [KH Method and Quantification Theory Type 1]


    • Master's papers:
  • "The relationship between self-control behaviors and problem-solving skills in Type 2 diabetics" F.Ohtori
    Master's thesis for School of Nursing, Osaka Prefecture University, 2007
    [KH Method and Quantification Theory Type1 and 3 with 105 subjects]

  • "A Study of grief of child-lost mothers: Through interviews with mothers who lost only child in the middle age" M.Hatano
    Mater's thesis for Department of Clinical Psychology, Sapporo Gakuin University, 2007
    [KH Method and Quantification Theory Type 3 with 3 interviewees]

  • "A study of determining factors of college students' helping behavior: Qualitative analysis by KH method and numerical analyses about multiple dimentional empathy scaling" Y.Kawamura
    Mater's thesis for Department of Clinical Psychology, Sapporo Gakuin University, 2008
    [KH Method and Quantification Theory Type 1 and 3 with 100 subjects]

  • "A basic research of anxieties and shyness for sexuality: Through questionaire study for college students and interview study for peer-education instructors" M.Takeuchi
    Mater's thesis for Department of Clinical Psychology, Sapporo Gakuin University, 2008
    [KH Method and Quantification Theory Type 3 with 3 inteviewees]

  • "A transcript study of experiences and experiencing process of novice therapists: Through associating experiences of the first interview of clients”Chiyomi Shirogane
    Sapporo Gakuin University, Graduate School of Clinical Psychology, Jan.2009
    [KH Method and Quantification Theory Type 3 with 5 interviewees]

  • "Analysis of the current relationship structures of teacher, school nurse, and school clinical counselor in school counseling activities”Akane Tonsho
    Sapporo Gakuin University, Graduate School of Clinical Psychology, Jan. 2009
    [KH Method and Quantification Theory Type 3 with 66 cases of 12 interviwees]


During 2006-2007, I tried to apply this appoach for several master papers and found it very effective in grasping factors in transcripts by giving a concrete procedure to integrate several transcripts among subjects and/or researchers. Also, by utilizing Hayashi's Quantification Theory (Type 1), it was rather a surpirising thing to see that this appoarch enabled us to find out the relationship between particular statements of subjects and some physiological indeces such as cholesterol level, blood pressure, HbA1c, non-HDL, etc. (which I am not familiar though) in an nursing study about coping behavior of diabetic patients. (Hayashi's type 1 is known as a qulitative multiple regression analysis.)

I am now interested in how to integrate this approach with Interpretative Phenomenological Analysis (IPA) or how to combine two approaches (including judgement abou how much and in which part it is possible), and would like to ask for cooperation of IPA researchers in order to promote qualitative study and get more fruitful understanding about what is really going on in the concerned research field.
I am ready to do a joint research by employing Relatedness Evaluation Qualitative Analysis based upon the IPA standpoint, and develop this approach more for tackling on Evidence Based Medicine/Psychology.
(Above was written for IPA mailing server. Jan.10, 2008)



(The followings are cited from email sent to the mailing list 
of Interpretative Phenomenological Analysis, Oct. 7, 2007)

...
In order to grasp a single case to ,at most, a dozen cases, I 
have developed a new approach that enable us to grasp contents 
of transcript/s.

I combined three Japanese approaches that have not been well 
known outside Japan, and have made a qualitative analysis named 
Relatedness Evaluation Qualitative Analysis.
(I call "Kanrensei" "Hyoutei" method, KH method in Japanese.)

Three components are as follows:

1)KJ method: 

 Kawakita, Jiro started this method to summarize and get an idea
 about unknown cultures as a cultural anthropologist in 1960s; 
Making cards from his field note of his observations and 
obtaining card-groups and a diagram of the contents almost as in
 GTA and IPA.

By placing several limitations on KJ method and you can obtain "
lattice structure" in the diagram that can be analyzed by the 
following two analyses.

2)Hayashi's Quantification Theory:

 Chikio Hayashi developed 5 different types of Quantificaiton 
method for qualitative data in 1970s. The outcome by the type 3 
analysis of this analysis is much like Factor Analysis, yielding
 several dimensional structures represented by numerals.
(The software is available for SPSS, and Microsoft Excel.)

3)Nagata's Formal Concept Analysis:

 Hiroyasu Nagata, who happens to be a staff of my college, has 
developed a qualitative analysis called Formal Concept Analysis, 
and its software.
The lattice structure can be analyzed and you can get a diagram 
from a different angle.
(This is programmed by JAVA, and is available upon request for 
academic studies.) 

      *Nagata's Formal Concept Analysis.
      (in Japanese with English abstract)

Related Evaluation Qualitative Analysis modifies the KJ method 
and makes above two (mathematical/logical) analyses applicable. 
This method summarizes and visualizes verbal qualitative 
materials by using above three tools.

An example of Relatedness Evaluation Qualitative Analysis.
The theme is "what your impression about qualitative approach is".

For example, eight comments were obtained...

  1. Qualitative analysis is difficult.
  2. It is not easy.
  3. I can't follow...
  4. I wish I could use.
  5. Is it useful indeed?
  6. Is it effective?
  7. Because it is significant, it has developed...
  8. Should I learn it?
Eight cards(descriptions) come closer one another based on the researcher's subjectively perceived relatedness, and card groups are made and given a label when the contents of the cards of each group are judged very related one another. This process is derived from KJ method.
The constellation of 8 cards may be as follows with 6 labels:




The correspondent table of 8 cards and 6 lables.
(L11,L12... stand for labels. Card-4 alone makes a category:L32)



The correspondent table between 8 cards and 6 lables is analyzed by Hayashi's Quantification Theory Type 3, yielding dimensional representation as follows:

*Your task is to give a name to each dimension as in Factor Analysis.
(Usually 4-5 dimensions are shown in Hayashi's Quantification program as add-in software for SPSS).
This is your "interpretation" of the qualitative data, but based on this mathematical analysis. If your categorization is not good enough or not following the ideas of KJ method, interpetation of each dimension tend to become difficult. You could refine your categorization process by checking how easy or uneasy to interpret each dimension. This reciprocal process can be used as another form of "triangulation".

*Hayashi's Quantification Theory Type 3 has the mathematically identical structure with Correspondence Analysis (eigen value calculation).

Nagata's Formal Concept Analysis gives a figure called Hasse diagram.


The percent numbers show how many cards are included in each label.


  • These graphs, obtained by Hayashi's Quantification Theory Type 3 and Nagata's Formal Concept Analysis, are not a statistical but mathematical/logical representation of the correspondent table. These methods can be used for any data consisting of member-category lattice sturucture, for example, when your final outcome obtained by IPA or GTA take the correspondent table format.

  • If you obtain both a) descriptive replies about research themes from many subjects and b) subjects' external data such as age, sex, psychological and/or physiological/medical personal traits, etc., you can make a quality-quantity integrated research by utilizing the categorical multiregression analysis of Hayashi's Quantification Theory Type I, or the categorical discriminant analysis of Hayashi's Theory Type II seamlessly in the framework of Relatedness Evaluation Qualitative Analysis.
    A integrated study utilizing Type I in the science of nursing (non-compliance behavior of diabetics) has been reported and highy appreciated by those reaserchers. (2008)

Papers about psychological methodology by T. Kasai

Relatedness Evaluation Qualitative Analysis Website
All Rithgts Reserved. (C)Toshiharu Kasai, 2007-2020