THE TOPIC:
My goal to identify specific character traits related to the occupation that may contribute to the underrepresentation of women in law enforcement is particularly relevant, given the slow growth of female officers in this field. However, my plan to use previous studies on police personality and current issues affecting law enforcement
Chapter 3 provides you multiple examples of types of quantitative designs for studies. Based on your work thus far which of these studies do you believe applies best to your study? If your study does not lend itself to a quantitative study then read ahead to Chapter 5 and see if you can land on a qualitative design that would work. Both of these categories of studies are appropriate. The key is to chose based on what method is best to explore your topic.
Quantitative Research Designs
In this chapter, we share with you an important component of the dissertation process—determining the type of inquiry and research design you will use for your study. This actually will not be a specific chapter in your dissertation, but is all important in developing your proposal and carrying out your research. The inquiry techniques and/or methods presented in this chapter all have their beginnings in basic human observation and curiosity. We are describing science in the broadest sense of the word—a way of reflecting on our world. Just as children experience science via attitudes, processes, and products, we do also as adult researchers.
Your attitude as a researcher is critical. First, you must think of yourself as a researcher and writer, and not just as a graduate or doctoral student. Your attitude will carry you far as a budding scientist. It will encourage you further in your own curiosity of your topic and of others’ topics; it will provide you with perseverance for the task of conducting the research; it will pick you up when you fail and help you learn from your mistakes; and it will aid your open-mindedness and assist you in cooperation with others. Furthermore, a positive attitude toward the research will provide you with a desire to seek reliable and valid sources of information; a desire to provide and to tolerate alternative viewpoints; an avoidance of overgeneralizations; a restraint to make a judgment until all evidence is examined or evaluated, or to make claims without having proof or descriptors; and an open mind toward questions related to your own research.
Processes of research will aid you in working through your study in a critical and creative way. In the simplest terms, processes may include observing, classifying, contrasting, communicating, measuring, estimating, predicting, and inferring. You also will use the processes of identifying and controlling variables, operationalizing definitions, hypothesizing, questioning, experimenting, investigating, interpreting data, or forming theories or models.
The product of your research, your dissertation, provides your chosen field with a greater knowledge base; therefore, because knowledge is considered power, you also carry with you during your research much responsibility as you plan your study, choose your method of inquiry, and conduct your research. You have an ethical obligation to do the very best research that can be produced. You ask: even at the dissertation stage? The answer is yes, at the dissertation stage of your research career. Though you may be questioned on your study, you also know that much of science changes over time and that knowledge is challenged as it is produced. Remember this about your research, as Slavin (1992) said, “the best research design is one that will add to knowledge, no matter what the results are” (p. 3). Your research product, whether you find a small or large effect size along with significance, may be interpreted or used differently by different audiences, depending on their circumstances and experiences.
Your dissertation, as a product, may take the form of a hierarchy, such as basic factual or uniconceptual research, principles of research that relate to multiple concepts, or theories, the highest level of research. This hierarchy can be observed in Figure 3.1. The closer the products of the research are to the top point of the pyramid, the more complex the study. In the development of theory, we note that the complexity may involve dual methodologies, both quantitative and qualitative.
A couple of important questions should be considered as you determine your method of inquiry. First, you must ask yourself: what is my intent or purpose of the research? Second: what are my research questions? These initial components will drive the method you select. You may determine that a quantitative analysis will suffice in answering your research question or will respond to your purpose. On the other hand, you may conclude that it is qualitative, the deeper understanding of the topic, that responds to your purpose, or ultimately, you may decide that a mixed methods approach, using both quantitative and qualitative analyses, is the best method of inquiry for your dissertation.
In this chapter, we provide you with an overview of each of the data analysis techniques or inquiry methods you could use if your questions or hypotheses necessitated a quantitative approach. Certainly, this chapter is not all inclusive, and we know that you will want to “dig deeper” on your own once you settle on a specific method of inquiry. For example, you will need to consult statistics and research methods textbooks and review your proposed method with your dissertation advisor. You must be thoroughly familiar with your method of inquiry and the assumptions for the statistical procedures.
Your research design or method of inquiry for quantitative research will fall into one of four categories of research (a) descriptive research, (b) correlational research, (c) causal-comparative research, and (d) quasi-experimental and experimental research. In addition to providing you with a brief overview of each category, we also provide examples of research purposes or questions that would justify a specific type of quantitative research design or method of inquiry.
DESCRIPTIVE RESEARCH DESIGN
Descriptive research is one of the most basic forms of research. It lies at the bottom of the hierarchy as depicted in Figure 3.1. Even though descriptive research is at the base of the pyramid, it does not mean it is unimportant, unscientific, or unworthy. This type of research involves the description of phenomena in our world. In this type of inquiry, the phenomena described are basic information, actions, behaviors, and changes of phenomena, but always the description is about what the phenomena look like from the perspective of the researcher or the participants in the research; it is not about how the phenomena function. Prior to beginning the research, you will need to have established from theory or from prior research what it is about the phenomenon you will study; from this perspective, descriptive research is theory or research driven.
Descriptive research studies are important to the public and to educators. For example, many reports produced by the federal government are good examples of necessary descriptive research (the U.S. Census Report or the many reports from the National Center for Education Statistics). Such reports provide the impetus for many other research studies.
Descriptive research tends to answer basic informational questions as indicated in the following examples. What leadership behaviors do superintendents who serve rural, suburban, and urban school districts exhibit? What pedagogies can be observed in teachers who teach in bilingual education classrooms? What types of technology are implemented by teachers who have had technology classes at the university level? Which leadership and/or organizational textbooks used in educational or business administration courses include gender-inclusive leadership theory? Such questions in descriptive research will generally ask the question of who, what, when, where, how, or which.
When conclusions are drawn from descriptive research, there is a reporting of the facts, but there should also be researcher-based conclusions connecting the data to theory or prior research.
… illustrating the importance of how a phenomenon is conceptualized in descriptive research, Thurstone, Guilford, and others promoted a different conception of intelligence that emphasized instead the specific nature of human abilities. Thurstone, using a somewhat different mathematical model of factor analysis than had the previous researchers emphasizing general intelligence (Gould, 1981), found, instead of general intelligence, several distinct dimensions of intelligence that he termed “primary mental abilities.” Subsequently, Guilford developed and described a complex “structure of intellect” model that first posited 120 separate factors, theoretically derived (Guilford,1967), and later 150 (Guilford, 1977). Notice that the strategy was theory driven: Guilford’s complex conceptual model of intelligence, for example, largely based on logically derived theory, provided a basis for designating the specific tasks he then used to measure each specific dimension of intelligence as described in his model. (Anastas, 1999, pp. 128–129)
Guilford used prior research and theory to build his structure of the intellect model that has been used many times in developing curriculum and in research related to gifted children.
Instrumentation
Instrumentation is critical to descriptive research. If you are using archived data, you will likely use an instrument that measures achievement among students or classrooms or schools. Perhaps you are asking the question: What is the achievement among various groups of students based in the state academic achievement test? You define the various groups of students as ethnicities, gender, and special-needs students. You would, of course, need to describe the instrument in detail, along with its norm reports on validity and reliability, particularly as it relates to the various groups of students. Because the research question is based in a standardized achievement test, the results of the study will only be as good as the instrument. Other standardized instruments may include personality inventories, intelligence measures, or attitudinal scales.
In most descriptive research studies, instruments must be developed by the researcher due to the fact that the study is related to a specific phenomenon. You will need to pay special attention to your instrument development and describe all the specifics of (a) how you developed your instrument, (b) where you obtained information to include in your instrument, (c) how you ensured validity and reliability, and (d) if others assisted you in your development. Surveys are some of the most common instruments you will use in descriptive research studies. In fact, sometimes descriptive research is even called survey research.
Data
Descriptive research reports data as measures of central tendency, which include mean, median, and mode, and as measures of dispersion, which include deviance from the mean, variation, range, and quartile. It is usually conducted as indicated previously through surveys, or it may be conducted via observation, interviews, portfolios, or cases. Polls, surveys, or questionnaires are typical examples of descriptive research studies. However, antiquities that are archived, film, video, Internet, and e-mail, may also be used for gathering data for descriptive research studies.
Longitudinal Studies
In descriptive studies, there may be more than one variable, but usually descriptive research includes only one variable. This variable may be measured at one point in time, but it also may be measured across time. The latter is considered as a longitudinal descriptive study, or it could be noted as a descriptive trend study. An example of such descriptive studies would be the reports on the National Assessment of Educational Progress (NAEP) or the Longitudinal Early Childhood Studies (for U.S. Department of Education reports see http://www.ed.gov). In longitudinal descriptive studies, data are taken from the same group or sample over varying points in time. In a trend descriptive study, data are gathered from a different group at various points in time, but within a population that changes as well. The NAEP study is an example of this type because students are assessed every year on reading and mathematics in specific grade levels. In other words, the population for all fourth grade students changes from year to year. Trends are established on a measure over time with changing populations.
Another type of descriptive study is a descriptive cluster study. In the cluster study, data are taken on measures from samples within the same population at various points in time. For example, all children beginning kindergarten in 2006 in a large urban school district or in a state would be considered the population. Those children would be followed throughout their school career; however, each year or at each point when data are taken, a cluster would be sampled within the same population of students who began their school careers in 2006.
Cross-Sectional Studies
A simulated longitudinal descriptive study can be completed by collecting data obtained at a point in time from samples of individuals who represent a cross section of the population. This is typically called a cross-sectional descriptive study. You may survey parents of students at grades K through 5 in a large urban district on their perceptions of teacher attitudes toward their children or on some variable of interest. You may administer a test to a sample of students from grades K through 12 at a point in time to gather data on a particular phenomenon. Mobility of students and attrition are not accounted for in this type of research and are considered limitations.
Several dissertation abstract examples follow that demonstrate effective use of descriptive research design. (See Examples 3.1, 3.2, and 3.3.) Following is an example of a dissertation dealing with a training program for school psychologists completed by Bridgewater (2006). (See Example 3.1.)
Example 3.1
Abstract
Despite the need for early identification and intensive intervention for social emotional concerns in young children, there exists a shortage of personnel trained to provide mental health services with early childhood populations (Klien&Gilkerson, 2000). The purpose of this study was to examine, through descriptive research, the current status of preservice school psychology training in relation to the assessment of social-emotional development in preschool children. Specifically, this study sought to explore the training school psychology programs offer, in preparing pre-service school psychologists to provide mental health services for preschool populations, and evaluate the perceptions of trainers and students regarding the coursework, competencies, and instruments necessary for preparing school psychologists to provide social-emotional assessment and mental health services in early childhood settings. Participants in this study included 108 program directors (trainers) of school psychology training programs within the United States, as well as 4 advanced level students from each program, randomly selected from the master’s/specialists (n = 151) and doctoral level (n = 74) of study in each program; a total of 225 students. Two corresponding forms of the “Preschool Social-Emotional Assessment Training” questionnaire were developed for this study. Each questionnaire was worded appropriately for the different groups of respondents. The questionnaire items were developed based upon two previous studies conducted by Boyer (1996) and Gettinger, Stoiber, Goetz, and Caspe (1999), and the current literature concerning the preparation of school psychologists and related service personnel in the delivery of early childhood metal health assessment and intervention services. (Bridgewater, 2006, p. ii)
The second abstract example is a dissertation that reports parental regulations on television viewing (Salvato, 2006). (See Example 3.2.)
Example 3.2
Abstract
Childhood obesity is the most prevalent chronic disease among North American children and has reached epidemic proportions. Increased television viewing has been shown to increase the chances of a child becoming obese. The purpose of this descriptive study was to determine the amount of media children are using and the parental practices being used to regulate television viewing among elementary school students. The sampling frame consisted of parents of children attending an elementary school in Connecticut. A questionnaire was adapted to determine the parental regulations being used to monitor children’s television viewing. Parents self-reported their child’s height and weight so that a correlation could be made between regulations and childhood obesity. The results of this study have increased the data on parental regulations regarding the use of television, and revealed that many parents are not aware of the effects that excessive television use can have on their child’s BMI. (Salvato, 2006, p. iii)
The third dissertation example relates to a higher education study on writing centers and the questions that are asked by the tutors. This dissertation was completed by Cook (2006). (See Example 3.3.)
Abstract
This dissertation examines two research questions: What questions do tutors ask of writers during tutorials, and what apparent effect do such questions have on the tutorial? Writing center theory has largely bulked question types together, causing many to hail questions as harmful while others to view them as essential.
In this study, 20 videotaped tutorials were observed wherein a total of 10 university writing tutors and 20 university student writers collaborated for nearly 427 minutes. During viewing, every question a tutor asked of a writer was logged and each context was analyzed. All sessions were viewed at least three times.
Tutors in this study asked 473 questions, allowing for 16 types of tutor questions to be identified. The types are labeled according to their effect and are further categorized into two categories. The first category is Interpersonal Tutor Questions, and it contains the following types: process, consent, rapport, gauging, filler, distracting, refocusing, and orienting. Effects include the ability to manage tutorials, gain permission, establish rapport, check writer understanding or mood, participate in chat, distract writers, refocus writers, and inform tutors. The second category is Making Meaning Questions, and it contains the following types: clarifying, verifying, transferring, suggesting, prompting, modeling, drawing, and exploring. The effects are the ability to clarify tutor understanding, verify that tutor understanding is correct, transfer expertise, suggest changes, lead writers through discovery, model thought processes, draw out information from writers, and challenge and stimulate writers’ ideas and views.
This study is a descriptive study, which examines tutor questions and their effects; it does not examine the mental processes behind questioning. The study’s conclusions are: tutors employ 16 tutor question types for various purposes; effective questioning requires the understanding of different question types; without an awareness of question types, tutors may find it difficult to know how to ask probing or challenging questions; the writing center field’s understanding of tutor questions can benefit from question research in other fields, but direct parallels made without consideration of the unique culture of writing centers may lead to tutor confusion. This study provides an important step toward tutor questioning training. (Cook, 2006, pp. iv–v)
CORRELATIONAL RESEARCH DESIGN
Correlational research is based in “relationship” just as its name implies. It is grounded in interactions of one variable to another; for example, as scores on one variable go up the related scores on another variable go down. You might find an inverse correlation between positive praise and acting out behavior—as the amount of praise increases, the amount of acting out behavior decreases. In correlational research, the degree to which the variables are related is important as well as the direction of the relationship. This type of research is not specifically causal-comparative research nor is it ex post facto research. Where correlational research actually relates scores from two or more variables from the same sample, ex post facto research compares scores from two or more groups on the same variable. Generally, correlational research does not signify causality, but some correlational research designs, such as path analysis and cross-lagged panel designs, allow for causal conclusions.
In correlational research, if a relationship is found between two or more variables, then the variables are correlated. This correlation must be interpreted based on the strength and direction via the correlational coefficient that ranges between –1 and +1. The direction of the correlation or the association between the two variables can indicate a positive association (0 to +1). This means that as one variable increases, the other variable also increases, and vice versa. Correlation coefficients between 0 and –1 represent a negative association. In this sense, as one variable increases, the other variable decreases. The magnitude or strength of the correlation or association between the two or more variables is signified as stronger by correlation coefficients closer to 1 (either positive or negative). A correlation coefficient of +1 indicates a perfect positive relationship, while a coefficient of 0 indicates no relationship, and a coefficient of –1 indicates a perfect negative relationship.
The correlational coefficient is represented by the letter r. In education, generally a correlation of 0.30 may be considered significant, and any correlation above 0.70 is almost always significant. A Pearson’s product moment correlation, represented by r, is the most widely used statistic of correlation. It is used when variables you want to correlate are normally distributed and are measured on an interval or a ratio scale. As you learned in statistics, the r is not a percentage score, but when you square the r (r2), then you have an index called the coefficient of determination that indicates how much of the variance of X has in common with the variance of Y. For example if r is .50, then the r2 would be 25%. This means that the two variables have 25% of their variance in common with each other. It can also be interpreted that 25% of the variance in Y can be accounted for by X. The coefficient of determination should be reported as it lends meaning to your findings by giving you the size or degree of the relationship. Cohen (1988) indicated that correlations can yield effect size in the r2 and determined that a small correlation effect size would be r2 = .01, a medium effect size would be represented by r2 = .09, and a large effect size would be r2 = .25. Effect size is recommended to be reported in the findings as it quantifies the distance from the null hypothesis (in correlational study the null hypothesizes that no relationship exists between the two variables) (Thompson, 2000).
Correlational research is used in a wide array of studies that intend, of course, to determine relationships, but also that aim to assess consistency, as well as predictions. As you may have learned from the chapter on statistics or your course in statistics, when the p-calculated is less than p-critical (in education it is usually established before a study is conducted and set as .05), such a relationship is significant. In correlational research, sample size is an important consideration. As you already know the smaller the sample size, the larger the size of the correlation has to be to be statistically significant; on the other hand, the larger the sample size, the more likely a significant correlation will be found even if it is with small magnitude. Therefore, in research with large sample size, reporting effect size is meaningful and critical.
Following are some types of correlations that may be used in correlation research when data are specific to ratio or interval scales.
Bivariate Correlation
For bivariate correlation, you would be investigating the relationship between two variables. For example, you may be interested in determining the relationship between the amount of time sixth grade students spend reading in structured partnership reading in school and their reading scores at the end of a six-week period. An example of a dissertation that used bivariate correlation was conducted by Siemers (2006). (See Example 3.4.)
Example 3.4
Abstract
Aggression negatively impacts children’s psychological and academic well being. The playground environment is more susceptible to aggressive behavior than more structured academic settings like the classroom. In this study, the relation between staff and student reports of student playground aggression was examined. Then, the relation between environmental playground factors (i.e., playground activities, playground supervisor ratios, active supervision, and playground rules) and student reports of aggression was examined. Finally, the relation of playground aggression to student self-reports of playground worries was evaluated. School Climate Theory provided the conceptual framework for evaluating playground characteristics, student aggression, and worry. Participants included 767 third, fourth, and fifth grade students and 57 playground supervisors from 10 Midwest elementary schools. Participants completed reports of aggression (students and staff), playground worries (students only), and playground environmental factors (staff only). Bivariate correlations were used to examine the relation between staff and student reports of aggression. Hierarchical Linear Modeling analyses were conducted to examine the relation among playground factors, aggression, and playground worries. This study used a novel approach to measuring the predictive relation of aggression on children’s worries through student self-report measures. Results of this study showed that staff and students’ reports of playground aggression were inconsistently correlated. Although overt physical and verbal playground aggression scores were significantly correlated, relational playground aggression and playground conflict were not associated. Additionally, four playground characteristics were linked to playground aggression: cooperative games, supervisor ratios, active monitoring, and playground rules. Finally, when students reported more playground aggression, their reports of playground worry were higher. (Siemers, 2006, pp. ii–iii)
Regression and Prediction
For regression and prediction, you would be assessing if a correlation exists between two variables. You might know the score on one and wish to predict the score on the second variable, and regression is related to how well you can make that prediction. The closer the correlation coefficients are to –1 or +1, the better your predictions become. A strong prediction is evidenced on packs of cigarettes—the more a person smokes, the higher the prediction of the person developing lung cancer or other cancers.
An example of a dissertation related to compensation strategy on nonprofit organizations is provided by Hoke (2006). (See Example 3.5.)
This study examined the effects of a total compensation strategy on the culture of a nonprofit. This study also examined employee perceptions of the new total compensation strategy. A paired-samples t test was used to determine whether a statistically significant difference existed between the means of the variables designated as productivity, innovation, and culture before and after the implementation of a new total compensation strategy. A Pearson r was used to determine the degree to which productivity, innovation, and culture variables were associated. A simple linear regression was used to examine how effectively the total compensation strategy variable allowed prediction of the value of the variables designated as productivity, innovation, and culture. The study was initiated in order to increase understanding of those elements that influence organizational productivity, innovation, and cultural change. The setting was a donative nonprofit in St. Louis, Missouri. The data analysis in this study revealed that the independent variable; i.e., the new total compensation strategy, had a statistically significant effect on the culture of the nonprofit organization. The findings presented in this study also illuminated ways in which employees viewed the new total compensation strategy as most helpful; i.e., providing a sense of ownership and motivation, and conversely, as least helpful; i.e., confusion about how the strategy works. An integrated analysis of the quantitative and qualitative data also suggested ways in which the effect of the new total compensation strategy could be increased. For example, the analysis suggested that more support for being creative and doing new things would increase organizational productivity and innovation. As a second example, the analysis suggested that more freedom for employees to exercise discretion in decision making would increase organizational productivity and innovation. Future research that addresses the application of organizational interventions developed in the private sector to the nonprofit sector and considers the impact of personal value systems will further refine the body of evidenced-based interventions. These interventions will assist nonprofit organizations to enhance sustainability, contribute to the effectiveness of organizational functioning, and finally the satisfaction and well-being of those employed by the organization. (Hoke, 2006, pp. i–ii)
Multiple Regression
For multiple regression, you would add more variables and more power in terms of making more accurate predictions. The variable that you would be predicting in multiple regression is called the criterion, outcome, or dependent variable, while the variables you use to make the prediction are called the predictor or independent variables. If you are trying to predict English reading scores for English language learners in a structured intervention, you might have as predictor variables the amount of time taught in English, the level of English of the parents, and the amount of time spent in academic oral language engagement with the child. You would have one criterion variable measured by some predetermined reading assessment and you would have three predictor variables. Do not forget to calculate your effect size for multiple regression and other correlational designs. An example of the effect size for multiple regression follows. Cohen’s f2 is the appropriate effect size measure to use in the context of multiple regression or multiple correlations. The f2 effect size measure for multiple regression is as follows:
(R2 is the squared multiple correlation derived from the regression model).
By convention, f2 effect sizes of 0.02, 0.15, and 0.35 are considered small, medium, and large, respectively (Cohen, 1988). Other multiple regressions, such as hierarchical multiple regression, will have a slightly different formula for the effect size. You will need to consult a statistics manual.
The following is an example of a dissertation in which multiple regression was used in a correlational research design (Bean, 2007). (See Example 3.6.)
Rates of obesity in children are rising at an alarming rate, particularly among girls and ethnic minorities. Engaging in regular physical activity can help reduce this risk. Little is known about factors associated with physical activity (PA) in preadolescent populations, an age when intervention is ideal. Guided by Social Cognitive Theory, this study used a repeated-measures design to examine PA and its correlates, including PA self-efficacy, outcome expectations, and social influences (from parents and peers), among participants (N = 57) in Girls on the Run, an innovative PA intervention for elementary school girls. Participants (M age = 9.4) predominately include girls from ethnic groups at highest risk for obesity, with 74% African American and 18% Hispanic. Multiple regressions indicated that, at baseline, girls with higher self-efficacy were significantly more likely to report greater intentions to be physically active (β = .40, p < .05). Further, although no mean changes in study outcomes were found, an examination of factors associated with the variance in PA behaviors and intentions at posttest can further understanding of PA in this age and ethnic group. Processes of change regressions suggested that, after adjusting for baseline levels, increases in both self-efficacy and social influences were significantly associated with higher physical activity behaviors and intentions at posttest (p < .05). Outcome expectations, or belief in the benefits of physical activity, was not a significant variable in the models (p > .05). Overall, findings suggest the importance of targeting physical activity self-efficacy and fostering high levels of peer and parental support for physical activity to help girls meet recommended guidelines. Implications for future interventions are discussed. (Bean, 2007, p. vii)
Canonical Correlation
Canonical correlation uses multiple correlations and has more than one criterion variable to a prediction equation. It allows you to see which of the variables are most important to relationships between the sets of independent and dependent variables. An example of a dissertation that included canonical correlation as its primary technique was done by McCoy (2005). (See Example 3.7.)
Example 3.7
Abstract
The purpose of this study was to examine hypothesized relationships between participants’ characteristics (i.e., academic interest area, prior exposure to persons with disabilities, the positive personal characteristics of humor, hope, and gratitude) and their self-report of cognitive, affective and behavioral responses to persons with disabilities, while controlling for social desirability. Undergraduate student participants at three Mid-Atlantic colleges were recruited from introductory courses in human services to complete survey packets.
The instruments used to gather data for this study consisted of a demographic survey designed by the author, the Marlowe-Crowne Social Desirability Scale (Crowne & Marlowe, 1960), the Humor Styles Questionnaire (Martin, Puhlik-Doris, Larsen, Gray, & Weir, 2003), the Adult Dispositional Hope Scale (Snyder et al., 1991), the Gratitude Questionnaire-6 (McCullough, Emmons, & Tsang, 2002), the Attitudes Toward Disabled Persons Scale—Form O (Yuker, Block, & Campbell, 1960), the Relationships with Disabled Persons Scale (Satcher& Gamble, 2002), the Interaction with Disabled Persons Scale (Gething& Wheeler, 1992), and the Situational Response Questionnaire (Berry & Jones, 1991).
A canonical correlation analysis was used to identify a combination of personal characteristics that are correlated with a combination of reactions to persons with disabilities. The primary multivariate finding of this study was that the combination of some, but not all, of the personal characteristics explored were correlated with the combination of some, but not all, of the measures of reaction toward persons with disabilities. A construct composed of the personal characteristics of affiliative humor and majoring in a human services field and, to a lesser degree, self-defeating humor, hope, and prior exposure to persons with disabilities was correlated with a construct composed of cognitive attitudes toward persons with disabilities, and to a lesser degree, affective responses toward persons with disabilities. Though the findings of this study were significant, the amount of variability in criterion variables that was explained by the predictor variables was relatively small. This suggests that other factors account for the rather large amount of variability unexplained by the predictors included in this study. (McCoy, 2005, p. iii)
Discriminant Analysis
Discriminant analysis is used to compare two or more groups on a set of variables. The criterion variable in this correlational analysis is the group membership. Predictor variables are input and a formula or model results. An example of discriminant analysis employed in a dissertation from Thanasui (2005) follows. (See Example 3.8.)
Example 3.8
Abstract
This study investigated the influence of past witnessed or experienced abuse on heterosexual cohabiting couples’ Premarital Personal and Relationship Evaluation for Cohabiting Couples (PREPARE-CC) couple relationship types. The researcher utilized preexisting data from 5,000 cohabiting couples who had previously participated in the PREPARE marriage preparation program and had completed the PREPARE-CC inventory including a demographic section that elicited information about past abuse. Discriminant analysis was conducted in SPSS to answer the question of whether the presence of past witnessed or experienced abuse could successfully predict relationship type among cohabiting couples.
Results of the discriminant analysis yielded no significant ability to classify cohabiting couples by individuals’ experience of past abuse, however, isolating females and males with the highest frequencies of past abuse indicated that males abused “very often” had a higher frequency of higher-satisfaction relationship types than the general sample consisting mostly of individuals with little or no history of abuse. Females reporting abuse “very often” did not follow this same pattern. Recommendations were made for future longitudinal studies and for strength-based research on healthy heterosexual cohabiting couples in an effort to understand what contributes to these couples’ success. (Thanasui, 2005, p. iii)
Factor Analysis
Factor analysis is another statistical procedure that uses correlations to identify basic patterns of variables. If a large number of variables having intercorrelations are present, then a common factor could be identified for that set of variables. Several factors could be identified among a large set of variables if you are studying leadership and organizations. For example, you might find two factors such as leadership behaviors and organizational structures. The two factors might be helpful in describing the success of superintendents. An example of factor analysis used in a dissertation by Evans (2006) follows. (See Example 3.9.)
Example 3.9
Abstract
The purpose of this study was to examine the validity and reliability of scores from a survey instrument, the Technology Integration Survey (TIS), designed to measure technology integration skills in preservice teachers. All of the items used in the survey were based on the ISTE NETS for Teachers (NETS-T) standards. Participants in the study were undergraduate students enrolled in the preservice teacher education program at a large university in southeastern North Carolina. Validity and reliability studies were conducted on the TIS including: (a) a study of representativeness and alignment of survey items to the NETS-T standards, (b) analysis of the internal structure using exploratory factor analysis, (c) pre/posttest testing after technology intervention, (d) correlations between the TIS and an instrument with known validity and reliability, (e) correlations between TIS and field experiences or online portfolios, and (f) analysis of internal consistency. Four factors were extracted using a principal components analysis. The factors (a) Planning and Designing Learning Environments and Experiences, (b) Teaching and Professional Practice, (c) Social, Ethical, and Human Issues, and (d) Assessment and Evaluation explained 63% of the variance. Reliability coefficients on the four subscales ranged from .83 to .89. Evidence from the study supports the use of the TIS as a viable tool to measure technology integration skills in preservice teachers. (Evans, 2006, p. iii)
Path analysis may be used to assess which of a number of pathways connects one variable to another. For example, in the relationship of smoking and lung cancer, there might be a small to medium path that runs through healthy activity, but a major path that goes through smoking behavior. In this, you could surmise that even with running, walking, and eating properly, if a person continues to smoke heavily, he or she has a stronger likelihood to develop lung cancer.
A dissertation example that used path analysis was completed by Choi (2005). (See Example 3.10.)
Example 3.10
Abstract
The purpose of this study was to identify the physical activity behavior (leisure-time, household, job-related, and transportation-related) and to examine the relationships between physical activity behavior and the correlates of physical activity including acculturation as well as environmental resource, social support, and cognition (self-efficacy, decisional balance) in 200 adult Korean immigrant women aged 20–64. Although the health benefits of physical activity are well established, the current physical activity level of the U.S. public, particularly for ethnic minority and immigrant women, does not reach the Healthy People 2010 goals. There are few studies of physical activity and its correlates for Asian-American women and none include Koreans. A descriptive cross-sectional survey design was used. Korean ethnic church-based recruitment was the primary means of accessing this population. Physical activity was measured with the long form of the International Physical Activity Questionnaire. Acculturation was measured with proxy measures (language, years of residence in the U.S., age at immigration) and the Vancouver Index of Acculturation. McAuley’s Barriers Self-Efficacy measure and Marcus’ Decisional Balance measure provided measures of cognition. The mean age of the participants was 41 years, 84.5% were married, and 71% had a college degree or higher. In general, the women were not physically active in their leisure time, but most of them were physically active when other domains of physical activity were examined together. A path analysis showed that years of residence in the U.S., self-efficacy, and decisional balance had direct positive effects, and number of children under five years had a direct negative effect on leisure-time physical activity. Age and American acculturation had indirect positive effects through self-efficacy while depression had an indirect negative effect through decisional balance on leisure-time physical activity. A total of 14 percent of the variance in leisure-time physical activity was explained. The findings showed that acculturation is an important correlate of leisure-time physical activity among Korean immigrant women. In addition, self-efficacy and decisional balance were essential cognitive correlates of leisure-time physical activity and they played mediating roles. Findings provide direction for developing targeted physical activity interventions for Korean immigrant women based on their self-efficacy, decisional balance, and acculturation. (Choi, 2005, n.p.)
Cross-Lagged Panel
Cross-lagged panel (longitudinal) correlational designs measure two sets of variables simultaneously at two points in time. For example, the correlation between early exposure to oral English literacy in prekindergarten and reading achievement in fifth grade is compared with the correlation between later reading achievement in fifth grade and later exposure to oral English literacy in first grade. Some of the data points in this design are treated as temporality delayed or lagged values of the outcome variable. According to Mackinnon, Nohre, Cheong, Stacey, &Pentz (2001):
Cross-lagged correlation approaches were criticized for several reasons, including problems due to measurement error (Kessler & Greenberg, 1981). Measurement error was later incorporated in the cross-lagged model by the estimation of latent variable models for the observed measures. These methods enjoyed several years of wide and convincing applicability (Bentler&Speckart, 1979, 1981; Kessler & Greenberg, 1981) until their limitations were clearly described (Rogosa, 1980). According to Rogosa and Willett (1985), these models do not explicitly specify individual change, do not easily generalize to more than two waves of data, and the data analysis only includes the covariance matrix and not the means. Stoolmiller and Bank (1995) added that autoregressive models fail when there is high-rank order stability over time and require the questionable assertion that the prior score on a variable causes the subsequent score on that variable. (p. 221)
An example of a cross-lagged panel correlational design was used in a dissertation by Coon-Carty (1998). Vygotsky argued that children’s development is most likely to occur when, in the course of collaboration, assistance is provided within their zone of proximal development—the distance between what a child can achieve independently and what he or she can do with the assistance of a more competent peer. This dissertation discusses the results from a longitudinal investigation exploring the impact of learning in either a multiage or traditional classroom on students’ subject-specific academic self-concepts. Results from a cross-lagged panel correlation analysis, testing the predominant causal flow from subject-specific self-concepts to academic achievement are also described. Participants in the study consisted of 189 first (106) and fourth (93) grade children who were drawn from a public elementary school in Salt Lake City, serving predominantly European-American middle- and lower middle-class students. Eighty-one of the participants were in a multiage classroom with the remaining 108 in a traditional classroom. Multiple self-concept measures and a standardized achievement test were given to the participants toward the beginning and at the conclusion of the academic year. Results indicated no significant differences for math and reading self-concepts between the multiage and traditional participants and the end of the academic year. Learning in a multiage classroom did not increase the participants’ math or reading self-concepts over the course of the academic year. Significant differences were found as a function of setting for male participants. Fourth grade males in the multiage classroom reported significantly lower control over performance scores at the end of the academic year compared to the fourth grade males in the traditional classrooms. Results from the cross-lagged correlation design demonstrated a causal flow from reading self-concept to subsequent reading achievement, suggesting that perceptions of reading abilities cause future reading achievement. The study did not show the same significant influence between math self-concept and math achievement. The pattern of associations between math self-concept and math achievement is more suggestive of a reciprocal relationship rather than a causal one. (Coon-Carty, 1998, pp. x–xi)
Other Correlation Coefficients
For data that are represented by rank scores, an ordinate coefficient of correlation, Spearman’s rho (ρ), is used. It is interpreted in the same way as the Pearson r and ranges from –1 to +1. A phi coefficient (Φ) is used when both variables are dichotomous scores (1 or 0). This is often used when sex of individuals is a variable. For example, you might want to determine on a particular college campus if drinking is more related to men than to women. The phi coefficient reveals both strength and direction of relationships. Although we have presented only three types of correlation coefficients, there are others that are available to you for use with nominal and ordinal data and that can be used with more than two variables.
Advantages and Disadvantages
There are advantages and disadvantages to correlational designs. First, the advantage is that almost any variable that you wish to study can be investigated. Additionally, it can be used in a predictive manner with one variable or more predicting another. Disadvantages that have been discussed are that correlational research is subject to extraneous variables and that causation cannot be inferred.(See Example 3.11.)
CAUSAL-COMPARATIVE RESEARCH DESIGN
Causal-comparative research, or ex post facto (after the fact) research, is the most basic design for determining cause-and-effect relationships between variables. Causal-comparative research is different from experimental research in that you do not manipulate the independent variable since it has already occurred, and as already indicated, you cannot control it. Additionally, causal-comparative research will not meet the experimental research requirement for random assignment of participants from a single population or pool of participants. At least two comparison groups are needed in causal-comparative research to be compared on a dependent variable. One group may be students who have been identified by their teachers as being potentially gifted, and the other group may be those students not identified by their teachers as being potentially gifted. These two groups could be measured on standardized IQ tests to determine if the group differences identified by the teacher perception are accurate as compared by the standardized measures.
When you discuss the groups, you will want to make certain you describe the groups and their characteristics in detail and the independent variable that distinguishes the groups. Why is this important? It is important because the group constitution and definitions can affect the generalizability of your findings. Give ample detail to understand all about each of the groups in terms of their comparability and differences. You may obtain your groups from two independent established populations, such as those students who have had public school prekindergarten and those who did not. You may want to compare these students on English reading achievement at the end of kindergarten or first grade. To obtain your sample you may conduct random sampling (not random assignment from one single population as in experimental research). You will want your groups to be as similar as possible on some variables. For example, in the prekindergarten versus no prekindergarten groups, you may want to go through the random samples and match the samples (to have better control) on age, ethnicity, socioeconomic status, and gender. You want the groups to be as similar as possible so that the difference or lack of differences you see in the end of your research is more likely to be attributed to your independent variable. Because causal-comparative research does not rest on randomization or manipulation or control of variables, you will know up front that is a weakness if you so choose this type of design.
Weakness Controlled
One way to control for that weakness has already been mentioned—match the samples. Another way to control for confounding variables is to compare homogenous groups. In our previous example, we could take all students who had been to prekindergarten and who had not who were all English language learners and who were non-English fluent upon entry to kindergarten. Another way to further control the study would be to take that same group of students and separate them into subgroups that would represent varying degrees of English language fluency. Perhaps you might group the students into non-English fluent and limited-English fluent. This would permit you to determine the influence of English fluency when checking the independent variable’s influence on the dependent variable. If you like this approach, you may wish to test control variables by using factorial analysis of variance (known as ANOVA). This approach will allow you to determine the effect of the independent variable, as well as the control variable (separately and combined), on the dependent variable. A factorial analysis provides information about the interaction of the independent variable on the control variable at the different subgroup levels. Students who are limited-English fluent upon entry to kindergarten may indeed show significantly higher beginning English reading (literacy) scores at the end of the kindergarten year.
Another statistical technique that may be used as well is analysis of covariance (ANCOVA). ANCOVA adjusts for initial differences in your groups. For example, if students who have better English skills at the outset of the study and demonstrate in a pretest score significantly higher in English literacy/reading, then the ANCOVA will adjust for that. It adjusts both groups to equal standing. As instruction is then given to both groups using the same curriculum, the final outcome can be better assessed in terms of the actual kindergarten year intervention.
All methods discussed here are univariate analyses. When you have more than one dependent variable you may want to resort to multivariate analysis.
Data analysis techniques used in causal-comparative research designs usually consist of inferential statistical techniques such as t test (comparison between two groups), analysis of variance (comparison among more than two groups), and chi square (when outcome variable is dichotomous). We have a couple of words of caution as you begin your analysis and conduct your statistical tests. First, be careful on your interpretation of causality. Look at your data carefully to make certain in which direction the causal relationship flows. Additionally, remember to conduct a test for effect size for each of the typical statistical tests for this research design. Effect size takes various forms, such as r2 (see previous sections on correlational research). We present here a basic effect size type for t test for means; however, you will want to consult your committee statistician on this topic for calculating or obtaining these via your statistical package you use for your dissertation. Typically for a t test for testing means, you would use Cohen’s d. Here is the formula:
You will also want to report confidence intervals as well. The bottom line is that, whether it is chi square or t test, all must be followed by an effect size calculation for your research findings to be rendered worthwhile, as this calculation provides you with the strength of the association to the larger population as it relates to your study.
Following are two examples of causal-comparative research designs used in recent dissertations. (See Examples 3.12 and 3.13.)
Example 3.12
Abstract
The purpose of this study was to determine differences in teacher preparation and retention for two groups of Texas A&M University–Corpus Christi graduates certified in 2002 and 2003. Factors that influenced teacher attrition, such as teaching assignments, working conditions, salary, benefits, scheduling, organizational skills, rapport with administrators, colleagues, parents and students were examined to determine whether graduates who completed a formal mentoring program stay in the teaching profession at higher rates than the graduates who did not participate in the Strategies of Success (SOS) formal mentoring program.
Participants included 95 certified TAMU-CC graduates who met the inclusion criterion of no prior teaching experience.
A causal-comparative research design was used to compare two preexisting groups, non-SOS and SOS TAMU-CC certified graduates applying for certification in 2002 and 2003. Outcome measures examined were retention, reasons for leaving the field of education and satisfaction with the teacher education program at this regional university. The TAMU-CC Teacher Education Survey (Arnold & Ramirez, 2003) was used to collect data for this study. Results of the analyses showed no statistically significant difference between the non-SOS and SOS groups. (Arnold, 2006, pp. iii–iv)
Example 3.13
Abstract
This study analyzed test scores of economically disadvantaged students who attended two elementary schools implementing different types of Title I models from 1999–2001. Test scores from the Texas Assessment of Academic Skills (TAAS), the Iowa Test of Basic Skills (ITBS), and the Stanford Achievement Test (SAT-9) were analyzed. One school implemented the targeted assistance model (less than 50% poverty), which focused resources on students identified as failing or at risk of failing. The other a schoolwide model (95% poverty), which used resources to help all students in a school regardless of whether they ware failing, at risk of failing, or economically disadvantaged.
The quantitative approach was used with a causal-comparative design. A cohort of continuously enrolled students was identified for the TAAS (n = 155) and the ITBS/SAT-9 (n = 135). Descriptive statistics such as the frequency, mean, and standard deviation, were used to measure differences on the Texas Learning Index (TLI) for the TAAS, and Normal Curve Equivalent (NCE) on the ITBS/SAT-9. Analysis of Covariance (ANCOVA) was used to partially adjust for preexisting differences among the groups and because randomization was not possible. The independent variable was type of Title I model, targeted assistance or schoolwide. The dependent variable was the achievement measure, and the covariate was the initial achievement scores in third grade (pretest).
The ANCOVA reports and descriptive statistics showed that economically disadvantaged students performed better in reading and math on TAAS and ITBS/SAT-9 at the targeted assistance school in 1999 and 2001, with mixed results in 2000. The academic performance of economically disadvantaged students at the targeted model was consistent all three school years. They scored slightly lower than the non-economically disadvantaged students, but higher than their peers at the schoolwide model. The students’ third grade pretest score was the most significant predictor of future performance. (Hinojosa, 2005, n.p.)
QUASI-EXPERIMENTAL RESEARCH DESIGN
Even though the best causal research is reflected in true experimental designs, most research in education that requires causal inferences cannot be conducted under true experimentation due to the inability to randomly assign participants to experimental and control groups or the inability to secure a control or comparative group. Additionally, much of true experimentation is expensive in that it often requires training in the intervention and always monitoring for fidelity. In this case, quasi-experimental designs are available to give you adequate control over threats to validity. There are several types of quasi-experimental designs, which we believe would be advantageous to the development of your dissertation research.
Nonequivalent Control Group Design
This design is used most frequently in educational research. This design includes at least an experimental group and a control group with random assignment of participants to groups; however, there is random assignment of intact groups to treatments. Cook and Campbell (1979) outlined 11 nonequivalent control group research designs. Table 3.1 shares several of those 11 possible designs.
Examples of nonequivalent control group research designs from dissertations follow. (See Examples 3.14 and 3.15.)
Research Design Explanation
One-Group Posttest-Only Design Because there is no accurate information on pretest data from the group, the results of this type of design cannot be considered completely valid because changes in the dependent variable may be due to treatment, or they may be due to any number of threats to validity, such as history or researcher expectation. A caution is noted to gather and report as much information as possible related to the pretest conditions. However, this design is fairly inexpensive. It allows you to conduct research when there is no comparison group readily available. Often this design is used in evaluation research.
Posttest-Only Design with Nonequivalent Comparison Groups Design This design consists of the administration of a posttest to two groups—usually a treatment group and a comparison group. Because the two groups may not be the same prior to the beginning of the instruction, it is difficult to draw valid conclusions about treatment effect based solely on posttest information on two nonequivalent groups because effects may be due to treatment or to nonequivalencies between the groups.
One-Group Pretest-Posttest Design This is a common design (0 × 0) comparing performances of a single group of participants, but this design has weaknesses. It is subject to such threats to internal validity, such as history, maturation, regression toward the mean, testing, selection, and mortality. It is difficult to determine if the results are from the intervention or from confounding variables between the pretest and posttest.
Two-Group Pretest-Posttest Design Using an Untreated Control Group In this design, threats to validity are reduced because there is a comparison group; however, the groups are not equivalent. Therefore, there remains the possibility of the threats of selection. Performance trends in both the treatment group and the control group, can be established by using multiple pretests. There should be a change in the trend line for the treatment group but not the control group.
Purpose. The purpose of this study was to describe Advanced Placement (AP) calculus teachers’ and administrators’ perceptions of AP calculus program characteristics that have increased student enrollment in selected San Bernardino County public comprehensive high schools.
Methodology. Descriptive research methodology and a posttest only, one-group design was used to identify the current characteristics of AP calculus programs in the county of San Bernardino. The study examined the responses of twenty-five AP calculus teachers and twenty-seven site administrators in the county of San Bernardino. Teachers and administrators were surveyed with a questionnaire to obtain information on student preparedness for AP calculus classes, student learning, access, and placement. The study examined the educational/training of AP calculus teachers, as well as the types of instructional, administrative, district, and parental support AP calculus programs received. Free response questions were elicited to determine the teachers’ and administrators’ perceptions of AP calculus programs.
Findings. Administrators and teachers both perceived that there were gaps to students’ preparedness for AP calculus classes. Teachers were more pessimistic that AP calculus classes were supported from their school district. Teachers and administrators perceived that they received a lot of support from their site principal, but teachers were more pessimistic on parental support. The findings confirm that teachers and administrators perceived that students receive AP practice exams for support, after-school sessions, and individual tutoring, but students are not perceived to receive Saturday support sessions or guidance counseling for AP calculus classes.
Conclusions and Recommendations. AP calculus teachers and site administrators must generate strategies to increase student enrollment in AP calculus classes in selected San Bernardino County public comprehensive high schools. The study recommends that AP calculus teachers and site administrators collaborate to develop and implement strategies on student preparedness for AP calculus classes, student learning, access, placement, and the implications of those strategies for the design and integration of curriculum, instruction, assessment, and professional development. The study presents a set of recommended actions that could significantly improve existing programs for approaches to AP calculus courses and serve to promote programs for advanced study in the mathematics discipline. (Rodriguez, 2005, pp. iii–iv)
Example 3.15
Abstract
This study assessed the effects of general and specific supervisory feedback on counselors’-in-training ratings of counseling self-efficacy (CSE). Fifty-four students in counseling-related graduate programs from two universities in the Southeast and one in the Midwest volunteered as participants. Thirty-seven participants were female, 14 were male, and three did not indicate their sex. Forty out of 54 participants indicated they were Caucasian-American, five were African American, one was Hispanic-American, one was Asian-American, and three indicated “other.” The median number of months of previous clinical supervision for the participants was one month. This study made use of a two-group pretest-posttest design. The independent variable was performance feedback, with two levels (specific and general). The dependent variables included posttest scores on the Counselor Activity Self-Efficacy Scales (CASES), (Lent, Hill, & Hoffman, 2003). State anxiety was also assessed with the state scale of the State-Trait Anxiety Inventory (STAI-S), (Spielberger, 1982). After completing pretest measures, participants performed a ten-minute mock counseling session with a confederate. After the mock counseling session, the confederate provided either specific feedback or general feedback statements to the participants. Participants then filled out posttest measures. After the measures were collected, participants received an extensive debriefing. The three hypotheses for counseling self-efficacy were: (a) participants receiving specific feedback would obtain higher counseling self-efficacy scores; (b) posttest counseling self-efficacy scores would be significantly higher than pretest scores; and (c) while both groups were expected to score similarly in the pretest measure of counseling self-efficacy, it was predicted that the participants receiving specific feedback would report higher counseling self-efficacy scores in the posttest measures. There was no difference between counseling self-efficacy scores of participants who received general or specific supervisory feedback. Postmeasure counseling self-efficacy scores were significantly higher than the premeasure scores. There was no group × time interaction. There was no difference between anxiety scores of participants who received general or specific supervisory feedback. Postmeasure anxiety scores were significantly higher than the premeasure scores. There was no group × time interaction. Implications and issues to be considered in future research were discussed. (Clark, 2005, pp. vi–vii)
Time Series Design
The time series design is a variation of the one-group pretest–posttest design. Such a design is diagrammed in Figure 3.2. The one- or single-group time series design is where one group of participants is tested repeatedly both before and after treatment. The repeated testing establishes stability prior to the treatment, and the repeated testing after the treatment establishes confidence in the effectiveness of the treatment. You would essentially investigate the pattern of test scores pre- and post-treatment. There are threats to validity related to history and instrumentation if the instrument is changed during the research. A multiple time series design, adding a control group, can eliminate the threats.
Chapter5
Qualitative Research Designs
In this chapter, we continue the discussion of methods of inquiry from Chapter 3; however, in this chapter we guide you through qualitative and mixed methods research designs. Just as quantitative research begins with basic human observation and curiosity, qualitative research does also. The science of qualitative research seeks to have the researcher look deeply into the world of individuals and phenomena.
Your dissertation, as a product of qualitative research, will likely be based on small nonrandom samples, or it may relate to multiple concepts, phenomena, or theories, or it may yield theory. Basic factual information or uniconceptual research in qualitative designs report information on one concept or phenomenon and does not attempt to make broader commentaries on the data. It is confined to one dimension and the small sample. It may be applied to similar groups in similar situations. Qualitative research can become more complex as the researcher uses methods of inquiry that promote cross-analyses or comparisons of data. Theory development, grounded theory, is often the most difficult, but most rewarding outcomes of the methodological techniques.
Just as in quantitative research, also in qualitative research, you must ask yourself: what is my intent or purpose of the research? Second, you must ask: what are my research questions? Again, these initial components will drive the method you select. If you conclude that you need to go obtain a deeper understanding of the topic, you will select qualitative methodology. Ultimately, you may decide that a mixed methods approach, using both quantitative and qualitative analyses, is the best method of inquiry for your dissertation. In this chapter, we provide an overview of each of the data analysis techniques or inquiry methods you could use if your questions or purpose necessitate a qualitative approach. Certainly, this chapter is not all inclusive, and we know that you will want to investigate more fully other options on your own once you settle on a specific method of inquiry. You will need to consult qualitative research methods textbooks and review your proposed methodology with your dissertation advisor. You must be thoroughly familiar with your method of qualitative inquiry.
Your research design or method of inquiry for qualitative will likely fall into one of four major categories of research: (a) phenomenological research, (b) case study research, (c) ethnographic research, and (d) grounded theory research. In addition to providing you with a brief overview of each category, we also provide examples of research purposes or questions that would justify a specific type of qualitative research design or method of inquiry. We also give several other types of research in qualitative inquiry. In a final section of this chapter, we discuss mixed methods research and provide some examples of it.
According to Creswell (2007),
A qualitative study is defined as an inquiry process of understanding based on distinct methodological traditions of inquiry that explore a social or human problem. The researcher builds a complex, holistic picture, analyzes words, reports detailed views of informants, and conducts the study in a natural setting. (p. 15)
Therefore, qualitative research emphasizes understanding by closely examining people’s words, actions, and records, as opposed to a quantitative research approach that investigates such words, actions, and records at a mathematically significant level, thus quantifying the results of observations. Qualitative research examines the patterns of meaning that emerge from data gathered; such patterns are often presented in the participants’ own words. Many students believe that qualitative research is easier than quantitative research; however, it takes not only a logical mind, but one that can take and make sense of ambiguity by searching for patterns and connections. Your task in qualitative research is to find patterns within words and observed actions, and to present those patterns for others to inspect while at the same time staying as close to the construction of the world of the participants as they originally experienced it. Again, your job as a researcher is not to make broad generalizations; rather, your job is to contextualize the findings.
PHENOMENOLOGICAL RESEARCH DESIGN
Phenomenological research is one of the most basic forms of research. This type of research involves the description of phenomena in our world. In this type of inquiry, the phenomena described are basic information, actions, behaviors, and changes of phenomena, but always the description is about what the phenomena “look like” from the perspective of the researcher and the participants in the research; it is not about how the phenomena function. Prior to beginning the research, you will need to have established from theory or from prior research what it is about the phenomenon you will study; from this perspective, descriptive research is theory or research driven.
Husserl, the twentieth-century philosopher and the father of phenomenology, was concerned with the study of “experience” from the perspective of the individual, and believed that the researcher could approximate those experiences through intuiting and rigorous examination of the subjects, objects, or people’s lived experiences, behaviors, or actions. He believed that researchers could gain subjective experience, essential realities, and insights into a person’s or persons’ motivations and actions; thus, researchers could minimize presuppositions and traditionally held beliefs. The researchers’ interpretations of the phenomena allows it to, as in action research, inform, support, or challenge policy, procedures, and actions in society or organizations.
In a phenomenological research design, the researcher is concerned with clarifying the specific and recognizing phenomena through the eyes of the participants. Deep and rich descriptions of the phenomenon or phenomena are usually gathered through inductive, qualitative methods such as interviews, focus group discussions, and participant observation. Although phenomenological research has been related to other essentially qualitative approaches including ethnography (discussed later in this chapter), hermeneutics (analysis of the written word), and symbolic interactionism (making meaning of individual’s interactions), it usually is akin to descriptive research in that it is about describing rather than explaining. Much of this type research has the researcher come to it without any preconceived notions or hypotheses; however, it is extremely unlikely that you, as the researcher, will not have a preconditioned paradigm, purpose, or hypothesis. Therefore, you, as the researcher, never can be removed from your own presuppositions about the phenomena; therefore, you should admit your own perspective under a heading in your dissertation methodology as “Researcher’s Perspective” or “My Personal Perspective.”
Techniques/Approaches/Methods
Phenomenological research can be based in either single-case designs or purposefully selected samples. A variety of qualitative techniques, approaches, or methods can be used in phenomenological research, including interviews, focus groups, participant or direct observation, and document or personal entry analyses. Of course in any of these approaches, the establishment of trust is important via good rapport and empathetic listening. The techniques are explained in more detail in Table 5.1.
Phenomenological Technique, Method, Approach Defined Explanation
Interview—The interview, both factual and meaningful, seeks to describe the meanings of central themes in the life world of the subjects. The main task in interviewing is to understand the meaning of what the interviewees say (Kvale, 1996). The interview can be structured, semistructured, or unstructured. In a structured or semi-structured interview, you will need an interview protocol that has been face validated. The interview is usually recorded and then transcribed with notes taken during the interview added later. Prior to the interview, you must: (a) secure a location with the least distraction, (b) explain the purpose of the interview, (c) address confidentiality and provide a consent form for signature, (d) explain the format of the interview, (e) indicate the length of the interview, (f) give your contact information, (g) allow interviewee to ask any questions about the interview, and (h) determine how you will record the interview data (if the tape recorder is used, verify that it is working periodically throughout the interview and if you take notes, be thoughtful of how and when notes are taken, not to make the interviewee feel uncomfortable—notes may be used for observations during the interview only).
During the interview, you will need to attend to: (a) careful listening, (b) nonverbal cues, (c) the progress of the conversation, (d) probing when needed, (e) taking notes, and (f) not responding during the interview.
Questions in the interview should be sequenced as follows: (a) ask factual, basic questions first, simply getting the interviewee involved, then place factual questions throughout the interview, (b) present-based questions should be asked prior to past-based or future-based questions, and (c) allow the interviewee to add any additional information at the end of the interview.
Note: the interview is time-consuming and resource extensive.
Focus Groups—This technique is a form of interview, but with a group. It focuses on data generated via observation and communication between and among participants. Denzin and Lincoln (1994, p. 365) stated that Merton et al., coined the term focus group in 1956 to apply to a situation in which the interviewer asks group members very specific questions about a topic after considerable research has already been completed, while Kreuger (1988) defined a focus group as a “carefully planned discussion designed to obtain perceptions in a defined area of interest in a permissive, nonthreatening environment.” (p. 18) You, or someone you train, can lead or moderate the group, which usually consists of four to twelve people. The less experienced the interviewer, the more people in the group may be likely to be frustrating or unwieldy; however, the less experienced the interviewer, the fewer people may likely not be encouraged to generate the information needed that a few more in a group could generate. In this technique, you would have an interview protocol established and validated. Additionally, it should be noted that you should not be alarmed if one participant’s discussion is influenced by another. This is normal in focus group research. We often call it “piggybacking.”
It is less time consuming and resource intensive than the individual interview. The protocol should go from more general questions to more specific and should consist of about 10 questions.
Participant or Direct Observation—The participant observation technique is an observational method in qualitative research that uses the five senses to describe the (a) setting, (b) people, (c) occurrences, and (d) meaning of what was observed. In other words, it examines the intricacies of the interactions and relationships of individuals. It investigates the phenomena of this. This technique is usually done onsite; however, with the virtual world, it may be conducted using photographs, videotaping or streaming, and audiotaping or streaming. You will, in participant observation method, intentionally put yourself in the context or location of the phenomena being studied over an extended period. As a participant observer, you will become a part of the group and become fully engaged in experiencing what the participants are experiencing. You may combine some unstructured interviews during the time you are also observing. The extent to which you participate and the length of participation is up to you as the researcher. It truly depends on your research purpose.
In direct observation, you will conduct systematic observation and documentation of phenomena in its usual setting or location. There may be limitations in terms of gender, time, language, age, socioeconomics, or culture that you, as the researcher, will need to consider prior to entering as a participant. You may consider having other researchers make observations at the same time you do. You may wish to observe at other times—vary your times of observation to make your finding more reliable.
In direct observation you will be a quiet observer and will not interact with the participants; you will not intervene in situations in an attempt to influence the outcome of the study. As participant observers, you become a part of the group and are fully engaged in experiencing what those in the study group are experiencing. You may, however, conduct focus groups or individual interviews at another time during the study. But, unlike participant observation, the interviews will be considerably more structured and will likely be based on things you have observed so that you can obtain more in-depth understanding about what it is that you observed.
In both participant and direct observation, you will want to make notes on: (a) the physical environment, (b) sociological aspects of community, (c) hierarchical structures, (d) language, (e) communication venues, (f) unspoken communications or norms, (g) activities, and (h) your own thoughts and responses. Take down as much of the direct conversations as possible for reproduction of thoughts of the individuals and group.
Document Analysis—This particular technique in qualitative research is related to the critique or analysis of documents for significance, meaning, and relevance within a particular context and phenomenon. Documents can consist of historical papers, personal entries, clinical records (if available), video, photos, electronic media, collections (stamps, clothes, other materials), books, newsletters, or newspapers (not an all inclusive list). Using this technique, you will (a) define your purpose(s), (b) collect your documents, and (c) review and analyze your documents. When you review your documents, you will want to label your documents and perhaps use differing colors of Post-it tabs for varying themes you find. You will want to use the constant comparison method as defined by Patton (1990) as “to group answers… to common questions [and] analyze different perspectives on central issues.” (p. 376)
Discourse Analysis—Discourse analysis is actually not a specific research method, but can inform the phenomenon being studied. It provides a deconstructive reading and interpretation of a problem or text. Discourse analysis can be stated as a set of methods and theories for investigating language in use and language in social contexts (Yates, Taylor, &Wetherell, 2001). It can be used as well in ethnography to understand and examine a person’s social world. Approaches to discourse analysis may include discursive psychology; conversation analysis; critical discourse analysis and critical linguistics; and sociolinguistics. The defining of discourse analysis is part of its explanation. Stubbs (1983) defined it as being (a) concerned with language use beyond the boundaries of a sentence/utterance, (b) concerned with the interrelationships between language and society, and (c) concerned with the interactive or dialogic properties of everyday communication. The validity and reliability of discourse analysis may be found in four approaches: (a) deviant case analysis, (b) participant understanding, (c) coherence, and (d) reader’s evaluation. (Potter, 1996)
Following are examples of phenomenological studies by Meyertons (2006) and Gitonga (2006). (See Examples 5.1 and 5.2.)
Example 5.1
Abstract
This phenomenological study investigated the experiences of a set of faculty who taught classes in hybrid format at a small liberal arts university in Salem, Oregon. For this study, a “hybrid format” course was defined as a course that includes elements of both traditional face-to-face and technology-enhanced (often Internet) course components. The study consisted of a set of heuristic interviews with faculty members identified through an empirical survey I conducted in Fall 2002 as part of my duties as Director of Instructional Design and Development for the university’s technical services department.
Higher education leaders have consistently identified technology integration as an important priority for their faculty. Since in many cases faculty have proven reluctant to do so, it is clear that there has been some dissonance between leadership expectations and faculty experiences. An extensive review of relevant literature indicates that little research has been conducted specifically on the faculty experience with educational technology, although much evidence has been gathered on the student experience and on learning outcomes. The goal of this study was to discover if there were any common elements that faculty experience in working with hybrid formats, and to try to distill these elements into a set of recommendations to higher education leaders for improving faculty experiences with educational technology. The broader goal was to help develop practices that might improve ways faculty use educational technology to enhance teaching and learning. (Meyertons, 2006, pp. 1–2)
Example 5.2
Abstract
The purpose of this study was to explore beginning counselors’ experience of their first counseling position in their first year. Four counselors, who shared a similar counselor education background but who also differed by their degree majors, described their experiences. Using phenomenological methodology to collect and analyze data, two rounds of interviews were conducted in order to allow the four participants to individually describe their experiences. A focus group composed of all the participants verified the accuracy of the research findings. The study resulted in a fusion of common themes. Within the theme of hope and expectations, the essence of the beginning counselors’ experience revolved around support, clientele, salary, time, and continuity. Other common themes that emerged from the study were: (1) the process of transitioning from students to counselors, (2) exciting as well as challenging experiences as beginning counselors, and their perceptions of their level of preparedness for the first counseling position. In these themes, there were commonalities as well as “uniquenessalities” of the essence of their experience. (Gitonga, 2006, p. x)
CASE STUDY RESEARCH DESIGN
CASE STUDY RESEARCH DESIGN
Case studies are specific explorations of individuals, but also such investigations can be on groups, cohorts, cultures, organizations, communities, or programs. If a case study is based in biography, it may be called life history, which focuses on major circumstances, situations, events, problems, celebrations, and/or decisions of a person, group, or organization. If a case study is focused on one individual or group, it is called single-case design. If the case targets multiple individuals and the same phenomena or it targets various communities related to a similar phenomenon, then this type of design is called multiple-case study design. It may even be called a phenomenological case study, combining as we indicated the case study and phenomenology.
More than likely, you will be using purposeful sampling in case study design. You must describe your sampling procedure and case selections in detail. Share all the characteristics of the selection criteria and distinctive and uncommon features. Also indicate the duration of the study, as that may have a bearing on the outcome.
Cases may also be classified as one of the following: deviant or extreme case, critical case, convenience case, typical case, or politically important case (Patton, 2002). An extreme case study would be if you were to study one or more persons at some extreme. For example, you might study superintendents who have served in urban school districts only as superintendents. A critical case would be one that makes a point. It is like this: if it happens in this case, then it will happen in any case, or vice versa. An example of a critical case is related to language. For example, if a letter is sent home in English to predominately Vietnamese-speaking homes and the parents understand the letter, then all parents of other languages and of English would be expected to understand the letter. A convenience case is one in which cases are selected based on availability. It is not purposeful or strategic in nature and is the weakest type of case. A typical case is one in which you would select to typify the norm when describing a group or program. A politically important case would be one in which cases are selected based on political sensitivity. For example, a case study of dual language programs in the state would be politically sensitive and informative in a year in which the legislature was considering funding or not funding dual language programs or other types of bilingual programs.
Your case data may include, but not be limited to: (a) basic demographic information about the individual that is written in narrative format, (b) family history (or if the case is about a program or organization, it would relate the program’s or organization’s history), (c) document analyses relating to the individuals or programs, (d) interview data, and/or (e) observational records. Once you have sufficient data, you will compile the case data that tells the story of the individual, program, or organization. You likely will have themes, which may be by type of phenomena revealed or from a chronological perspective. When you write, you will need to provide sufficient explanation with deep description, so nothing is left for the reader to surmise. If you have definitions specific to your study, be certain to explain those as well. Just as in a traditional dissertation, it is helpful to have a separate listing of definitions in qualitative dissertations as well in the Introduction Chapter. You must bring your reader right into the case itself. Perhaps if you visited the individual on location, you might take photos to place into your dissertation (see, for example, Christensen’s case study dissertation, 2003).
If you have multiple case studies, you will want to analyze each separately, then compare and contrast all of the cases for a final cross-case analysis (e.g., Bamberg’s case study dissertation, 2004). Of course, as in all qualitative research, you will need to report themes, categories, subcategories, or subthemes. Any anomalies found must be reported in detail as well. Data reduction, as it spirals from the more general to the specific, may include, as Miles and Huberman (1994) suggested, data matrices, tables, and figures to draw comparisons across cases. When you are writing your interpretation, you must establish the significance of the findings or themes, linking them to your theoretical framework. Implications for practice are also important in conveying the significance of your findings, but generalization to broader populations is typically not appropriate for this type of research. Even in writing your findings, you should be cautious in making what we call, “great leaps,” from the data to your implications. Your data represent exactly what they represent and nothing more, so make sure you have given the reader an appropriate amount of information in any claims that you make. Your findings must be credible as well as convincing and, again, based in a theoretical frame and previous literature and studies that are also credible. If you want to build theory itself from your case study, you will need to make certain you have a satisfactory number of case studies that will provide sufficient information to justify the development of a theory.
Examples of case studies conducted for dissertations follow. The first example is of a single case study conducted by Ozawa (2006). (See Example 5.3.)
The purpose of this single case study was to explore authentic communication for Japanese language learning in second-year Japanese classes at a small, private university in the Midwest. Types of authentic Japanese communication and materials in and outside of Japanese class were studied from four learners’ and one professor’s perspectives.
Data were collected throughout one academic year, the first semester of 2004 through the second semester of 2005. Multiple methods of data collection were used in this study including personal interviews, casual conversations, participant observations of classes and related events, and studying relevant documents, including the textbook, students’ study sheets, videos, oral exam transcripts, e-mail copies, reflection sheets, and web log copies.
Qualitative research procedures were used to study second-year Japanese language learners’ authentic communication. Data were analyzed by categorizing into codes, then themes and subthemes. Five themes emerged in this study: (1) the e-mail writing process, (2) the e-mail reading process, (3) the learning process, (4) learning through e-mail, and (5) authentic Japanese. Lastly, implications and recommendations based on the data were concluded. (Ozawa, 2006, n.p.)
The second example is a multiple case dissertation by Menconi (2006). (See Example 5.4.)
Example 5.4
Abstract
Looping is an instructional practice that allows the teacher to keep the same students over a two-year period. Multiage grouping is a more complex form of the multiyear configuration where the same teacher instructs students from two or three grade levels until they have completed the highest grade represented in that classroom. Both configurations present a logical approach toward developing long-term relationships which benefit students, teachers, parents, and administrators. Looping and multiage instruction beginnings go back to the one-room schoolhouse long before the Common School Reform Movement which introduced the graded school system of the 1860s. Even though the graded school system is over 150 years old, it still remains entrenched today. Consequently, looping and multiage instruction represent practices that require a paradigm shift from the graded school system.
While there are philosophical differences between looping and multiage instruction, there are also many similarities. Perhaps the most significant similarity is the long-term relationship that is developed between the child, teacher, and parent when the teacher has the same child for two years or two grade levels. This distinction develops many benefits such as curriculum coherence, safety, continuity, and student confidence. Two years with the same child also provides administrators and principals with many benefits. There is a tremendous savings of time because the students do not have to learn new routines and classroom management is easier. And there is a tremendous amount of professional development benefits for the teachers, since the teachers become better acquainted with the child and the curriculum.
Principals have indicated several concerns regarding the implementation of looping and multiage instruction, since it is a change from the standard graded school system. These concerns include the issues of more developmentally appropriate means for educating children, blending the curricula for two grade levels, and identifying staff members willing to undertake this paradigm shift.
The purpose of this study is to identify principal perspectives of the strengths and weaknesses of looping and multiage instruction from principals and lead teachers who have implemented them successfully. A multiple-case study design is used to investigate looping and multiage practices and provide a comparison of sites or cases so a range of generality or conditions is established. These sites include a wealthy, suburban elementary school using multiage practices, an integrated, middle-class, urban elementary school using looping, and a low-income, urban middle school using looping. Different information gathered from multiple sites and principal perspectives establishes a range of aspects or a conceptual framework that can assist principals in understanding these multiyear practices better. It is anticipated that this conceptual framework will be of value to those principals and administrators interested in implementing looping and multiage practices in their own schools or school districts. (Menconi, 2006, p. x)
ETHNOGRAPHIC RESEARCH DESIGN
Ethnographic research requires that you conduct fieldwork to become involved with the individuals or group in a personal manner, using participant observation as a technique for gathering data for telling the group’s or individual’s story via rich narrative description. You will typically gather the data via interviews during the participant observation, videography, photography, and document analysis. These techniques of data gathering will yield thick and rich descriptions necessary for your ethnographic dissertation in the form of quotations (low inference descriptors), descriptions of the group and the contexts, and parts of documents. You will, as in participant observation, investigate the behaviors of the people, their language, their actions, and their artifacts. You will look for norms, mores, and customs. Prior to going in to conduct such research, you must be clear about your own biases, about colonization, about who the “other” is, about your impact on the group to be studied, and about basic respect. To conduct fieldwork for the ethnography, you will need to gain the trust of the individuals in charge; these are the gatekeepers.
In ethnographic studies, depending on your purpose, you may use convenience sampling (as explained previously under case study method). You may use stratified sampling that seeks out groups from various levels such as socioeconomic levels. Snowball sampling may also be used where referrals from your initial contacts are made to add to the group.
Ethnography does not come without serious consideration. Some critical features to consider prior to the selection of an ethnographic design are: (a) your own understanding of culture and cultural anthropology, the foundation of ethnography; (b) your ability to write in a narrative style so that others may understand the cultural occurrences and norms of the group; (c) your ability to be a part of the group, yet remain apart from the group as the researcher, thus creating a fine line and balance between the researcher and the researched; (d) the ethical implications for studying the group or individuals; and (e) an extensive time commitment for the fieldwork.
Although conversation analysis was mentioned under the approach of discourse analysis within the phenomenological research method previously, we could consider conversation analysis as a separate approach and list it under ethnography and classify it as an ethnographic approach. In this type of analysis, one studies natural conversations in a variety of conventional settings (interviews, courts, schools, telephone conversation, restaurant, family conversations) to determine how the participants interact, construct conversation across time, identify problems, and/or exhibit gestures and eye contact.
We are placing narrative analysis within the larger context of the ethnographic research method, because according to Connelly and Clandinin (1990) narrative analysis is “how humans make meaning of experience by endlessly telling and retelling stories about themselves” (p. 4). Mishler (1986) frames the narrative through four categories: (a) orientation that describes the setting and character, (b) abstract that summarizes the events or incidents of the story, (c) complicating action that offers an evaluative commentary on events, conflicts and themes, and (d) resolution that describes the outcomes of the story or conflict (pp. 236–237).
Examples of ethnographic studies follow. The first ethnographic study comes from a master’s thesis by Quon (2006). (See Example 5.5.)
Example 5.5
Abstract
This study examines gendered patterns of participation in teens of classroom floor access and control with elementary school children. Primary participants in the study include approximately 97 students from four intact classrooms and 4 teachers. Eight students were also selected for observation.
Field notes from participant observation, informal interviews, and video/audio taped classroom interactions provided the data for this classroom ethnography. A portion of the data was analyzed in detail for time spent in effort and ratified speaking tone. The remainder of the data was analyzed through constant cross comparison for patterns of student talk participation in whole class interactions.
Findings revealed that success in talk is contingent on teacher organization of talk. Tightly organized floors rendered more equitable distribution of talk between boys and girls. Loosely organized floors rendered more differences in talk distribution in favor of boys. In all four classrooms, however, the girls are just as successful as boys in securing floor spaces, although they had to invest much more effort in hand raises. Data analysis of vocal children revealed that language proficiency can provide a context in which children gain self-confidence. Self-confidence allows the children to learn how to use language proficiency in order to gain status with other children.
The general improvement of women’s social status in the workforce and other contexts may help explain why gendered gap in participation is diminishing in the classrooms. (Quon, 2006, pp. 1–2)
The second example is from a doctoral dissertation by Thompson (2006), who employed an ethnographic approach. (See Example 5.6.)
Abstract
The purpose of this study was to describe and interpret the dimensions of performance creativity within a Montessori culture. Performance creativity is a reconceptualization of how creative activity is recognized and encompasses four elements: self, self-expression, meaning, and cultural significance. The conceptual framework for this study articulates creativity as a dynamic self-construction influenced by the meaning one draws from a cultural learning environment. The reconceptualization of creativity is built upon a frame of neuroscience research, Montessori philosophy, and selected Eastern/Asian, African, and Native American views of creativity.
This qualitative study was conducted in a Montessori primary and elementary school in the western United States, and it employed both the traditional ethnographic techniques of extended observations and artifact analysis. The study also used the performance ethnographic method of the collaborative dialogic interview.
The goal of performance ethnography is to produce a research-based performance narrative which views research participants as collaborators in the composition of the performance script. One of the facets of a performative study is that it initiates a dialogue within a larger community. The goal for this study is that it will invite educators to begin to reexamine how and why children express creativity and how creativity can be envisioned through a multicultural lens.
The findings of this work present emerging connections among neurological, Montessori, and multicultural perspectives of creativity that were observed in a Montessori learning environment. Educational implications related to this study include the potential for the fields of neuroscience and education to be drawn closer together, by using a multicultural avenue through which to describe creativity. The goal of designing curriculum and teacher education that is supported by neurological evidence is one that can be supported by further research in this area. (Thompson, 2006, n.p.)
McFadden (2006) used narrative analysis to conduct her dissertation study. (See Example 5.7.)
Abstract
Abundant research exists on the relationship that religion or spirituality has played in the lives and recovery of individuals diagnosed with physical ailments, while little research is found regarding the role that religion or spirituality has played in the lives of individuals diagnosed with a chronic or severe mental illness. Therefore, the researcher undertook to explore, investigate, understand, and describe the lived religious or spiritual experiences of these individuals, using a qualitative research inquiry and narrative analysis.
Data were collected during unstructured, informal interviews with seven adult volunteer participants. Data analysis of the interviews provided three major relationship themes common to all participants: with self, others, and God.
All participants acknowledge some form of religious or spiritual upbringing, and all report that their religious or spiritual practices improved once they were diagnosed. The participants identify a holistic approach to their recovery that incorporates their physical, mental, emotional, and spiritual well-being, and they further suggest a need for health care professionals to include a spiritual aspect in their treatment. Finally, all participants report a renewed sense of hope, meaning, and purpose for their lives, which they attribute to their religion and spirituality. (McFadden, 2006, p. iii)
GROUNDED THEORY RESEARCH DESIGN
Grounded theory, first described by Glaser and Strauss (1967), is intended to generate or discover a theory inductively from data gathered about a specific phenomenon. Three elements of grounded theory are concepts, categories, and propositions. Concepts are the basic units of analysis. Corbin and Strauss (1990) stated:
Theories can’t be built with actual incidents or activities as observed or reported; that is, from “raw data.” The incidents, events, happenings are taken as, or analysed as, potential indicators of phenomena, which are thereby given conceptual labels. If a respondent says to the researcher, “Each day I spread my activities over the morning, resting between shaving and bathing,” then the researcher might label this phenomenon as “pacing.” As the researcher encounters other incidents, and when after comparison to the first, they appear to resemble the same phenomena, then these, too, can be labelled as “pacing.” Only by comparing incidents and naming like phenomena with the same term can the theorist accumulate the basic units for theory. (p. 7)
Catagories is the second element of grounded theory and are defined by Corbin and Strauss (1990) as:
… higher in level and more abstract than the concepts they represent. They are generated through the same analytic process of making comparisons to highlight similarities and differences that is used to produce lower level concepts. Categories are the “cornerstones” of developing theory. They provide the means by which the theory can be integrated. We can show how the grouping of concepts forms categories by continuing with the example presented above. In addition to the concept of “pacing,” the analyst might generate the concepts of “self-medicating,” “resting,” and “watching one’s diet.” While coding, the analyst may note that, although these concepts are different in form, they seem to represent activities directed toward a similar process: keeping an illness under control. They could be grouped under a more abstract heading, the category: “Self Strategies for Controlling Illness.” (p. 7)
Propositions, the third category of grounded theory, indicate generalized relationships between a category and its concepts, as well as between discrete categories. As an iterative process, grounded theory is not generated a priori and then subsequently tested. Rather, it is,
… inductively derived from the study of the phenomenon it represents. That is, discovered, developed, and provisionally verified through systematic data collection and analysis of data pertaining to that phenomenon. Therefore, data collection, analysis, and theory should stand in reciprocal relationship with each other. One does not begin with a theory, then prove it. Rather, one begins with an area of study and what is relevant to that area is allowed to emerge. (Corbin & Strauss, 1990, p. 3)
Four research criteria are important in grounded theory for validation of the study and for establishing reliability of the findings of the study. Actually, these are recommended for review for all qualitative studies. They are: (a) construct validity, (b) internal validity, (c) external validity, and (d) reliability. Construct validity is accomplished by establishing clearly specified operational procedures. Internal validity can be reached by establishing causal relationships in which certain conditions are shown to lead to other conditions, as distinguished from false relationships, and it addresses the findings’ credibility or “truth value.” External validity is the extent to which the study’s findings can be generalized and to the extent that you, as the researcher, establish the context in which it can be generalized. Generalization is usually to some broader theory or other valid studies (quantitative and qualitative) and not the population. Finally, reliability requires demonstrating that the operations of a study—such as data collection procedures—can be repeated with the same results.
Triangulation is another method for ensuring that the study is robust, valid, and reliable. Triangulation may appear as four basic types: (a) data triangulation, involving time, space, and persons, (b) investigator triangulation, which consists of the use of multiple, rather than single researcher/observers, (c) theory triangulation, which consists of using more than one theoretical frame in the interpretation of the phenomenon, and (d) methodological triangulation, which involves using multiple methods. Multiple triangulation may be used when you combine in one dissertation, multiple observers, theoretical perspectives, sources of data, and methodologies.
You may also use reflexivity in the study to establish better credibility and trustworthiness.
Reflexivity requires an awareness of the researcher’s contribution to the construction of meanings throughout the research process, and an acknowledgment of the impossibility of remaining “outside of” one’s subject matter while conducting research. Reflexivity then urges us to explore the ways in which a researcher’s involvement with a particular study influences, acts upon and informs such research. (Nightingale &Cromby, 1999, p. 228)
Willig (2001) indicated:
There are two types of reflexivity: personal reflexivity and epistemological reflexivity. “Personal reflexivity” involves reflecting upon the ways in which our own values, experiences, interests, beliefs, political commitments, wider aims in life, and social identities have shaped the research. It also involves thinking about how the research may have affected and possibly changed us, as people and as researchers. “Epistemological reflexivity” requires us to engage with questions such as: How has the research question defined and limited what can be “found?” How has the design of the study and the method of analysis “constructed” the data and the findings? How could the research question have been investigated differently? To what extent would this have given rise to a different understanding of the phenomenon under investigation? Thus, epistemological reflexivity encourages us to reflect upon the assumptions (about the world, about knowledge) that we have made in the course of the research, and it helps us to think about the implications of such assumptions for the research and its findings. (p. 10)
Two examples of grounded theory methodology used in dissertation research follow. The first example is from Jodry (2001), and the second example is from Ericksen (2006). (See Examples 5.8 and 5.9.)
Abstract
Purpose. The overarching purpose of this study was to develop a grounded theory that can guide action in advancing the academic achievement for Hispanic students. Specifically, the study focused on the discovery of relationships and factors within the contexts of the home, school, and community that positively influenced the academic achievement of six academically able Hispanic students in an urban area, nonborder, high school advanced diploma program.
Methods. Following an extensive review of pertinent literature, I developed a conceptual framework of factors of support, motivation, and education within the contexts of the home, school, and community found to have an impact on Hispanic student achievement. The research questions focused on identifying the factors that were present in relationships within the home, school, and community that positively influenced the achievement of academically able Hispanic students in an urban, nonborder, high school advanced diploma program. Using ethnographic case study methodology, findings were applied to the development of a theory and an accompanying model titled, The Hispanic Academic Advancement Theory.
Results. From a research-based conceptual framework and data gathered in an ethnographic study, a grounded theory that can guide action in advancing the academic achievement of Hispanic students was developed. The Hispanic Academic Advancement Theory includes three factors, 18 subfactors, and 12 relationships. The three factors are: (a) supportive factors, (b) motivational factors, and (c) educational factors. The 18 subfactors are (a) positive communication, (b) positive adult relationships, (c) a climate of caring, (d) collaboration, (e) value for academic goals, (f) students viewed as assets, (g) advocacy orientation, (h) positive adult role models, (i) service learning, (j) safe environment, (k) high expectations, (l) programs, (m) development of self-advocacy, (n) respect for language and heritage, (o) culturally and linguistically responsive pedagogy, (p) shared culture and language, (q) parents as assets, and (r) culturally and linguistically responsive leadership. The 12 relationships are: (a) mother, (b) father, (c) older siblings, (d) grandparent, (e) aunts/uncles, (f) elementary bilingual educator, (g) secondary special area teacher, (h) secondary counselor, (i) principal, (j) peers in the advanced program, (k) college/university relationships, and (l) church relationships. (Jodry, 2001, pp. iv–v)
Example 5.9
Abstract
This dissertation utilized primarily qualitative methods of data collection and analysis to examine how 11 early elementary special education teachers in 10 schools in a large school district in the Mid-Atlantic region of the United States approached the task of developing and providing literacy instruction to students with significant disabilities who received most of their education in a self-contained setting. An environmental observation, semistructured interviews, and classroom observations were used to (1) describe the teachers’ existing practices related to early emergent reading, writing, and communication, (2) explore the extent to which teachers followed recommended best practices, and (3) to develop a beginning theory of how early elementary special education teachers approach early literacy learning given the many challenges of their students.
Six critical features were identified from the literature related to literacy instruction for students with significant disabilities. Those features were: (a) a responsive and supportive literacy rich environment, (b) integration of computers, assistive technology (AT) and alternative and augmentative communication (AAC) strategies, (c) direct instruction, (d) social engagement and meaning-making, (e) individualization based on local understanding, and (f) high expectations. Most of the teachers were actively engaged in teaching literacy skills to their students. In general they provided a responsive and supportive literacy environment and direct instruction. Computers were widely used as were AT and AAC; however, AT and AAC strategies were not systematically integrated across activities. Teachers had high hopes for their students but did not expect many of them would achieve conventional literacy. Emphases on meaning-making and social engagement were not observed, nor did teachers share these perspectives during interviews.
Using grounded theory methodology, a theory was constructed of five related concepts: (a) instructional outlook, (b) institutional expectation, (c) instructional set, (d) instructional fit, and (e) vigilance-adaptation. This theory was used to explain how teachers approach literacy instruction. It was also found useful for predicting poor instructional fit, which was defined as the match among the teacher, her students, and the instructional demands and supports in an instructional setting. (Ericksen, 2006, p. x)
MIXED METHODS RESEARCH DESIGN
Mixed methods research can refer to those studies that have engaged both quantitative and qualitative research questions and/or that have used both probability and purposeful sampling. It is a field of research that is still emerging.
According to Johnson and Onwuegbuzie (2004), mixed methods research is “the class of research where the researcher mixes or combines quantitative and qualitative research techniques, methods, approaches, concepts, or language into a single study” (p. 7). They further indicated:
Mixed methods research offers great promise for practicing researchers who would like to see methodologists describe and develop techniques that are closer to what researchers actually use in practice. Mixed methods research as the third research paradigm can also help bridge the schism between quantitative and qualitative research (Onwuegbuzie& Leech, 2004a). Methodological work on the mixed methods research paradigm can be seen in several recent books (Brewer & Hunter, 1989; Creswell, 2007; Greene, Caracelli, & Graham, 1989; Johnson & Christensen, 2004; Newman & Benz, 1998; Reichardt& Rallis, 1994; Tashakkori&Teddlie, 1998, 2003). (p. 5)
Johnson and Onwuegbuzie further iterated eight distinct steps in mixed methods design: (a) determine the research question, (b) determine whether a mixed design is appropriate, (c) select the mixed methods or mixed-model research design, (d) collect the data, (e) analyze the data, (f) interpret the data, (g) legitimate the data, and (h) draw conclusions (if warranted) and write the final report.
Examples of mixed methods research in dissertation studies follow. The first example is from Christopher (2006) and the second is from Wilbur (2005). (See Examples 5.10 and 5.11.)
Example 5.10
Abstract
Secondary English language learners (ELLs) spend most of their school day in classes with regular education teachers who have little or no training in methods for teaching English for Speakers of Other Languages (ESOL). This mixed methods (qualitative and quantitative) study investigated teacher use of accommodations and strategies for ELLs and assessed how the format of the resources made available to them affected their teaching. The participants were mainstream secondary teachers at the district-designated ESOL high school in an urban Midwestern city. The participating teachers were randomly divided into 2 groups and presented with a lesson plan addendum designed to help them increase their use of accommodations and strategies for ELLs. The first group received a CD version of the addendum which included Internet-linked resources explaining the strategies and accommodations. Participants in the second group received a paper version of the lesson plan addendum, and printed versions of the Internet-linked resources were available in the school media center. The researcher observed the teachers’ classes before and after receiving the lesson plan addendum. She recorded the strategies and accommodations demonstrated in their teaching, (before and after), in order to determine if the teachers in both groups changed their strategy use. Additionally, she compared the results from each group to ascertain if there was a difference in the frequency of strategy use between the two groups. Finally, the teachers were interviewed after the quantitative data were collected to determine if their self-perceptions about using strategies for ELLS aligned with the researcher’s observation of their teaching.
The results of this study showed that providing mainstream teachers with a lesson plan addendum focusing on English language learners does, in fact, lead to an increase in strategies and accommodations used. The results on the preferred way to provide the lesson plan addendum are inconclusive, although there is evidence that the resources linked via the Internet are more likely to be accessed than those that are made available in the media center at the school. Finally, based on interviews with the participants, the researcher concluded that they had a good understanding of how they changed their teaching as a result of using the lesson plan addendum. (Christopher, 2006, p. ii–iv)
Abstract
This dissertation focused on the application of a known organizational development strategy to facilitate the conversion of a vast administrative software system at an institution of higher education. Castle and Sir (2001) reported on how to use what they called the five I’s of instructing, informing, involving, intervening, and incenting to bring employees directly involved in the change through the adaptive work they needed to do during the process. In applying this 5 I strategy, I used as a platform the theoretical framework of Bennis, Benne, and Chin (1965), using people technologies during change in thing technologies to assist workers in adapting to the change. Using the participatory action research methodology of planning, acting, observing, and evaluating results, this study employed a mixed method approach to data collection, using both qualitative and quantitative methods, as defined by Johnson and Onwuegbuzie (2004). The purpose of the study was to discover the fears, needs and concerns of the staff involved in the conversion through the collection of field notes, journal entries, interviews, and surveys. I then applied the appropriate five I strategy to the change process to mitigate feelings of fear and loss during an imposed organizational change process. The impact of the action research project was the result of intentional interventions to mitigate feelings of fear and loss during an imposed organizational change process. This study also sought to engage the researcher in the practice of reflective leadership. A mixed methods approach to the data collection for this part of the study included taking field notes, journaling, and the administration of the Leadership Practices Inventory (LPI) of Kouzes and Posner (2003). The study showed a firstorder change at Orchard University as a result of the intentional insertion of organizational development strategies into the learning process. The study also showed how reflective leadership practice changed a leader’s self-perception and set the stage for individual leadership growth. (Wilbur, 2005, n.p.)
Your choice of methodology, whether it is qualitative or mixed methods, will certainly depend on your purpose and research questions. A mixed methods study offers a way to lend credibility to your study and triangulate your data while providing rigor to your study. Be sure to discuss this option with your dissertation chair or advisor.
SUMMARY
In this chapter, we provided an overview of several data analysis techniques or inquiry methods you could use if your questions or purpose necessitate a qualitative approach. We offered you four major categories of qualitative research: (a) phenomenological research, (b) case study research, (c) ethnographic research, and (d) grounded theory research. In addition to providing you with a brief overview of each category, we provided examples of research purposes or questions that justified a specific type of qualitative research design or method of inquiry. We also offered several other types of research in qualitative inquiry and briefly discussed mixed methods research, and provided some examples of it. In the next section, we share the typical approach to writing the dissertation chapters.
Lunenburg, F. (2008). Writing a Successful Thesis or Dissertation: Tips & Strategies for Students in the Social/Behavioral Sciences
