Goubil-Gambrell, P.: A Practitioner's Guide to Research Methods

Goubil-Gambrell, Patricia. “A Practitioner''s Guide to Research Methods.” Technical Communication 39.4 (1992): 582-91. Print.


Goubil-Gambrell''s premise: members of the field know that research is important, but they don''t know how to evaluate it. Names two objectives for article: "first, to identify the main types of methodology and business and technical communication research, and second, to help technicall communication practitioners understand the distinguishing characteristics of these methodologies" (583). Argues major problem in reading research is "overlooking the methodology and concentrating on the conclusions" (583). Cites MacNealy--those who don''t include info as to methodology damage ethos.

 

Guide for Analyzing Research Methodology

 

1. The Problem

What question is the researcher trying to answer?
2. The Data

What kinds of data need to be collected to answer the question?

What kind of data did the researcher collect?  Was it the type that will be needed to answer the question?

3. The Subjects

Who were the subjects for the research?  How many subjects were there?

What are their demographic characteristics?

What background do they have that makes them appropriate subjects for this research?

How were these subjects recruited to participate in this research? Were they volunteers?  Were they paid?

4. The Methods

What were the subjects asked to do that produced the data that will provide an answer for the research question?

Did the task seem appropriately related to the research question being asked?

5. The Analysis

How were the data analyzed?

Were appropriate statistical techniques used?

If data were to be interpreted by raters, how many raters were there?  How were they trained?  What criteria did they use in interpreting the data?

6. The Conclusions

How much does the researcher generalize in the conclusions?

Did the researcher generalize beyond the bounds that are appropriate for the research method used?

  • Quantitative research can establish cause-and-effect relationships
  • Qualitative research can generalize only about the particular subjects that were studied and can make an argument that they are representative of others with similar characteristics.

 

Types of Research on Writing:

 

Two empirical methodologies: Quantitative (attempts to quantify key aspects [variables] and relate them; manipulates variables [called treatment] then analyzes results. Purpose is to determine cause and effect) and Qualitative (observational and non-obtrusive). Two other approaches: Scholarly Inquiry (more of a literary approach; analyzes writing from various perspectives [e.g., historical, critical]) and Practioner Inquiry (essentially "how I solved this." "Experience-based testimony" [584]).Issue: How to measure/ analyze the results. Qualitative is situational and quantitative unnatural. Difficult to generalize results.

 

Characteristics of Quantitative Research (585):

 

Experimental: 1. Random sampling. Random allows results to be generalized. Samples can also be stratified (e.g., by gender); 2. introduction of a treatment; 3. use of control group.

 

Quasi-Experimental: Subjects not randomized (e.g., students in one classroom). To establish comparable groups, research should develop a pretest (background info or an actual questionnaire/test)

 

Hypotheses: Must have. Null hypothesis = no difference between control and experimental groups. If a difference, -> cause and effect statement.

 

Evaluating hypothesis: should be conceptually clear, with concepts defined operationally, if possible; should have empirical referents/ no value judgments; specific so can be determined if testable; related to available testing techniques; related to a body of theory.

 

Statistics: Used to determine whether measurable/significant differences between/among groups on variable; differences not by chance; determine/measure differences between variables.

 

Two kinds: descriptive (present data in orderly fashion; commonly frequency distributions like mean, mode, average); inferential (allow researcher to infer that relationships exist among variables-- ex. chi-square, t-test, F-test).

 

Stastitical view of causes: 1. May occur in sequence to produce effect; 2. several causes may converge to produce effect; 3. effects of single cause may affect many areas; 4. preceding three may all create a complex network of causes for an effect (586; qtng. Miller). Identify independent (the treatment; the activity that makes a difference in the outcome) vs. dependent variable (the effect; the change thanks to independent variable).

 

Validity (does it measure what it says it will measure?): Internal (difference in dependent variable actually a result of independent variable) and External (results can be generalized to other groups)

 

Reliability: "whether the experiment precisely measures a single dimension of human ability" (587).

 

Problems with Quant Research: Isolated variables in structured situation. Not like real life. May eliminate other variables that would affect results. Unnatural, and not always possible to actually conduct quant study. Notes that sex, socioecon status, ethnic background, or previous experiences cannot be randomly assigned.

 

Characteristics of Qualitative Research (587):

 

Case Study: Researches particular individual/small group.

 

Ethnographic Study: whole environment in which individual or group of individuals function as communicators

 

Subjects: Purposeful sampling (as opposed to Random). Examples:

 

Extreme case sampling: subjects are unusual or special in some ways (novices and experienced writers)

 

Intensity sampling: "those who manifest phenomenon of interest intensely but not extremely"-- e.g., skilled writers but not the very best.

 

Maximum variation sampling: focuses on common patterns/ core experiences (e.g., what do both novices and skilled writers experience?)

 

Data Collection and Analysis: Identify salient features or variables. Researcher needs full disclosure: method, setting,data-collection techniques, data-analysis strategies. Researcher usually participant-observer; should indicate role in setting (it could influence). Main method of working with data is triangulation:

 

Data triangluation: comparing and cross-checking consistency of information gathered at different times with the same method (interviews at beginning, middle, end)

 

Methods triangulation: Check consistency of info gathered by different means-- field notes, interviews, documents produced in setting

 

Researcher triangulation: more than one researcher interprets data

 

Theory triangulation: look at same data from different perspective

 

Judging a Qualitative Study (Marshall and Rossman): 1. Data collection methods are explicit; 2. data are used to document analytic constructs; 3. negative instances of findings are displayed and accounted for; 4. biases are discussed; 5. strategies for data collection and analysis are made public; 6. field decisions altering strategies or substantive focus documented; 7. competing hypothesis presented and discussed;  8. data preserved; 9. participants'' truthfulness assessed; 10. theoretical significance and generalizability made explicit.

 

Advantages and Disadvantes of Qual research: Reflects reality; cannot generalize findings. Not cause and effect, but co-occurrences.

 

Other Research Methods: writing studies often hybrids of qual and quant.

 

Developing Methodological Literacy: Both forms have merit and purpose. Goubil-Gambrell, however, provides guidelines for evaluating qual methods:

 

Credibility: demonstrate that study conducted accurately identifying/describing subject and parameters

 

Transferability: (helped by use of theoretical framework and triangulation of data)

 

Dependability: researcher "attempts to account for changing conditions in the phenomenon chosen in the study" (? qtd. in Goubil-Gambrell 590)

 

Confirmability: Findings are objective. Document methodology, use more than one researcher to record/interpret data, archive data so it can be re-analyzed.

\