Selecting a topic for your thesis or applied research project should be one of the more deliberate choices you make in your graduate experience. The term papers and written assignments in your classes up to this point have been generally prescribed by the content of the course and ten week duration of the quarter. In contrast, the thesis/research project is defined primarily by your interests (within the telecommunications discipline of course), and is a relatively long term project of nine to twelve months (e.g., minimum of two semesters). Moreover, it involves exploring new information and developing skills that may serve you professionally for years to come. In selecting a topic, consider the following:
Pick a topic that is intriguing, stimulating, and so captures your interest that you can unerringly pursue it and immerse yourself in study for about a year.
Consider a topic that will require you to explore it in such depth that you will gain expertise in the subject. Such expertise can be used to develop a consulting practice in a particular area.
Select a topic that will enable you to be more effective on your present job--give you greater understanding, insight, or flexibility in dealing with the demands and problems that confront you in your work.
Explore a topic that will increase your promotability or at least the capacity for assuming new responsibilities and task assignments. These may help make your job more secure, enable you to establish a stronger power base, or provide leverage for needed change.
If you are considering further graduate education, such as a doctorate, use the thesis as an opportunity to explore a topic related to your educational direction.
Choose a topic that will increase your marketability and mobility for a job or career change in the near future. Consider where you would like to be in five years and what you would like to be doing, then work backwards to identify a topic that could become a stepping stone in the direction.
Identifying and stating the problem
The initial step in writing the final research project is to decide upon the question or problem that is driving your inquiry. Most often this is stated as a problem statement or research question, but may also be called the "thesis" or theme of your paper. It is most importantly used as the controlling idea to help you decide exactly what should and should not be included in the text. For example, if you find a stray article that is of personal interest and you are wondering whether to include it, ask, "does this clearly fit within the purpose and scope of my thesis?" If it does not--drop it! The thesis is much like the mission statement of an organization: It specifies "what business are we in--and NOT in." That clarity is also what you need to help sort through all the articles you will find.
The thesis should be specific enough to communicate to the reader (and clarify in your own thinking) the general direction and parameters of the study. For example:
What are predictors of success with respect to new technologies?
What are the barriers to competition in the telecommunications industry?
Notice that the statement usually identifies the major variables, although these may be subdivided into additional variables later. The statement or question also asks about the relationship between variables. Questions can be framed in terms of these relationships:
How could I describe the presence and frequency of certain variables in a specific setting (descriptive study using frequencies and percentages).
How are variable-1 and variable-2 related? (correlational study).
What is the effect of an intervention on a variable? (pre-post comparison study).
Is one method more effective than another to accomplish a goal? (comparative significance study).
How does a variable change over time? (time series study).
How do variable-1 and variable-2 interact under conditions A, B, and C? (multi-factorial study).
Closely related to identifying the research question is defining the variables. As soon as possible this should be done "operationally." That is, what operations or measurements must be done to explicitly define the variable? Remember that a "variable" is some component of the study that "varies" in some way--it is not a constant. Here are some examples of operationally defined variables:
Autocratic leadership is defined as verbal behaviors toward subordinates that are directive (specify action to take) and are non-negotiable.
Academic performance is defined as the cumulative grade point average for the last two years of college attendance at the time of graduation.
Default will be used in this study to mean the borrower does not have the interest on the loan repaid within two months of its due date.
It is important to define variables early so that you can more clearly pursue your literature search and can prevent going off on tangents related to other meanings. As your study progresses and you learn more about your variables, you may redefine them as necessary. The key idea here is to clearly define them for your use and the reader's understanding. Don't use more than one word to refer to a given variable. While synonyms makes for pleasant reading in prose, research requires the use of specific terms for specific reasons.
Literature review and your model
The purpose of the literature review is to collect relevant information to help you build your case for the study. In formal research, ideas do not come out of the air--they are derived from other ideas and based on previous research. In building your case you will need to cite studies that support your reasoning in having selected certain variables, proposed their relationships, and offered your hypotheses. The literature review commonly covers the findings of previous research, trends and themes, gaps and inconsistencies. It also provides an argument regarding the need for the current study.
In effect, you are providing information to lead the reader to understand why and how you have developed this study. You are building a "model" that proposes how something works, and your study will test that model against what you actually find with your data. Remember that you are NOT trying to "prove" anything--that's biased research. Even results that show "statistical significance" may not be "practically significant." You are trying to find whether your model does an adequate job accounting for the phenomena you are studying--not prove!
A model is simply a stated relationship between or among variables. Your research question, defined variables, and hypotheses are a verbal expression of that model. You might also think of a model as a diagram or flow chart of your variables with lines and arrows showing how they are connected to each other. A model is nothing more than your argument about how variables are related to each other.
Your hypotheses are statements that you offer to answer your research question. Each hypothesis usually specifies a single relationship between two clearly defined variables which will be tested. The hypotheses are really the key ideas in your model--everything else has just been foundation. Your hypotheses are the foci around which the rest of your study will develop. Here are some examples of hypotheses:
There is an inverse relationship between outsourcing and job satisfaction among telecommunications managers. (Interpretation: as a contractor even working for the same company and in the same cubicle, workers will feel less satisfied; as a "full fledged" member of the company, workers will feel more job satisfaction).
Post graduation college loan default are predicted by grade point average, timeliness of registration, incompletes, and level of co-curricular involvement. (This suggests a multiple regression in which default can be predicted by a combined weighting of each of the variables).
Managers who provide employee wellness programs will receive higher employee ratings on selected leadership qualities than managers who do not provide wellness programs.
Avoid the words "significant" or "significance" in your hypotheses since these terms refer to a test of statistical significance. Most readers will assume anyway that you will be reporting significant findings in the results section.
Also be cautious of the words "effect" and "influence" since these imply cause-effect relationships. For example: "the effects of an independent variable (e.g., interventions or treatments) on a dependent variable." Unless you plan to use a well controlled experimental or quasi-experimental design, choose a different wording.
Synthesis reflects the ability to construct new knowledge. It is not simply reviewing other studies and summarizing their results. Synthesis requires you to review previous information and then to construct your own answer, a new or different answer, that somehow extends and builds on what you have learned. This also is the nature of your model--to offer ideas how variables are connected, and how that connection describes or explains something.
Originality is more difficult to define. While your having an idea for the first time may be original for you, it may not be original in your field of study. It is the latter sense in which we are using originality. Your study does not have to produce a revolution in scientific thinking, a breakthrough in managerial practice, or be publishable in a top tier journal. It does need to be distinctive from previous research in some way. It might build on or extend a previous study. If you have reason to believe that another study was done poorly and different results might be obtained, then it might be reexamined. If you have a new twist to a procedure that might do something better, faster, or differently you could try it.
Related to originality is the relevance question. A criterion against which you might check your question is "So what?" Is this information already common knowledge? Has this topic been so thoroughly researched that my findings would be passe? You should seek some way to find an interesting angle to "hook" the reader's attention and interest. You will probably deal with this question when you write the section that argues "need for the study" or "relevance of the study." Be sure to ask this question early in the process so you won't collect vast amounts of information and end up throwing it all out for some "ho-hum issue." Here are some examples of dead (and dying) horses:
What are the stages of group development for decision making teams?
Does autocratic management lower morale?
Will support groups reduce stress during transition?
Can hygiene theory increase worker motivation?
One part of the problem with the above questions is that they are vague. The other part is that just about everyone familiar with the field knows their answers. If you were to pursue any of them you would need to find some aspect that had not been sufficiently explored. For example:
What are the effects of and democratic management among immigrant Indonesian workers? (This is a twist since Indonesian workers usually prefer autocratic styles. Although autocratic and democratic styles has been exhaustively researched for Western worker, there has been little done on other cultures. It could be legitimate to see how their immigrant experience in a democratic culture might change their preferences or create adjustment problems).
A comparison of self-paced computer tutorials in successfully getting across technical concepts, for an audience consisting of telecommunications managers with little or no technical background compated with those that have a technical background. (This could be an interesting comparison of interventions. First each group has been studied in isolation, there may not be comparisons among them).
Pick a topic that you think you can complete within six months to a year. You can always spend the rest of your professional life examining the big picture of your model. For this study, however, pick a manageable scope and examine it in detail. Be cautious of long term or time series studies that might require you to track a group over many months or even years.
Find a population to whom you have relative convenient access. For example, if you are interested in studying "Women Telecommunications Executives in Fortune 500 Companies" you are confronted with several potential problems: There are not many women yet in top managerial positions; many top level executives are too busy to respond to surveys, etc. Define a population and context to which you have more access, and where you might obtain more subjects if some decline to participate or drop out. Be prepared to gather demographic and other relevant data to adequately describe the sample to give readers a clear picture of who is being studied and represented.
Related to population is the number of subjects you plan to have in your study. In a single case study involving a work team of six people, you would not be dealing with a large "n" (number of subjects), although you might be using multiple sources of measurement or taking multiple single source measurements. The problems are these: If you have a small sample (generally <30) it can be difficult to obtain statistical significance, even when it's really there.
Small groups may be skewed (not normally distributed) thereby not enabling you to use parametric (normal curve) statistics.
Small groups may not allow you to fill all cells in cross-tabulations. This prevents you from even doing even simple comparisons with a chi-square.
Small groups produce scores which are notoriously unstable--that means the reliability will be low and you may not be able to depend on your scores from time to time.
A small "n" does not make your study impossible, you just may have to use a type of statistic designed for smaller n's--nonparametric statistics. These are also covered in SPSS, but it just means that you may not be able to generalize to a larger population, determine the magnitude or direction of the difference you are hypothesizing.
Archival (previously collected) data is acceptable for many types of studies. If you have reason to believe that currently collected data (e.g., census data, employee database, etc.) is complete, current, and relevant enough for your purposes you may use it. This technique is most commonly used in financial analyses, market demographics, and employee records. If you are examining a rapidly changing or unique situation, or are testing the effects of an intervention, archival data is probably not useful.
Consider designing your study for a worksite other than or in addition to your own. For example, the Human Subject Review Committee would challenge you to show that your in-house study of your subordinates would not place them at risk by disclosing their feelings about your management style. Although there are ways to increase anonymity of responses, this can become a sensitive issue and require elaborate controls. It may be easier to search for outside sources of subjects and not those with whom you have a questionable relationship.
When possible, use existing theories as a foundation for your model. Most current theories of management are still emerging, and may not even hold in a changing business or multicultural environment. There are usually interesting ways to extend or test a current theory that will make it easier for you to develop your support argument.
Consider operationally defining your variables in terms of currently available measuring devices and conventions. This does NOT mean to choose your instrument first, and then to decide on your topic--that's putting the cart before the horse. Pick your topic first, clarify your variables, and THEN search actively to find an instrument (e.g., tests and inventories) that best measures them. Using a previously constructed instrument saves much time and effort. There are databases and library search resources available for locating instruments, most of them have adequate information on validity and reliability, and are available at reasonable cost. If you construct your own instrument (to do it correctly), this requires a much more complex understanding of the domain you are studying, generating a large group of items, pretesting a pilot version of the instrument, and constructing a final version to use in the study.
Always keep dated backup copies of your drafts. You never know when a copy will be misplaced or lost, your computer will crash, or when you and your advisor will need to refer to the same document in a phone consultation. The computers in the department and lab are IBM compatible. If you use a different system you may not be able to share disks with your advisor or print pages or graphics.
The "design" of your study refers to the way in which you make comparisons among the variables and test the hypotheses. It also refers to the way in which you structure the inquiry so that you can confidently say whether there is a difference between certain variables or interventions. The idea of structuring is called "control." You should return to your statistics and research text (Campbell & Stanley's Experimental and Quasi-Experimental Research Designs is an excellent one) for a more thorough discussion, but basically control refers to any of the following:
Using a pre-test and post-test on a group of subjects to see whether an intervention (between pre- and post-) had a significant difference on some outcome.
Using a "control" group (who experienced no intervention at all) against which to compare an "experimental group (the one you tried an intervention on).
Using parallel (two different but equivalent) forms of a test to avoid "practice effect" of using the identical form twice.
Randomizing selection of subjects and their assignment to different groups in order to avoid bias. Randomization allows you to generalize your conclusion to a larger population.
The Human Subjects Review Committee (HSRC)
If the purpose of your project is to propose a model, or if your data is archival or does not involve human subjects, you will likely not need to submit a proposal for Human Subjects Review. Confer with your advisor regarding whether a proposal will be necessary.
Every college and university, and many organizations, have a HSRC to review ethical and safety factors potentially involved in a research project. This committee is staffed by experienced researchers who represent a variety of disciplines. Their role is to make certain that you have given adequate thought and action to avoiding physical, psychological, or social risks to subjects.
After your proposal has been accepted by the department faculty, you will complete a detailed proposal to the HSRC explaining your study, describing any potential risks, and how these risks will be managed. Forms are available through the Graduate School Office, and your advisor will assist you in completing them. You should keep subjects' safety in mind during your formulation and design of the study. Remember that "risk" is broadly considered to consist of embarrassment, social discomfort, abuse of power or even perceived intimidation by a supervisor, as well as physical injury.
The Human Subjects Review Committee will review your proposal first. If the proposal is in order, does not violate any of the conditions for waiver, and represents "good science" (e.g., meets criteria for a well designed study), it can be approved at the departmental level and does not need to be forwarded to the institutional level HSRC. The conditions for waiver are available from the Management Department and your advisor.
Other considerations
Date all materials (drafts, FAXes, notes, etc.) to avoid miscommunication due to mistaken documents.
Make at least two and perhaps three backups of your data and written documents. Keep titles and labels clearly sequences and dated, and consider using a master record of all documents that briefly list titles, dates, and brief comments on contents. Of all the times you need to be compulsive, this project demands it the most. Regularly backup your documents and statistics files while you are working with them.
Use your peer support group and other support resources (e.g., family, friends, advisor, other faculty, etc.). Share your excitements, frustrations, questions, resources, and ideas freely.
Make a clear decision regarding the priority of this project compared with your other obligations, and negotiate these decisions with others involved (e.g., employer, family, sports teams, etc.). There is no need to experience problematic guilt and frustration. Pace your efforts so that you can adhere to your timetable. If you get off schedule, question what is needed to get back on track. Use your advisor or support group to explore distractions or problems.
Become very familiar with your word processor (but DON'T practice on your original document).Purchase additional books on using its special features that make it work faster and more conveniently. Practice on unimportant documents to mastery before you risk your project document. The time you invest in this practice usually pays off in your time on writing and formatting the document. A version of the APA style format is also available for Microsoft Word software.
Always come prepared for a phone conversation or meeting with your advisor by having all materials with you and having at least three questions to which you need answers.
The faculty are committed to turning around manuscripts within two weeks of receiving them (unless delayed due to unforeseen or previously scheduled events). If you do not have a rough draft of your final manuscript in to your advisor by the beginning of March, there is no guarantee that you will be completed by May (however, you may decide to graduate during a later quarter).
Recognize that this kind of project involves multiple revisions of the manuscript. From nine to a dozen or more is quite common. This is not nitpicking, but a necessary process in professional and technical writing to make certain that the final document is consistent, clear, and accurate for the reader. In some cases, restrictions in methodology may require you to go back and make modifications in the literature review, either adding or dropping some arguments. In the end, most writers agree that the final document is a better product.
Last point: Expect the unexpected!
The following resources are recommended as background reading as informative guides:
1. STATCON: The expert system statistical consultant. This is PC-based public domain freeware and can be distributed to you. It can be supplemented by a copy of a statistical decision tree.
2. Alreck, P. A., & Settle, R. B. (1985). The survey research handbook. Homewood, IL: Irwin. 427 pp. (ISBN: 0-256-03174-6 paper). Cost circa $28. Excellent text on planning and design of surveys, instrumentation, data collection and analysis. May not be as appropriate if your methodology is unrelated to the survey approach.
3. Isaac, S., & Michael, W. B. (1971). Handbook in research and evaluation. San Diego, CA: Robert R. Knapp. Cost circa $17. Excellent overview of basic concepts, definitions, and brief examples of statistical designs and terms. Recommended text for all types of projects.
4. American Psychological Association. (1994). Publication manual of the American Association (4th ed.), Washington, DC: Author. This is the required reference for writing your project.
Writing is a reflection of the writer's thinking style; one's mastery of concepts, use of logical argument, and ability to bring together diverse ideas. The latter aspect of synthesis involves constructing new knowledge by assembling separate ideas into an original form.
For example, the following separate concepts might be joined in various ways:
Hersey and Blanchard's managerial style and Tuckman's stages of group development may be combined into a model that describes how style might vary with developmental stage.
French and Raven's three types of power might be combined with Argyris' concept of worker maturity in a way that applies varieties of influence under different conditions of worker initiative.
Schein's process consultation and Likert's four systems might be compared to determine the conditions under which process consultation is optimized.
The ideas for such research topics can be initially stimulated by either concepts or applications. Many writers are intrigued by an idea, concept, or principle in management theory. They begin with the abstraction and then work toward an application of it. For example, you might be intrigued by literature on organization conflict, its hazards and benefits. Such interests might lead to exploring the practical question of how conflict might be constructively used in department restructuring in the health industry. Alternately, other writers start with a practical problem and seek to find a theoretical explanation to better account for it. For example, you might be puzzled why a manager seems to vary his/her behavior in successfully dealing with different employees. This direct observation might lead to a study of Fiedler's leadership style contingencies. In both cases, however, the process of deriving a researchable topic and arguable theme may be less orderly than expected.
Research writing involves two stages: the disorderly before the orderly. The initial stages of formulating a working idea for a paper is often convoluted. One idea may lead to another, leading to still more branching ideas and references. Both divergent and convergent thinking are involved in focusing on a topic, finding multiple sources of information, going back and refining the topic, and resuming the search for more specific information. Pages of notes and complexly connected flow charts accumulate. This first stage of writing is circuitous, complex, and often confusing.
The transition to the second stage of writing emerges (eventually) when a clear purpose for the paper and conclusion are finally identified. The writer can retrace the previous steps and identify the most important and relevant segments of the argument. These segments are then presented in a sequential, logical, and orderly fashion (see Figure 2). The main idea is to inform the reader of the purpose of the paper and then to systematically use writing cues to lead the reader to the conclusion. The writing cues that guide the reader include clear headers and subheaders, well stated research question or problem statement, thesis sentences at the beginning of paragraphs, consistent use of key terms that are clearly defined, and transitional sentences between paragraphs.
The paragraphs are the building blocks of the argument. The first sentence should contain the thesis sentence about what the paragraph is about. The sentences in the paragraph should support and elaborate on the idea in the thesis sentence. In addition, the sentences should logically be related to each other. When uncertain whether to add a sentence to a paragraph ask, "how is this related to the paragraph's thesis sentence?" If it is not related--drop it.
In a similar manner, the paragraphs should be sequenced so that they systematically elaborate on and provide support for the header of a section. The paragraphs should have transitional sentences so that the reader is gently guided from key idea and support to the next key idea and support.
The argument in the paper is composed of your assertion of key ideas which are sequenced to build a case leading to the conclusion. Supportive studies, quotes and citations should be used to document your argument, but do not let the other authors speak for you. Quotations should be used sparingly, usually only when an author says something more eloquently or concisely than you ca. Minimize citing other authors and the details of their studies directly (e.g., "Smith and Wesson (1993) found that..."), since this distracts the reader from your argument. It is sufficient to cite them as a source showing their contribution in part to your synthesis of ideas. Since most synthesis writing combines ideas from several sources, citations often include multiple authors (e.g., Adams, 1992; Cronbach, 1990; Petrich, 1989).
Remember that writing is recursive or nonlinear. It is comprised more of revising than just writing linearly, step by step. Take time to ask the following questions:
Does the reader know from the beginning precisely what the paper is about and what to expect?
Are there clear landmarks (e.g., headers, thesis sentences, etc.) To guide the reader to the conclusion?
Are the contents within the paper, sections, and paragraphs appropriately limited to what is implied by the thesis statement?
Are there clear transitions between sections, paragraphs, and sentences so that it reads smoothly?
Are the key ideas in the argument clearly presented and supported by citations?
A final check might be to put it aside for a few days and then return
to review it refreshed. Alternately, you might read it into a tape recorder
and listen to it, or pass it on to a friend for critical reading and feedback.