Research Methods and Processes (Encyclopedia of Management)
In any organization, managers at all levels need accurate and timely information for managerial decision making. Whether the decisions made are at technical, tactical, or strategic levels, good, accurate, and timely information always leads to a better decision. Gathering of information is done through a sound and scientific research process. Each year organizations spend enormous amounts of money for research and development in order to maintain their competitive edge. Accurate information obtained through research leads to enormous benefits.
APPLIED VERSUS PURE RESEARCH
Research can be defined as scientifically and methodically delving into the unknown in order to provide information for solving problems. The heart of this definition is the concept of problem solving. Both applied and pure (also known as basic) research attempt to solve problems. In applied research, the researcher attempts to solve a known problem and find answers to specific questions. In other words, the emphasis of applied research is on practical problem solving. For instance, when a paper recycling company wants to determine whether or not their recycled papers meet the required specification as to the thickness of the paper across the roll, they might design a systematic procedure for answering this specific question. The research in such a situation represents applied research. Another example of applied research might be that of prediction. As an example, consider a trucking company that is interested in predicting the tonnage of material shipped in the next quarter. The practical problem is predicting the tonnage and determining which variables are good predictors of tonnage for the next quarter.
Applied research can help make a decision about the following, including a variety of other business and management decisions:
- pricing a new product
- where to locate a new retail store
- how many employees to hire
- how many products to offer
- what to pay employees
Applied research can be used to collect information about markets, competitors, and customers. For example, research can help pinpoint the optimal business location and the size of markets. It can also be used to monitor competitive actions. Customer research determines customer loyalty, customer satisfaction, and customer preferences.
On the other hand, pure, or basic, research does not necessarily try to answer specific questions or solve specific problems. Pure or basic research is done in order to expand knowledge and probe into the unknown. For example, when a researcher is interested in determining how employee demographics and tenure on the job relate to preference for flexible work schedules may represent pure research. Both pure and applied research deals with problem definition and problem solving. Most basic research is conducted by professors in academic institutions (i.e. colleges and universities), by the government, or by consulting firms. Few business organizations will engage in pure research related to business problems. However, understanding the process and methods used for both applied and basic research are important to interpreting research results.
Any research involves several chronological steps, but that does not mean each step must be completed before the next step is undertaken. Furthermore, the process of research is dynamic and the process may change as the research progresses. The steps involved in most research endeavors are shown in Figure 1.
THE RESEARCH QUESTION
Managers' needs for information are the primary source of problem definition and the research question. Managers need information to make educated decisions arising from unanticipated as well as planned changes. As such, managers must select between different alternatives and thus require information about the organization and its environment. The question to be answered or the problem to be solved must first be clearly defined. Questions to be answered could be very specific or extremely broad. The more specific the questions, the easier it will be to answer the research questions. There might be hypotheses that could be tested scientifically. Once the questions to be answered are clearly defined then the value of the research must be assessed. Clearly, if the costs of performing the research project exceed the value that the research will provide, then the project should not be continued.
THE RESEARCH PROPOSAL
Research endeavors require a proposal that explains the problem to be address and the procedure by which the questions will be answered. The researcher's proposal tells the managers what they should expect from the research. It is a contract between the managers and the researcher. For instance, if a company wants to know the degree to which its new incentive program is effective in improving employee performance, then the consultant or employee conducting the research will create a proposal that indicates to that company how the question will be addressed and what specific information the company will have at the end of the research process. The proposal may indicate, for example, that the research will indicate the level of satisfaction of employees with the new incentive plan, the increased firm performance with the plan, and the individual increases in performance (as measured by managers) with the incentive plan. The purpose of the research proposal is to effectively guide the researchers in their development of the research design and data collection to answer the specific research questions.
Once the proposal is approved, the researcher has a foundation for development of the research design. The plan for conducting the research is the research design. There are two general forms of research design, namely non-experimental (ex-post-facto) and experimental. In a non-experimental design, the researcher does not control or alter any of the independent variables. The researcher merely studies existing situations, variables, and the interrelation among variables and reports the results of his or her findings. The two major non-experimental designs are field studies and surveys. Field studies combine literature review and possibly analysis of some case studies. For example, if one is interested in determining the effectiveness of total quality management (TQM), there will be a thorough literature search on the topic as well as a study of the firms that have applied TQM and have been successful. A literature review means that a researcher identifies previous writings and research on a topic, summarizes the current knowledge on the topic, and assesses the value of that prior research on the current problem. On the other hand, surveys deal with the formulation of a questionnaire (survey instrument) by which one can measure the magnitude of the desired variables as well as the interrelation among the variables. Non-experimental designs are primarily exploratory in nature and provide descriptive measures and can also be used for predictive purposes.
There are two broad categories of experimental designs: field and laboratory. In both field experiments and laboratory experiments, the researcher controls and may alter and introduce some variables in order to determine the effect of a given variable. Field experiments are done in a natural setting, whereas laboratory experiments are undertaken in a simulated setting. Studies on the effectiveness of different configurations of teams and their level of effectiveness can be undertaken in both field and office settings. In an office setting, a researcher might organize workplace teams, using different criteria to establish each, then measure the success of their group interactions and their productivity on real work tasks. This would be a natural setting, except for the way in which teams were organized. Team composition could also be studied in a laboratory in which the researchers had complete control over more variables. To study team effectiveness in a laboratory setting, individuals would be placed in teams using different criteria, then asked to perform a series of tasks specially designed to measure team interactions and performance. This laboratory setting would allow the researcher more control, because the types of individuals involved could be chosen, rather than using only the employees available in a field setting; by designing tasks specific to the study, rather than using existing work tasks; and by having more ability to watch and measure team performance without hindering organizational performance.
Data collection is the process of gathering the specific information used to answer the research questions. There are a number of issues associated with data collection, including the use of primary or secondary data, survey design, sampling, survey administration, and increasing response rates.
PRIMARY DATA AND SECONDARY DATA.
Data can be primary or secondary, and whether one or both are used, and which is used, depends largely on the research question and the availability of these data sources. Secondary data refer to data gathered by others or from other studies. Secondary data is generally less costly and less time consuming than gathering primary data, typically is accumulated before primary data is gathered, and may even help determine the course by which primary data is pursued. An example of secondary data is if a company uses data from the U.S. Census or data collected for another organizational activity (e.g., performance information for individuals from the company's annual performance appraisal). While secondary data can be used for background information about specific research, it may also answer some specific research questions. However, because secondary data was collected for another purpose, it may not adequately address the new research question. In today's world of rapidly growing information technologies, secondary data are available from numerous sources. A researcher should explore the existing data before starting the research process, since there are datasets for many different types of information currently available. There are abundant data available in literature, company records, government publications, trade associations, and through the Internet.
Primary data is that which is collected by the researcher to address the current research question. Types of primary data include subject demographics, lifestyle characteristics, attitudes, knowledge, intentions, motivations, and behavior. Demographic data includes statistics regarding populations, such as age, sex, income, level of education, and so forth. Lifestyle characteristics describe a respondent's activities, interests, and opinions. Attitudes refer to views and opinions about things, events, or ideas. Knowledge is the degree to which respondents are aware of these things, events, or ideas. Intentions generally refer to a respondent's planned future behavior. Motivations describe the reasons behind a respondent's behavior. Behavior is related to what respondents do.
Primary data can be collected in the field or the laboratory through communication and observation. Communication generally requires the direct questioning of respondents via a paper-and-pencil survey (i.e., questionnaire) or telephone survey. Observation involves the direct recording of respondent behavior. Surveys are probably the most common design in business research. For instance, if one is interested in determining the success of TQM, a survey can be designed that encompasses questions regarding elements of success, strengths, weaknesses, and other questions dealing with TQM. Then the survey can be sent to companies that have been successful in implementing TQM. The survey results could shed light on many aspects of TQM.
Survey design is of major importance, because is a survey is poorly designed, it will not provide the researchers with the data that addresses the research question. Survey questions, called items, must be properly chosen to in order to elicit appropriate respondent answers. The steps involved include determining the information that will be sought, the type of questionnaire, the method of administration, the content of individual questions, the form of response to each question, the wording of each question, the sequence of questions, the physical characteristics of the questionnaire, and, finally, pre-testing the questionnaire.
Some items for certain areas of interest already exist. For instance, there are existing surveys that measure employees' satisfaction with pay and benefits. If survey items do not already exist in the published literature, the researchers must create their own items, based on their review of the existing literature and their own expertise. Often, a focus group of experts can also help to create items. For example, if a company wants to assess its employees' attitudes towards an intended change in work rules, the researcher may lead a focus group of several experienced company managers to capture all of the relevant ideas that need to be addressed by the survey. Before the survey instrument is sent out, it must be tested for reliability and validity. Reliability refers to how consistently the instrument measures, and validity refers to whether the instrument is measuring.
One concern when designing a survey is how to word the items. One of the most popular ways to measure attitudes on a survey is by using the Likert scale. This method presents a series of statements to respondents for which they are asked to indicate the degree to which they agree with the statements. An example of a statement might be "The sales people are helpful." Respondents are asked to indicate the degree to which they agree with the statements by checking either SA (strongly agree), A (agree), N (neither agree nor disagree), D (disagree), or SD (strongly disagree). Respondents' answers would then be scored where SA = 5, A = 4, N = 3, D = 2, and SD = 1. A total score would be computed by average or summing scores on related items.
When administering a questionnaire there are two options as to who should complete the survey. Option one is to give the questionnaire to everyone in the targeted population. This is called a census. However, a census is usually not practical or cost effective. For instance, you may not be able to survey every one of your customers from last year to determine levels of customer satisfaction with your products. Consequently, in order to save time and money, only a sample or subset of the target population receives the questionnaire.
When selecting individuals for a sample, either a probability approach or a nonprobability approach can be used. Probability samples are those where each element of the population has a known probability of being selected. A random sample, for example, is the case where each element has the same probability of being selected. There are some specific types of nonprobability samples: convenience samples, judgment samples, and quota samples. Convenience samples are chosen at the convenience of the researcher. For example, a researcher might distribute a survey to all customers who enter one retail store in a one-week period to determine their level of customer satisfaction with the company's products. This sample is rather easy to select, but it may not represent the full range of customers who have used that product. In a judgment sample, individuals are selected by the researcher because they are believed to represent the population under study. Quota samples attempt to make the sample representative of the population under study where quotas are set for specific groups of people, which are generally selected on the basis of demographic characteristics.
The chief advantage of a probability sample over a nonprobability sample is the ability to assess the reliability and the amount of sampling error in the results. For example, if the goal were to estimate the annual household income for a given county, probability sampling would allow an accuracy assessment of the estimate. This could not be accomplished with a nonprobability sample.
After the survey has been designed and its reliability and validity assessed, the company must decide the administration method that it will use. Each administration method has its own advantages and disadvantages in terms of cost, information control, sampling control, and administrative control. Information control refers to the possible variation in responses to questions. Sampling control is the ability to select cooperative respondents. Administrative control refers to factors affecting the efficiency of the survey, including timing, quality control, and standardization.
Personal interviews are generally the most expensive means of data collection. In a company, this would mean having researchers meet with employees one-on-one to ask them the survey questions and record their responses. One of the main advantages of the personal interview is the ability to ask any type of question, including an open-ended question, and to adapt to the respondent's answers. However, in addition to being expensive and time consuming, this method is not anonymous, and therefore respondents may be reluctant to answer questions that they feel are sensitive or invasive.
The mail questionnaire is usually the least expensive method of data collection. Besides cost savings, another advantage of the mail questionnaire is its wide distribution potential. However, mail questionnaires cannot control the speed of responses, and the researcher cannot explain ambiguous questions. Mail questionnaires are probably best utilized when asking personal or sensitive questions, particularly if the survey can be made anonymous. Questionnaires can be circulated using various methods, such as post, electronic mail, and fax.
The telephone interview is associated with relatively low cost and higher response rates, and is one of the fastest methods of data collection. While there are methods to address the problem, unlisted numbers make it more difficult to obtain representative samples. Establishing rapport is also more difficult in telephone interviewing than in the personal interview.
One survey administration method that is growing in popularity is the Internet survey, in which respondents answer items on a survey that is located on a web site. Newer, specialized software products are making it easier to conduct online surveys, even for those people with little to no computer programming skills. Studies indicate that Internet research can result in faster responses, lower costs, higher response rates, and better flexibility. Additionally, this method aids in data administration, since survey responses can be directly inserted into a data spreadsheet by the web survey software.
INCREASING RESPONSE RATES.
One of the main concerns of survey research is the response rate, or the number of people who are asked to complete a survey who actually do. Nonresponse error is a source of bias because of the failure to get answers from some of the sample. "Not-at-homes" plague the telephone survey and uncooperative respondents affect telephone, mail, Internet, and personal interview surveys. While research results are mixed regarding effective means for increasing response rates, the following represent some ideas for increasing response rates:
- give respondents advance notice of the survey
- guarantee confidentiality or anonymity
- provide monetary incentives
- provide a postage-paid return envelope for mail surveys
- personalize outgoing envelopes
Research provides data, and it is the task of the researcher to transform the collected data into useful information for management. The first step in data analysis is preparing the data by editing it for several factors, including:
- completenesshecking for any omissions
- legibilityaking sure that handwriting is understandable so that answers will be coded correctly
- comprehensibilityaking sure the answer is understandable
- consistencyhecking for consistent answers from the respondent
- uniformityhecking to see that responses are recorded in the same manner
Once the data is edited it is ready for coding, which is determining how survey responses will be transformed into numerical data. The first step in coding is the development of a codebook. The codebook formalizes the coding process by listing answers and their accompanying codes. After the data is coded and entered into a data spreadsheet, statistical analyses can be performed to create useful information for the researchers. If there are hypotheses to be tested, the researcher is in a position to use the gathered data to test the hypotheses. Data analysis could be as simple as reporting descriptive statistics such as averages, measures of variability, and percentages, or if needed, advance statistical techniques could be applied.
The research report can be as simple as a short report of a few pages giving the overall findings of the research, or it can be a long report with numerous parts. The degree of formality required by management dictates the type of report to prepare. Figure 2 presents the order of inclusion of the various parts of a long formal report.
In this part of the report, first a title fly needs to be prepared. The title fly only includes the title of the report. The title should be carefully worded so it tells the reader exactly what the report is about. Following the title fly is the title page. The title page should include the title of the report, the name and the title of the recipient of the report, and the name and the title of the individual who prepared the report and the date. The letter that authorized the undertaking of the research project, followed by a letter of transmittal indicating the completion of the research report are the next items included in the report. Include a table of contents followed by an executive summary. The executive summary, summarizing the report's major findings, should be brief and to the point. This summary should briefly explain the conclusions.
INTRODUCTION TO THE RESEARCH.
This section of the report provides a clear background and statement of the research question and provides information about the objectives of the research. Included in this section would be a literature review about previous studies with the same or similar problem. If there are hypotheses to be tested, population parameters to be estimated, theories to be considered, they will be incorporated into this section of the report
This section will provide a detailed explanation of research design and will provide answers to many questions. What type of design was used? What instruments were used for the collection of data? Were there any subjects involved in the study? What did the subjects do? How was the sample selected? What kind of statistical or non-statistical techniques were used for data analysis? Finally, in this section of the report the limitations encountered in the study should be presented.
This section is probably one of the most important parts of the research report. Provided in this section would be the results of the data analyses and explanation of all the findings. At this point, all the raw data have been analyzed and converted to meaningful information for management's use. This is the section where the original research question is answered.
SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS.
A concise yet precise summary of major findings will be included in this section, followed by any recommendations that the researcher considers important and meaningful.
APPENDICES AND BIBLIOGRAPHY.
Statistical tests, large tables of information, copies of measurement instruments, and supporting documents should be included in the appendices. Finally, the report should end by providing a bibliography of all sources of information.
Babbie, Earl R. The Practice of Social Research. 10th ed. Belmont, CA: Thomson/Wadswoth, 2004.
Cooper, Donald R., and Pamela S. Schindler. Business Research Methods. 8th ed. New York: McGraw-Hill, 2003.
Davis, Duane. Business Research for Decision Making. 6th ed. South-Western College Publishing, 2005.
Hoover, Kenneth R., and Todd Donovan. The Elements of Social Scientific Thinking. 8th ed. Belmont, CA: Thomson/Wadswoth, 2004.
Kerlinger, Fred N., and Howard B. Lee. Foundations of Behavioral Research. 4th ed. Fort Worth, TX: Harcourt College Publishers, 2000.
Pedhazur, Elazur, and Llora Pedhazur Schmelkin. Measurement, Design, and Analysis. Hillsdale, NJ: Lawrence Erlbaum Associates, 1991.
Salkind, Neil J. Exploring Research. 6th ed. Upper Saddle River, NJ: Prentice Hall, 2006.
Schwab, Donald P. Research Methods for Organizational Studies. Mahwah, NJ: Lawrence Erlbaum Associates, 1999.