1.7 Research Procedures in Organizational Behavior

The term "correlation" refers to the relation between two variablesAn element, feature, or factor that is liable to vary or change. or attributes. Most attributes of interest to OB researchers vary or change, hence the term "variable." Leadership traits, productivity, communication, and motivation are all variables.

Virtually all correlations are calculated through the use of a mathematical procedure devised by Karl Pearson. The full name of such a correlation is Pearson product-moment correlation, abbreviated as r, and ranging from 0 (indicating no correlation whatsoever) to 1 (indicating a perfect correlation). The numbers between 0 and 1 are termed correlation coefficientsA number between −1 and +1 calculated so as to represent the linear dependence of two variables or sets of data. where the higher the number, the stronger the relationship. A minus sign in front of the coefficient means that as one variable goes up, the other decreases.

The higher the correlation between two variables, the more accurately we can predict the amount of one variable from knowing the level of the other. For instance, if we knew the amount of training given to employees and their motivation level, and if the training and motivation had a correlation of 1, then if I knew your training history I'd know your motivation level too.

Correlational researchThis allows us to state how two variables are related mathematically, but not why they are related. allows us to state how two variables are related mathematically, but not why they are related. For instance, if we know that there's a high correlation between the amount of praise given to employees and their productivity and — the most productive are the most praised, and the least praised are the least productive and — then from knowing how much praise a worker receives, we can predict that person's productivity. What we cannot do is determine the reason for this correlation, since there is more than one possible explanation. Maybe the praise caused the high output (or vice versa), or perhaps some third variable, such as the employee's attitude, was responsible for both the level of output and the level of praise. We are entitled to guess as to the reason, but not to conclude. Examples of correlation research that are popular among OB researchers include field studies, surveys, archival research, factor analysis, longitudinal studies, and case studies.

The surveyThe most common procedure for collecting data in OB research. It involves questionnaires and interviews. is the most common procedure for collecting data in OB research. Surveys that involve written responses are called questionnaires; those that involve spoken responses are called interviews. Each of these survey types has advantages and disadvantages.

• The questionnaire is the most cost-efficient means of surveying a large group of individuals. It is inexpensive to draft and to type a page of questions, and the page can be reproduced, distributed, and collected efficiently.

• The procedure is standardized. The same questions are presented in the same way to each respondent, and all respondents answer in the same way on the same answer form.

• A variable that can be tapped in this standardized fashion is writing ability. For instance, a question could say, "In the space provided, explain why we should offer you the position you are seeking."

• The questionnaire is inflexible. Some questions do not lend themselves to a "yes or no" answer, or information that the respondent wants to or should ask may not appear. Of course, skillful design of the instrument can reduce such problems.

• Questionnaires are subject to demand characteristics. That is, no matter how neutral or non-leading the questions were designed to be, the respondent is prone to interpret (from the wording, from the effect of previous questions, from the title of the document, etc.) what the "right" or "best" or "expected" answer is.

• Because of the ease with which questionnaires can be written, reproduced, administered, and collected, they invite carelessness in their design and administration. It is easy to collect more information than can be properly processed, interpreted, or used.

• Only a small percentage of those who receive a questionnaire are likely to complete and return it. The problem here is not one of numbers. If a researcher wants 100 completed questionnaires and the response rate is only 10 percent, it is still cost- effective to administer 1000 forms to yield the required 100 answers. The problem is non-responder biasThe systematic difference between responders and non-responders., which is the systematic difference between responders and non-responders. For instance, people opposed to an issue are more likely to return a questionnaire dealing with it than are those in favour of or neutral toward the issue. Thus, responders may not be representative of the population surveyed.

• It is relatively easy to amass an impressively large array of data from questionnaires. However, achieving a high response rate does not eliminate problems. Respondents may answer carelessly or even randomly. This pattern is hard to detect and counteract. Some questionnaires contain repeated (reworded) questions throughout, so that internal consistency may be measured.

• No matter how clearly the researcher tries to word the questions, there remains the chance of misinterpretation.

Whereas correlation research reveals the extent to which two variables are related, factor analysisA statistical procedure that shows the extent to which any number of variables are related. is a statistical procedure that shows the extent to which any number of variables are related. Factor analysis also shows how much "clustering" there is in a group of variables by revealing which variables are linked into groups. That is, it tells which variables are so strongly correlated with each other (and only weakly correlated with the other variables) that they represent some common trait, ability, or factor. For instance, suppose you obtain measurements on these five variables in a group of employees: performance, motivation, incentive, health, and age. Suppose that there's a strong connection among the first three variables, and that these variables are not related to the last two variables on the list. Factor analysis would show the extent to which performance, motivation, and incentive form a single cluster, sometimes also called a "loading."

As statistical methodologies become more sophisticated, organizational behavior researchers are incorporating methods in their work such as Structural Equation Modeling (SEM) and Multi-level analysis.

The simplest type of scientific research is the field studyA study performed when the individuals being observed are unaware of the researcher., also known as naturalistic observationSee field study. These terms stem from the early history of animal psychology, when some researchers argued that because bringing animals into a laboratory for study would disrupt their normal behavior, scientists should observe the animals in the natural environment.

Field studies are routinely chosen for OB research when the objective is to find the correlation between two variables, such as the quality of the cafeteria food and worker contentment. By definition, when a field study is performed, the individuals being observed must be unaware of the researcher, a requirement that raises ethical issues when the subjects are human beings.

One of the most popular qualitative designs is the use of interviews. OB researchers have used interviews to study organizations at all three levels.

Advantages of the Interview

• In contrast with the questionnaire, an interview yields high response rates because it is harder to ignore a survey-taker than a piece of paper. When potential respondents are not available, the interviewer can return or make contact by telephone.

• In most face-to-face encounters, the interviewer can assess the respondent's confidence, attitude, and anxiety level. Similarly, the interviewer can note physical characteristics such as dress, grooming, and posture.

• By establishing a rapport with the respondent, the interviewer can elicit richer and more complete answers than would be possible in the context of a questionnaire.

• The interview can be highly flexible. Although flexibility is minimal in the structured interview, which involves reading the same questions to each respondent, the interviewer can probe (i.e., ask for clarification or expansion) in the semi-structured interview. In the unstructured version, the interviewer may ask anything in any order.

Disadvantages of the Interview

• Even a structured interview lacks standardization, because it is hard to remain unchanged in manner, voice, and appearance over a period of time. An interviewer may start each interview in exactly the same way, but each response of the interviewee is likely to change the behavior of the interviewer.

• A more serious criticism is subjectivity. For instance, when one person meets another, an impression, perhaps having to do with physical appearance, tends to be formed immediately, and it resists change.

• A selection interview is susceptible to the "first date" syndrome, wherein each party tries to impress the other by exhibiting uncharacteristic behaviors.

• The interview is an expensive procedure for collecting data. To gather 1000 hours of responses requires 1000 hours of interviewing, yet 1000 respondents could complete a questionnaire at a cost of a few hours of research time.

• Respondents tend to be more nervous in a face-to-face encounter than in a paper-and- pencil encounter, especially if the interview is seen as important. A selection interview favors the applicant who is verbally articulate and skilled in self-promotion. When the interview is designed to gather research information, the respondent may feel intruded upon, self-conscious, or concerned about the lack of anonymity; such feelings can distort the responses.

• The prime liability of the selection interview is that success in the interview, a highly controlled situation, is not a reliable predictor of success on the job. It's likely that you have been asked to complete various surveys. And if you continue in OB, you may be involved in designing, administering, and interpreting surveys. No other research technique is as useful, and no other technique has as compelling a list of both advantages and weaknesses.