Research Methods in Organizational Behavior

Abstract

Page 87 RESEARCH DESIGN 6 Page 91 BASIC RESEARCH DESIGN Page 92 Experimental Designs O1 X O2 Time 1 History Events that occurred between O1 and O2 other than X which may have been responsible for the observed differences between O1 and O2. 2 Maturation These are effects that occur systematically with the passage of time. 3 Testing: Observed differences between O1 and O2 may be a function of simply being exposed to a measure at two periods in time. 4 Instrumentation: Here the ability of our measure to accurately index the measured variable changes over time. 5 Statistical Regression: If individuals are selected for a treatment because of extreme scores on a pretest that is less than perfectly reliable, it is likely that those with extreme observed scores actually have true scores that are nearer to the mean of the score distribution for all measured subjects. 6 Selection: If post-test scores for a group that has received some treatment differ from scores on the same variable for a group that has not been treated, the inequality may be due to initial effects of the treatment. 7 Mortality: Here, observed O1-O2 differences are the results of individuals dropping out of a study between pretest and post-test periods. 8 Interactive Effects: It is possible that two or more of the above-mentioned phenomena may be responsible for O1-O2 differences. The eight factors just mentioned are known as threats to the internal validity of an experiment. An experiment is internally valid when we as researchers are able to conclude that an experimental treatment has indeed had an effect. Page 105 Nonexperimental Designs In ex post facto research the investigator has virtually no control over the study's independent variables. Nonexperimental research generally takes two forms: (a) the cross-sectional study and (b) the correlational study. Cross-sectional Studies In cross-section research the investigator compares scores on the study's dependent variable(s) for groups that the researcher assumes have been differentially exposed to the study's independent variable(s). Page 105 Inherent all such studies is the problem of lack of control over independent variables. As a consequence, statements about causal relationships between independent and dependent variables cannot be safely made. Correlational Studies the investigator obtains data on the study's independent and dependent variables. These data are then used to assess strength of relationship between the two variables. It should be obvious from these two examples that correlation alone cannot be used to support arguments of causality. Page 108 The researcher who uses an ex post facto or nonexperimental design must be in a position to rule out numerous rival hypotheses for any relationship he or she uncovers. Data on potential confounding or nuisance variables are critical in ruling out such competing hypothesis Note, however, that in the case of ex post facto research we are only able to rule out competing hypotheses associated with potentially confounding variables that weve measuredIn sum, nonexperimental studies leave us in the uncomfortable position of not being able to state with any degree of certainty whether two variables are causally related to one another or whether an observed relationship between two variables is a legitimate one (ie, not spurious). In spite of the weaknesses of nonexperimental research, this type of study strategy has been and continues to be widely used in organization researchFirst, ex post facto studies may result in hypotheses that can be tested in a more rigourous fashion (ie, quasi-experimental and experimental research). Second, the findings of experimental studies are made more credible to the extent that they can be corroborated in less rigourously controlled field research. Thirdin many instances we cannot manipulate independent variables. If we relied exclusively on experimental and quasi-experimental research for generating knowledge, numerous important relationships uncovered in the behavioural, biological, and physical sciences would not be a part of the current body of knowledge. INTERNAL AND EXTERNAL VALIDITY in considering whether or not a study has internal validity one is, in effect, asking the question, "Did the study's treatment variable have the hypothesised effect?" An equally important consideration is external validity. A given study has external validity to the extent that its results are generalisable beyond measures, subjects, and other conditions associated with the study. Page 111 EMPIRICAL RESEARCH STRATEGIES Page 128 Field Studies Page 129 Description of the Strategy: The field study strategy can be characterised in the following way: (a) the research is ex post facto in nature, no independent variables are manipulated by the researcher; (b) intact, naturally occurring systems are the object of study; (c) variables are systematically measured; (d) the investigator attempts to minimise his or her intrusion upon the system being studied; and (e) the focus of such research may be exploratory, descriptive or hypothesis testing. Page 131 The field study strategy has a number of advantages, including: a Field studies are high on realism b The intrusiveness of this strategy is relatively low c Data on a large number of variables can be obtained from subjects; d Complex phenomena can be studied e Socially significant (ie, application-oriented) problems are amenable to study f Field studies are often valuable for their heuristic qualities; and g The strengths of independent, intervening, and dependent variables are generally greater than would be found in a laboratory experiment. Disadvantages a The co-operation of organisations is often difficult to obtain b Causal inferences from field study dataare highly tenuous c Data from most field studies are likely to contain unknown sampling biases d limits on the number of variables that can be measured in a field study e Independent variables are not manipulated in field studies f Measurement is not as precise as it is in the laboratory g expensive to conduct h The "dross rate" or proportion of irrelevant data yielded by measures may be high Page 135 The Case Study Page 136 Description of the Strategy The case study approach to research can be described as follows: (a) the researcher intensely examines a single unit (eg, a person, group, or organisation); (b) data are often collected by multiple means; (c) no attempt is made to exercise experimental or statistical controls; (d) phenomena are studied in natural settings; and (e) the strategy is suited more to the generation of hypotheses than their testing. Page 137 Advantages a The full complexity of the unit under study can be taken into consideration b Data collection is flexible c The case study is a useful vehicle for the generation of hypotheses and insights d Data are collected in natural settings e less expensive Disadvantages a the least systematic of all research strategies b Causal inferences from case study data are impossible since there are no control over confounding variables c Data collection may alter the setting under study d Hypothesis testing is not possible e likely to have substantial amounts of bias f Generalisationis not possible g time-consuming