When conducting research, a given research design must be chosen with the ultimate goal in mind to solve a problem or issue that is pertinent to management. There are various strengths and weaknesses associated with different research designs and this paper will focus on three research designs, providing a brief analysis and explanation of the benefits and limitations imposed by each. The designs analyzed will include: surveys, laboratory experiments, and field studies. Additionally, the four common measurement scales utilized in research today will be discussed, with real world examples for each of these measurement scales provided.
Surveys use questionnaires and can be used to describe or predict phenomena based on their results (Davis, 2005, p.146). A survey is a type of ex post facto design, which is a study designed to determine what the pre-existing causal conditions are between groups (University of Port Elizabeth, n.d.). Surveys require interaction with respondents and are likely the most widely used research design today (Davis, 2005, p.146). Surveys are beneficial because of their ability to target large populations more cost effectively than a field study and they are also adaptable to any type of research need (Davis, 2005, p.146). Further, REACT (2000) explains that surveys can be repeated in the future to assess any changes that have taken place. However, surveys have their drawbacks as well. Davis (2005) explains that surveys don’t allow researcher intervention in an effort to control the independent variable, but instead measures the relative level of the concepts focused on in the study, so a survey wouldn’t be useful if the researcher needs to control the independent variable (p.146). Further, REACT (2000) expands on survey weaknesses by explaining that they often require special skills from the researcher in sampling, proper question design, and analysis. Further, REACT (2000) explains that surveys don’t always uncover straightforward answers to problems where there is an underlying factor or cause.
However, surveys can still be beneficial, both due to their low cost and wide applicability in a number of different situations; thus the survey shouldn’t be overlooked when designing research.
Laboratory experiments are “conducted in an artificial setting, where the researcher intervenes and manipulates some independent variable(s) in a highly controlled way” (Davis, 2005, p.148). Because laboratory experiments are highly structured, they ensure the greatest chance against design error and are the most rigorous of all models available in the research arena (Davis, 2005, p.148). The variable the researcher manipulates is the independent variable, whereas the variable being measured for changes is the dependent variable (Maricopa Community Colleges, n.d.). Laboratory experiments allow the researcher to see cause and effect relationships more clearly than other research methods since they can manipulate the independent variable(s), while holding all other conditions that impact the dependent variable constant (Maricopa Community Colleges, n.d.). However, laboratory experiments are also subject to their own limitations. Davis (2005) explains these include the artificial setting of the research can cause changes in the dependent variable that may not occur in a real world setting (p.149). Maricopa Community Colleges (n.d.) elaborates on the limitations of the laboratory experiment, explaining that researchers can only conduct real world research deemed ethical and practical. Davis (2005) elaborates by explaining that the artificial setting for the experiment can change the behavior of those in the study (p.149). Although there are limitations with the artificial setting and possible behavior changes of research participants, laboratory experiments are still the most advanced technique for allowing the researcher the greatest control over his or her research.
A field study is a type of ex post facto design that combines literature searching, experience surveying, and single or multiple case studies, which allow researchers to attempt to identify variables of importance as well as their relationships (Davis, 2005, p.144). Field studies are considered ex post facto designs because there is no manipulation or control exercised and data is often gathered in the most nonintrusive way possible (Davis, 2005, p.144). Maricopa Community Colleges (n.d.) explain that field studies are useful “when researchers want to get a detailed contextual view of an individual’s life or of a particular phenomenon.” Further, field studies are an option for the researcher when conducting a laboratory experiment would be considered impractical or unethical (Maricopa Community Colleges, n.d.). However, the weaknesses of field studies include the lack of control by the researcher since it is an ex post facto experiment and the difficulty of obtaining all useful information due to noise and other interruptions in the study setting (Davis, 2005, p.146). As Maricopa Community Colleges (n.d.) explains, behavior can only be described, not explained with a field study. Furthermore, field studies often involve a small number of participants, so it is hard to make generalizations about entire populations or large groups based on findings (Maricopa Community Colleges, n.d.). Despite the drawbacks of the field study design, they allow researchers to gather insight in a natural environment and these findings can be compiled with findings from other research studies to gain a more comprehensive understanding of the phenomena being studied.
The four measurement scales described include nominal, ordinal, interval, and ratio scales. Nominal measurement is used in the classification of objects, individuals, or groups (Davis, 2005, p.180). Nominal measurement scales allow for division to be made by classifying equalities and inequalities (Davis, 2005, p.181). An example of a nominal measurement would be using 1 to indicate power is on and 0 to indicate power is off. Ordinal measurement allows elements to be ordered by rank; however, the distances between the elements have no meaning (Trochim, 2002). An example of ordinal measurement would be creating a survey to learn about income levels in a survey. It would look as such: 0= less than $20,000; 1=$20,001-$40,000; 2=$40,001-$60,000; 3=$60,001-$80,000; 4=$80,001-$99,999; 5=$100,000 or more. Interval scales have the properties of nominal and ordinal scales, while also having equal points on the scale of measurement (Davis, 2005, p.183). As Trochim (2002) explains, that attributes with distance among them have meaning and averages of interval variables can be computed. Taking the income situation a step farther to illustrate interval measurement, you have the following example: Assume person A and B have incomes of $55,000 and $100,000 respectively. Person A could determine using intervals that person B makes $45,000 more than him. This is computed by subtracting $100,000 from $55,000. Additionally, the average income of A and B can be computed by adding both figures together ($155,000) and dividing by two, to come up with an average income of $77,500. Ratio measurements maintain all of the attributes of the prior three measurement scales, with the addition of an empirical absolute zero point (Davis, 2005, p.184). Simplifying the concept, Trochim (2002) explains that one can create a fraction (ie: ratio) using the measurement technique. For example, a store could take a count of customers making purchases in January and July of a given year. Let’s assume that the January customer count is 25,000 because the store recently opened and the July customer count is 50,000. The store could explain that from January to February there was a 2/1 increase in customers from January through July. This is possible because the store could have zero customers and uses absolute zero as the basis to generate their computation of the ratio.
References
Davis, D. (2005). Business Research for Decision Making. (6th. Ed). Mason: Thomson South-Western
Maricopa Community Colleges. (n.d.). Research Methods. Maricopa Center for Learning and Instruction. Retrieved June 20, 2006, from http://www.mcli.dist.maricopa.edu/proj/res_meth/rmvl/index.html
REACT. (2000). Data Tools for the Community Profile. Regents of the University of Minnesota. Retrieved June 20, 2006, from http://www.epi.umn.edu/react/main/community_org/meetings_surveys.html
Trochim, W. (2002). Levels of Measurement. Research Methods Knowledgebase. Retrieved June 20, 2006, from http://www.socialresearchmethods.net/kb/measlevl.htm
University of Port Elizabeth. (n.d.) Research Method. Retrieved June 20, 2006, from http://www.petech.ac.za/robert/resmeth.htm
