Statistical Analysis in Reaching Their

¶ … statistical analysis in reaching their conclusions and presenting their findings in published literature. These techniques are necessary in order to provide consistency and validity to the data, as well as to provide grounds for comparison with other studies and other data. It is from these needs for consistency and comparability that specific statistical methods and tools have been developed and put into common use by researchers and statisticians of all stripes, including those working in the area of human resource management. Though this field of study also contains a great deal of qualitative information and conclusions, as with any good science there are measurable and quantifiable elements to the study and application of human resource management, and this leads to statistical analysis.

This paper will examine eight different statistical measures and tools of analysis as they appear in literature pertaining to the study of human resource management. The application of these statistical tools in real world contexts and practical situations will add a great deal of clarity to their use, and will also provide an opportunity to examine the different ways in which these statistical tools can be used, and indeed are used in current research. A brief critique of the usage of each statistical tool in the context examined will also be given, to provide a broader framework of understanding regarding the use of each statistical tool.

Regression

A study concerning the perceived work climate and turnover rate for different occupational categories in a Korean hospital provides an excellent example for a discussion of regression analysis. The study was conducted by comparing observed turnover rates to responses from a questionnaire given to employees, which measured a wide array of attitudes and perceptions that were classified into a total of 32 variables (Hwang & Chang 2009). The number of variables necessitated regression analysis in order to determine which specific attitudes and perceptions were most correlated with turnover rates.

This is an instance where regression analysis was absolutely necessary to answering the central research question. Regressions are used to compare the effects of two or more independent variables on an dependent variable (in this case, the turnover rate), and with 32 identified independent variable a regression was definitely called for. This is a simple, compelling, and straightforward example of an application of regression analysis.

Time Series Analysis

The title of this set of statistical tools is not at all misleading, but is in fact completely accurate and descriptive of the methods and manners of the tool itself. An examination of the effects of fear on worker morale over a relatively long period of time is an excellent current example of this type of tool. Research shows that using fear as a motivator is hugely detrimental to employee performance, with the effects increasing dramatically as exposure to workplace fear continues for longer periods of time (Edwards 2009).

This use of time series analysis is not as clear cut or as necessary as the example of regression given above, but it still serves as a useful indicator of the scope of applications that time series analyses can have. The effects of fear on employee performance were measurable in discrete instances, but the researcher's decision to perform a time series analysis showed a different set of results -- i.e. A changing effect -- than these discrete instances alone would have provided. This is therefore a highly useful application of the analysis.

Double Blind Study

In order to eliminate bias as much as possible, many research studies are designed in such a way that the researchers are unaware of precisely who or what they are measuring until after the study's completion. In a study of the role of employee competency in service profitability, for instance, researchers analyzed questionnaires from respondents without knowing who they were or where they worked, only making these connections after their data had been compiled and the variables analyzed for correlation (Xu and Heijden 2005).

The postponement of the full knowledge of a respondent's participant's status is what makes the study double-blind; this makes it impossible for the researcher to unconsciously alter their interpretation or perception based on knowledge of criteria outside the scope of what the study tries to measure. Double blind studies are more common in medical research, but this is an effective example of the use of a double blind study in the human resource management field. The researchers wished to eliminate any suggestion of bias, and by eliminating their knowledge of certain personal details regarding their respondents they did exactly that.

Triangulation

The use of several methods and/or approaches to the same research question in a given study is known as triangulation, and it provides both a greater understanding and a heightened validity to the results of the study. A recent literature review of nonprofit human resources practices covering a broad range of situations and effects necessarily incorporated several different methods of analysis in order to determine the most common and accurate ways of analyzing and addressing nonprofit human resources concerns (Ridder & McCandless 2010).

The triangulation employed in this analysis of human resource management vs. strategic management in their effects on nonprofit success allowed for a better understanding of the issue from a multitude of perspectives, making the need for appropriate human resource management much more clearly and strongly stated than any of the single methods employed could have (Ridder & McCandless 2010). This makes the example a very clear demonstration of the use of triangulation as a method of analysis and presentation. Triangulation enhances all of the contributing methods of analysis, creating a whole that is greater than the sum of its parts in terms of its analytical clarity and impact.

Hypothesis Testing

Though very common in studies in the "hard sciences," hypothesis testing is somewhat underutilized in human resource management literature. A study concerning health worker motivation in Kenya, however, began with the hypothesis that non-financial incentives and human resource management practices would have a greater influence on motivation than financial incentives, and through the course of a two-phase study the researcher's confirmed this hypothesis in part, but finding that certain human resource management techniques had a detrimental effect on motivation, contrary to their prediction (Mathauer and Imhoff 2006).

This provides an excellent example of hypothesis testing because it demonstrates both the confirmation and the rejection of a hypothesis. The researchers also started with years worth of observations, allowing them to develop a reasonable hypothesis and then set about testing it. Their openness to being wrong in their hypothesis is also an important aspect of the scientific process, and especially of data analysis and presentation.

Focus Group

At times, more open-ended and qualitative information is needed, and focus groups can provide a great deal of information in such instances. Airline safety, for instance, can be measured by statistics, but reporting accuracy is a problem. A focus group consisting primarily of airline pilots led to the fairly certain conclusion that safety is sacrificed for personal and financial gain, and the regardless of conflicting motives safety in airline operations could never really be guaranteed, but only forecast (Gill 2004).

There are other methods that might have allowed these pilots to divulge the same information -- this study also utilized a mailed questionnaire, in fact -- but the focus group allows for ideas to be put forth and debated in a group, leading to more comprehensive and more clearly-defined (through the progress of the discussion) terms and interpretations, thus leading to a greater validity of findings. By utilizing a focus group in this study, the researcher yielded far more comprehensive and detailed results than through the questionnaires (Gill 2004).

Snowball Sampling

Snowball sampling is a process whereby subjects recruited to participate in a study recruit suitable friends or acquaintances of theirs to join the study also. This is typically done when researchers have trouble identifying or approaching members of the group they are trying to study, and as such this practice is almost unheard of in the area of human resource management, which is not especially controversial or hidden. A research study involving nursing mistakes with medication administration, however, used snowball sampling as a way of obtaining participants otherwise unwilling to "rat out" their peers (Sheu et al. 2008).