Three Stages of Data Analysis

Three Stages of Data Analysis

In all areas of scientific endeavor, the collection and analysis of relevant, objective data is a rather obvious necessity. In the hard sciences such as physics and chemistry, this process is often relatively straightforward -- phenomenon are observed, measurements are taken, and analysis yields possible correlations and causal relationships. This same basic process hold true for the social sciences, but determining what to measure -- and what has been measured -- can itself be a daunting and uncertain task, let alone the analysis of such data for theoretical or practical conclusions and applications. This can be especially true in the field of psychology, where much of the data received will by definition be subjective on the part of the respondent, and which allows for widely different interpretations of such data depending on the theoretical perspective one is working under. For this reason, a clearer understanding of the correct procedure for obtaining and analyzing data is necessary in psychological research.

Getting to Know the Data

The first step in any data analysis, especially in psychological research, is to come to a full and complete understanding of the data (Shaughnessy et al. 2006). This requires cleaning up the data, removing any incomplete entries, impossible figures, and taking careful note of any outliers (Shaughnessy et al. 2006). In this way, the data can be made more uniform, making it more likely to yield accurate interpretative results in further analysis. Different methods can be used to accomplish this clean-up of a data set; stem-and-leaf displays have been found to be especially useful in this regard (Shaughnessy et al. 2006). These displays provide both numerical and pictorial representations of data, making it easy to spot outliers and to see other trends in the data; certain correlatives might even begin to emerge at this point.

Summarizing data is essential not just for the presentation of research and/or findings to peers and colleagues, but also for one's own edification, familiarization, and understanding of the data (Shaughnessy et al. 2006). Such summary can take place numerically, pictorially, or verbally, and often all three methods of presentation and summarization are used for a more complete and effective understanding of the data (Shaughnessy et al. 2006). This enables the researcher and any peers to familiarize themselves with the data from several angles before the true in depth analysis occurs, making this an essential part of the data analysis process in providing the preliminary framework for future analysis (Shaughnessy et al. 2006).

Summarizing the Data

Once the data has been summarily arranged in any or all of the above three methods, the statistical summary can (and should) begin. Important measures in data summarization include measures of central tendency (i.e. "averages" or the mean, median, and/or mode) and measures of dispersion or variability -- the range of the data and the standard deviation of the points within the data set (Shaughnessy et al. 2006). These statistical staring points can be used to derive a wealth of information form the data, including correlations to other related studies/data sets, reliability and consistency of the data set at hand, and other summary statistics that provide the necessary measures to begin to understand the implications of a given data set (Shaughnessy et al. 2006). These basic figures must be known before any statistical analysis can occur.

Effect size is also a very important measure in the summary of a data set (Shaughnessy et al. 2006). Rather than simply showing a correlation between various features of a data set, effect size measures determine the strength of such relationships; some things that appear to be correlative might have effect sizes that are quite small, suggesting perhaps a different causal agent common to the two phenomenon, or leading to questions requiring firther research (Shaughnessy et al. 2006). In this way, even when data analysis points away from desired conclusions, a carefully conducted research project with proper data analysis will never be entirely fruitless. Summary provides the clear and concise results of an experiment as represented through the data, and is therefore quite essential to the overall process of data analysis.

Confirming What the Data Reveals