Descriptive Statistics Article Critique
Descriptive Statistics Article Critique In the article Do Men with Excessive Alcohol Consumption and Social Stability Have an Addictive Personality? the authors investigate the validity of the statement, which is mentioned in the title. There were applied certain descriptive statistics methods, while at the same time a few significant techniques were omitted. The researchers provided an overall analysis of the raised issue, acceptable argumentation of investigation results, notwithstanding left a research field wide enough for future studies.
In this paper the authors set a goal to discover, whether there is any correlation between excessive alcohol consumption and behavioral traits of men with certain social status. As the hypothesis, it is stated an idea that there are no specific personality patterns, which can be observed among individuals with excessive alcohol consumption. Considering the criteria for evaluating research questions, the ones in this investigation are formulated clearly, not too abstract or wide and researchable in general, so they can be accepted as appropriate.
The investigation was conducted for two groups – 100 individuals with excessive alcohol consumption and control group, consisted of 131 individuals. The research included an assessment by the Karolinska Scales of Personality (KSP) and the Principal Component Analysis (PCA). These methods were based on the descriptive statistics techniques including using the mean of samples, variance of the given values and determination of outliers.
While evaluating statistical analysis applied in this research, it is necessary to affirm that the authors’ choice can be approved. Firstly, the mean in that analysis is suitable and more precise than other measures of central tendency – the median, for instance. It is preferred in cases like that in the article given within nearly equal-sized samples. In order to make a research more specific and to get significant results for the aim stated, the techniques of variance and, as the result, defining outliers were also used. In case, when scores are spread out, it is appropriate to use such measures of variability. The authors have applied this method successfully, although a comment regarding outliers analysis should be considered. As to the theory of statistics, it is much more useful to conduct the research again after revealing and excluding outliers. Thus, defining the aspects that should be added to this particular research, the one mentioned above would be the most significant.
The second disputable question is the selection of variables. There are three groups of variables presented in the article: anxiety proneness, aggressiveness and hostility, as well as those related to social vulnerability. Accordingly, these groups include 15 different items of research. It cannot be said that they are completely comprehensive, but as to this particular analysis it is sensible. There is no use in including a lot of variables, since it can be avoided in such research.
The authors of the study indicate some limitations, which are connected with the samples within the research. The most considerable gap in it is relatively small samples for the PCA analysis. It could significantly influence the result received and statistical options, like variance. Moreover, the researchers noticed that the study included only middle-aged men, recruited be the advertisements, and also there is no evidence in the absence of alcohol-dependent individuals among the controls.
The last, but not the least aspect that should be highlighted is insufficient use of the problem statistical testing. Unfortunately, the procedures of measuring reliability and validity, which should be determined, are often not followed. There are a couple of statistical tests (for example, Mann-Whitney U-test), considering aspects of distribution of the values, revealing differences in samples and others, which should be applied in this particular case. In addition, the research would be more scientifically significant with certain weights, given to separate variables.
To crown it all, this analysis can be defined as significant; however it has features of superficiality without deep problem exploration and testing of the results achieved. Whereas the problem needs more comprehensive investigation, several improvements, such as samples expanse, excluding outliers, in-depth distribution analysis could be suggested as well.
