Article critique and analysis

PC's, Central-System Computing, and Government Functionality

In this article, the authors use correlational methods to prove their point-of-view. The way that this is accomplished is through a survey. They used percentages comparing two sets of subjects as well as a statistical analysis based on Cramer V scoring. The importance of the correlation was based upon whether or not the values in the study were greater than.20 on the Cramer V scale. If the score was under.20 it showed a weak relationship, while scores greater than.20 on the Cramer V scale showed a moderate relationship.

The authors address the computing problem quickly. The problem being the widespread belief that PC's can easily and cheaply replace mainframe computers in governmental operations (page 1, paragraph 3, lines 1-4). The problem itself was deduced from research studies completed over the last decade and is answered by the study conducted by the authors. The objective of the study is stated as, "Our aim is to assess whether the two computing environments are functionally equivalent, as many of the PC advocates would have it, or whether PC-only environments provide only some of the functions, applications, and benefits of their bigger brothers and sisters -- the mainframe and minicomputer environments" (page 2, paragraph 3, lines 1-4). There are two types of variables used in this study. The first being the independent variable defined by the authors as the computing environment (page 7, paragraph 5, line 2). The second is the dependent variable, defined as the relationship between specific impacts (page 7, paragraph 5, line 1). These variable were used throughout the study, except for one instance where due to the wording of the questionnaire these variables where unable to be used.

The scope of the study conducted by the authors clearly addresses all of the concerns raised in the article. However the population of this study is defined as, "...two sets of cities..." (page 2, paragraph 4, line 1). The description of this population is very vague in that there is no clear definition as to the size of the cities involved. It is impossible for the reader to know whether this study is based on small cities or a large metropolis. The population used in studies referenced by the authors range from small to medium cities, but no reference is made as to whether or not the authors used the same size population in their own study. The authors specify the sampling of this study to be, "Assistant managers, department heads, technical staff, and administrative staff..." (page 4, paragraph 3, line 3), and referred to studies used for comparison that used the same sampling. It is unclear as to whether the same population is used throughout the study given the statistical differences. For example, the authors defined the scope of the study as using two sets of cities, but later in the study they refer to responses given from 10 different cities. "When central-system cities rated their experiences with minicomputers and/or mainframes, more than 8 in 10 (83.4%) indicated that these systems had met or exceeded their expectations" (page 9, paragraph 2, line 1). Due to these inaccuracies it is impossible to determine if the same methodology is used throughout the survey. In the final two pages of the study the authors address the significance of their study in comparison to other studies. Each point of the problem is addressed in these pages followed by the authors deductions of their study and backed by polls or studies completed by various groups.

Only certain areas of this study can be replicated. The study's treatment can be replicated in that the methodology used is clearly defined as PC vs. mainframe and the Cramer V scoring method. The data collecting methods cannot be defined due to the inconsistencies of the population used. The size of the sample was not adequately described, however the characteristics of the sample were clearly laid out (Assistant managers, department heads, and technical staff). The treatments administered for the study were held constant for all the treatments in that the authors define the parameters of the Cramer V scoring method and used them consistently throughout the study. The alpha level in this study is a priori of p=. 20 defined as a moderate relationship between the variables.