Using SPSS to Understand Research and Data Analysis.

6.3 Interpreting the Output

The descriptive statistics we requested are displayed in the first table of the output file (Figure 6.5). The mean, median and mode of Task Skills 1 are 5.12, 5.00 and 5.00, respectively. These three measures of central value are in close agreement, indicating that the middle of the distribution happens to be the midpoint of our 9-point scale of task skills. Thus, the typical employee was about average on task skills at the beginning of the study. The standard deviation (1.90) indicates there is variability in task skills (on average, about 2 points from the mean).

Note that after employees attended the leadership training workshop, the measures of central tendency indicate that there was a subsequent overall increase in the mean, median and mode on Task Skills 2 (5.50, 6.00 and 6.00, respectively). Determining whether or not this is a statistically significant increase in task skills requires use of inferential statistics (a topic we will discuss in a later chapter). However, this comparison of task1 to task2 does provide some initial evidence that the workshop was effective, in that the typical employee now has a task skills score one point above the midpoint of the task skills scale. The standard deviation of task2 (1.94) indicates that there wasn't much change in the variability of task2 scores compared to that of task1.

Examination of the frequency table of task1 (Figure 6.6) confirms that the highest frequency of scores was 5 (53 employees, or 23.2%, received this score), and the Cumlative Percent column shows that 59.6% of employees received a score of 5 or lower (and 40.4% scored 6 or higher). We get a sense of the variability in scores by examining the other frequencies/percentages. The scores are fairly evenly distributed within 2 points of this middle, with few scores at either the extreme high or low end of the scale.

Examination of the frequency table of task2 (Figure 6.7) confirms that the highest frequency of scores increased to a value of 6 (55 employees, or 24.1%, received this score), and the Cumlative Percent column shows that now only 42.1% of employees received a score of 5 or lower (and now 57.9% scored 6 or higher). Examining the other frequencies/percentages, we see that the scores are fairly evenly distributed within 2 points of this middle, with fewer scores at either the extreme high or low end of the scale. However, there appeared to be a substantial increase in scores at the high end of the scale compared to task1. That is, 17.1% have scores of 8 or higher on task2, compared to 10.9% with 8 or higher on task1. This is additional evidence of the effectiveness of the leadership training workshop.

The histograms we generated visually depict this shift towards the higher end of the task skills scale (Figures 6.8 and 6.9).

Examination of these graphs reveals that the distribution has shifted to the higher end of the scale on task2 scores. The bars depicting frequencies of scores 6 or higher are much taller in the second figure than in the first, reflecting an increased number of employees with high scores on task skills after attending the workshop.

More could be said about these results, but this has shown the value of the Frequencies procedure and computing descriptive statistics in helping the researcher gain an understanding of the data from a project such as this. This procedure is often the first step in analyses, and can be done for most variables in the data file. It can lead to some preliminary conclusions early on in the analysis phase, and it is sometimes used for exploratory purposes to generate suggestions for subsequent analyses (e.g., determining whether or not the increase in mean task skills scores is a statistically significant one).