Page 88 -

74 CHAPTER 4 Statistical analysis

example, in Table 4.2, I used “1” to represent “high school degree” rather than “0.”
However, when the data is processed by a statistics software, a coding scheme of “0,
1, 2” is exactly the same as a scheme of “1, 2, 3.”
In various studies such as surveys, interviews, and focus groups, content analysis
needs to be conducted in which text reflecting different themes or critical events
is coded and counted (Stemler, 2001). Detailed discussion on content analysis is
provided in Chapter 11. Event coding is also quite common in usability tests or
lab-based studies. For example, Hu and Feng (2015) used extensive coding schemes
to analyze the causes for failed browsing or search tasks in an online environment.
The coding scheme allowed the authors to further understand the difficulties that us-
ers experience when finding information online.
When coding your data, it is critical to ensure the coding is consistent. This is
particularly challenging when the coding is completed by more than one person. If
the coding is inconsistent, the validity of the analysis results will be greatly affected.
Various statistical methods, such as Cronbach's alpha, can be used to assess the reli-
ability of coding completed by multiple coders (Weber, 1990). Please see Chapter 11
for more details on this topic.

4.1.3 ORGANIZING DATA
Statistical and other data-processing software normally has predefined requirements
for how data should be laid out for specific statistical analysis. In SPSS, for example,
when running an independent-samples t test to compare two groups of data, the data
of the two groups need to be listed in the same column. In contrast, when running
a paired-samples t test to compare two means, the two groups of data need to be
laid out parallel to each other in two separate columns. Similarly, other statistical
methods such as ANOVA, repeated measures, and correlation all have different data
organization requirements that need to be followed closely.

4.2 DESCRIPTIVE STATISTICS

After the collected data is cleaned up, you may want to run a number of basic descrip-
tive statistical tests to understand the nature of your data set. For instance, you may
want to know the range into which most of your data points fall; you may also want
to know how your data points are distributed. The most commonly used descriptive
measures include means, medians, modes, variances, standard deviations, and ranges.

4.2.1 MEASURES OF CENTRAL TENDENCY
When we study a data set, we often want to find out where the bulk of the data is lo-
cated. In statistical terms, this characteristic is called the “central tendency.” Various
measures can be used to describe the central tendency of a data set, including the
mean, the median, and the mode (Rosenthal and Rosnow, 2008).