Practical Data Analysis

Practical Data Analysis

Copyright: © 2015 |Pages: 26
DOI: 10.4018/978-1-4666-8116-3.ch012

Abstract

In this chapter, students will learn “what to do” with their quantitative data once it has been collected. The chapter begins with a discussion of data coding, which is the process of preparing one's data for statistical analysis. What follows is a discussion of basic univariate, bivariate, and multivariate data analysis techniques. These techniques are presented in such a way that students with limited statistics backgrounds can understand and employ. Emphasis in this chapter is placed on giving students a working knowledge of statistical techniques that are most widely used when interpreting quantitative data.
Chapter Preview
Top

Pre-Analysis: Levels Of Measurement And Coding Your Data

Most public administration students and future public administrators know very little about practical data analysis techniques. Spanning the spectrum of statistical know how, there are those on the far left who have little or no understanding of what to do with quantitative information. On the far right side of this spectrum, there are those capable of complex statistical analyses. Students reading this text, and public administrators, should strive for a basic and useful understanding of what to do when you have collected information using one of the three deductive empirical tools discussed throughout this book.

Prior to analyzing your data, you need to code it. Coding is the process of making your data ready for statistical analysis. By and large, coding involves taking any qualitative label and transforming it into a numerical value. For example, if you had gender as one of your variables, you would obviously note whether a research subject was male or female. You cannot conduct-statistical analyses using word labels. Those word labels need to be transformed into numerical values. Thus, you might code males as “0” and females as “1” – or vice versa. Key to coding data is to understand the different levels of measurement. There are four levels of measurement (see Figure 1):

Figure 1.

Levels of measurement

  • 1.

    Nominal

  • 2.

    Ordinal

  • 3.

    Interval

  • 4.

    Ratio

Nominal measures are simply names or labels that have no rank order – rank order meaning highest to lowest, large to small, etc. Gender is a nominal measure, as is race and ethnicity, for example. Ordinal measures are names or labels where there is rank ordering. There is implied in each of these names or labels a greater than or less than rank ordering. The classic ordinal measure is the Likert scale. A typical Likert scale question is the “agree or disagree” question:

The writer of this text book presents the material clearly.

  • 1.

    Strongly Disagree

  • 2.

    Disagree

  • 3.

    Neither Agree nor Disagree

  • 4.

    Agree

  • 5.

    Strongly Agree

Since there are five choices, this would be considered a five-point Likert Scale. There can be seven point scales. A seven-point Likert scale might look like this:

Rate your experience reading this textbook.

  • 1.

    Very Dissatisfying

  • 2.

    Somewhat Dissatisfying

  • 3.

    Slightly Dissatisfying

  • 4.

    Neutral

  • 5.

    Slightly Satisfying

  • 6.

    Somewhat Satisfying

  • 7.

    Very Satisfying

Likert scales provide no objective basis for differentiating numerical values between, for example, very dissatisfied and somewhat dissatisfied. Even though these labels are “one unit away” from each other, you cannot infer that the difference between “very” and “somewhat” is one unit. Ordinal measures allow for subjective comparisons – that is, the relative magnitude.

Complete Chapter List

Search this Book:
Reset