Notes from Week 3 Readings (Part 2)
Vogt, W. Paul et al. Selecting the Right Analyses for Your Data: Quantitative, Qualitative, and Mixed Methods. New York: The Guilford Press, 2014.
What is the author's argument?
Chapter 1 of Selecting the Right Analyses for Your Data examines the challenges and benefits of coding survey data, while Chapter 8 addresses the use and vital terminologies associated with the regression and correlation visualized by statistical analyses.
Coding and measurement in surveys usually focusses on quantitative data and has two distinct phases: first before you collect the data as you write your questions and second as you sort and categorize the responses.
Surveys differ from most designs in that the first phase is the most extensive and often focuses on establishing content validity (that the survey measures what you intend for it to measure).
To move from your research question to your first draft of survey questions, the first step is to determine what your variables are.
Consider using or refining questions from previous surveys and limit the number of open-ended questions.
There are three main ways to administer surveys: face-to-face, telephone surveys, and self-administered surveys. It is important to note that survey experiments have shown that very small differences in question formats can produce major discrepancies in the results.
Many methods exist for depicting the association among variables, and most are types of correlation and regression.
Correlation and regression have many uses in social research; the concepts are closely tied to one another; they both can be understood when the patterns they describe are studied visually and statistically; and the two are the foundation of nearly all advance analytical techniques for quantitative data.
There are only two kinds of variables in regression analysis: one side is the outcome or dependent variable and on the other are predictor or explanatory variables.
The chapter reviews whether the right variables are included and the wrong ones excluded; checking for outliers and departures from normality; determining that the relations are linear when the technique assumes it; and investigating whether the sample size is adequate for the job.
Some Key Terms/Concepts
Two broad categories of survey questions: those that ask for facts (objective data) and those that ask for attitudes and beliefs (subjective data).
The most common form of survey format is the Likert Scale where participants are given a series of statements to which they agree or disagree.
Regression line is the straight line that best fits the dots of a graph. The steeper the slope of the regression line, the stronger the regression effect.
The correlation is a method of analysis that measures how close the dots are to the line of a graph. The closer the dots, the stronger the correlation.
Regression coefficient is a measure of the size of the relation between the outcome variable and the predictor variable.
Positive correlation indicates a direct relationship, which means that values on the two variables tend to move in the same direction. Negative correlation describes an inverse relation, which means two variables tend to move in opposite directions.
“Answers to survey items can only be as good as the questions.” (22)
“A ‘good’ survey is one in which the questions are valid and truly ask about what you want to know.” (31)
“Drawing causal conclusion has almost nothing to do with particular statistics…Drawing conclusion about cause is based more on reasoning about evidence than on a particular method of coding and analyzing the evidence.” (290)
“Regression is and has always been at the heart of quantitative analysis [for a variety of fields]. It always answers one version or another of the same basic question: for every 1-unit increase in a predictor what happens to the outcome?” (293)
Strengths and Weaknesses
As noted previously, Selecting the Right Analyses for Your Data begins each chapter with a helpful outline of the chapter’s objectives, a summary of the chapter’s key points, and suggestions for further reading. This aspect was beneficial as a review for Chapter 1, which was rather straight forward in its examination of survey design. Chapter 8, however, was considerably more complicated as it delved into the intricacies of statistical analysis. Despite the myriad of terms and tables presented, these were not included in the summary table at the end of the chapter. As an entry level researcher (and one not likely to utilize these methods directly), I made note of the tables with their corresponding page numbers for future reference.
How does this relate to your research?
Again, as an entry-level researcher most of chapter 8 would not be applicable to methods I plan to employ. Chapter 1, however, was useful in its examination of survey design. As the authors state “your survey is only as good as your questions” – a sentiment that encourages me to think critically about the application and intent of surveying a target audience. As with When to Use What Research Design, the authors provide a useful framework for the development of surveys in Selecting the Right Analysis for Your Data and emphasize the responsibility of researchers in formulating, gathering, and analyzing data.