SOCI 217 Sociological Research Methods Week 9 Readings - Quantitative Analysis Social Research Methods, Chapter 8: Quantitative Data Analysis Introduction Though data analysis is performed after it is collected, decisions regarding data analysis must be made before doing so because: 1. certain types of measurement may not allow for certain types of analysis later - keep in mind what and how it is you want to analyze data before collecting 2. Nature and size of sample impose other analysis limitations and specifications Missing data can be an issue in research; where respondents may skip over one or more questions and thus create a gap in the data · Missing data points must be coded accordingly so (i.e. a code that cannot be mistaken for an answer) Dichotomies or dichotomous questions are those which have only two options · A form of forced choice - ppts must side with one or the other There are three main types of variables in research 1. Nominal variables - those whose only difference is categorical (it is only measured by the fact 2. Ordinal variables are those which can be rank ordered by category, however difference between ranks is not measured a. E.g. "always, usually, sometimes, never" b. Another order would be illogical 3. Interval and ratio variables are those which have units of measurement and thus the difference between each rank can also be measured and kept equal a. You can explain how much one data point differs from another numerically b. The difference between ratio and interval is that ratio has a true zero which makes it possible to make claims like "data point X is 2/3 of data point Z" Univariate Analysis Univariate analysis refers to analyzing one single variable at a time, to which there are several different approaches: · Frequency table: shows number of and percentage of people in each category of variable in question o Can be used to all three types of variables
· In frequency tables: valid percent should be used rather than percent because that includes missing values Remember that categories must be mutually exclusive (people can only belong to one) and all people must be categorized, even if it is under a 'missing' or 'void' category because they did not (properly) provide their data · Diagrams take many different forms and are often used as a way to visually display quantitative data ? Bar graphs & pie charts are popular for ordinal and nominal data · Show the size of each category of a variable, especially in relative size to whole sample · Histograms appear similar but are used for interval/ratio variables and do not have spaces in between each bar · Measures of central tendency (most common being mean, median, mode) provide a single number that tells us an average about a data set o Mode - can be used with any data but most common with nominal o Median - midpoint in a set of ordered data; useful for all types of