Lecture 8: Descriptive Statistics and Statistical Inference on SPSS Research Methods I Lecture 8: Descriptive Statistics and Statistical Inference on SPSS
Introduction This lecture has several purposes: To reinforce notions of statistical analysis encountered elsewhere To highlight conceptual issues in descriptive statistics in particular To introduce you to some basic features and functions of the SPSS package Lecture: conceptual and practical
Conceptual issues Descriptive statistics: describing the sample Statistical inference: making (probabilistic) claims about the population, based on the sample Inference from samples is always a gamble (Bryman and Cramer: 110) but it is one which is usually taken!
Remainder of the lecture The remainder of the lecture material here will be lists of issues you should be familiar with However, the lecture time will be spent demonstrating how SPSS can assist in dealing with these concepts You should practise with SPSS in order to get familiar with it. SPSS is a good package used widely in social science
Types of data There are four types of data: Nominal or categorical data: e.g. gender Ordinal: e.g. ranking preference but not on a scale Interval: measured on a scale with equal intervals: e.g. 4>3 = 10>9 Ratio: same as interval but has a true zero Type of data constrains type of test
Univariate frequency distribution Discrete categories required Interval/ratio data and midpoints Relative frequency: pie, bar, histogram Cumulative frequency Frequency distribution Normal distribution Skewness and kurtosis
Measures of location and dispersion Mode Median Percentiles and quartiles Mean (including grouped and frequency) Measure of qualitative variation Range Interquartile range Variance, SD and coefficient of variation
Parametric data Conditions for parametric data: Normality Homogeneity of variance Independence of observations Interval/ratio data NB non-parametric statistics Most tests economists use are parametric
Hypothesis testing in data Arbitrary nature of statistical testing Covariance Correlation Partial correlation Testing between two means Regression will be dealt with in lecture 10
Methodology and Ho testing Positivists generally regard hypothesis testing as crucial to science and advocate statistical testing (with a range of methods); inference crucial Interpretivists: generally do not engage in inferential statistical work Critical Realism: testing involves ‘closure’; but many CR have used stats, including non-parametrics