SPSS training for Doctoral students of Psychology, University of Calcutta

GENERAL OUTLINE

SPSS is the statistical package for scientific researches in social sciences like psychology, sociology, economics and in engineering sciences. In social sciences, there are large number of variables and cases. Specific distribution of variable or case or association among set of variables or cases can be examined by SPSS.

    SPSS provides both menu and syntax driven approaches in analysis of data. In menu driven approach, researcher uses the icons of the SPSS tool bars. In syntax approach, researcher writes the programme in the syntax window and runs it for the output. Syntax approach is always better than menu driven,as researcher gets freedom to analysed the variables. Syntax archive helps researchers in locating specific files and analysis of data. 

There are several function in SPSS - file management, variable creation, variable transformation,data visualization,and analysis of text and numeric data.

File-Management:
SPSS accepts both MS-Exel and Text file as input and Spss output can be inserted in the MS- office files.

Variable Creation : New variable can be created in SPSS following specific scales of measurements.Saved output variables can be inserted in the original files.

Variable Transformation :Variable properties can be transformed from text to numeric or vise - versa.One numeric property of variable can be transformed another numeric through SPSS, for the same, one can create new transformed variable or replace original variable through recode into same or different variables.New variable can be created by manipulating more numbers of variables.

Case/ variable Selection: Single or multiple cases can be analysed through select cases or if command.Similarly descriptive statistics of single variables or set of variables can be extracted through SPSS.

Data Visualisation: Visual display of data is useful in examining data quality.This is possible in SPSS.Structured or unstructured ,  large text and numeric data can be summerized an visualised through graphs and tables.


Outlier Detection: Presence of outlier or the extreme data affects the distribution of data adversely.This is specifically dangerous for pervasive statistics like correlation. Through box-whisker plot location of extreme data can be identified and be manipulated to serve the quality of data.

http://www.slideshare.net/ddroy/box-whisker-show

Analysis of data,SPSS is useful for non- parametric and parametric statistics.

ANOVA: In SPSS, ANOVA follows few nomenclatures. Independent variables is called factor. This is categorical in nature. This is determined by experimental membership. Dependent list includes list of dependent variables. These are metric or scaled variables.  Results provide three sums of squares - between, within and total sums of squares; degrees of freedom, mean square, F ratio and level of significance. When more than one IV interact with each other to change in DV, it extracts interaction sum of squares.
   The effects on DV are of two types - fixed and random effects. 
Random effects cause errors. It usually happens in sampling errors, instrumental errors and environmental errors. Confidence intervals are accounted in accepting regions of random effects. 

Non-parametric ANOVA :  Kruskal-Wallis H test  is used when  original data set actually consists of one nominal variable and one ranked variable. In SPSS, it is not important to rank the scaled data when scaled data are dependent variables. SPSS provides Mean rank, chi-square, df and significance level for interpreting the result. 

see more:http://www.ats.ucla.edu/stat/spss/whatstat/whatstat.htm

Regression : http://dss.princeton.edu/online_help/analysis/regression_intro.htm