STATISTICAL DATA ANALYSIS SOFTWARE

1
STATISTICAL DATA ANALYSIS SOFTWARE

• By
• Johnson Lubega Kagugube
• Director, District Statistics and Capacity
  Development
• Uganda Bureau of Statistics

2
OUTLINE OF THE PRESENTATION

• Meaning of data analysis
• Purpose for data analysis
• Reason for statistical analysis
• Issues to consider in data analysis
• Statistical data analysis softwares
• Issues to consider when choosing a Statistical
  Package
• Conclusion

3
MEANING OF STATISTICAL DATA ANALYSIS

• Collection of methods used to process raw data
  and report the overall trends. 
• Process of systematically applying statistical
  and/or logical techniques to describe and
  illustrate, condense and recap, and evaluate
  data.

4
REASON FOR STATISTICAL ANALYSIS

• Transform raw data into information
• The general purpose of statistical analysis is to
  provide meaning to what otherwise would be a
  collection of numbers and/or values.
• Provide a way of drawing inductive inferences
  from data and distinguishing the signal (the
  phenomenon of interest) from the noise
  (statistical fluctuations) present in the data
• Statistical analysis procedures are categorized
  according to the type of statistics generated
  i.e descriptive, associative, and inferential.

5
REASON FOR STATISTICAL ANALYSIS-Cont..

1. Descriptive statistics portray individuals or
   events in terms of some predefined
   characteristics, like measure of central tendency
   and dispersion Mean, Median, Range, Standard
   Deviation, etc.
2. Associative or relative statistics seek to
   identify meaningful interrelationships between or
   among data. Such statistics include univariate,
   bivariate and multivariate analysis. For
   instance, "Is there a relationship between salt
   intake and diastolic blood pressure among
   middle-age women?" is a problem definition
   suitable for analysis by associative statistics.

6
REASON FOR STATISTICAL ANALYSIS-Cont..

• Inferential statistics seek to assess the
  characteristics of a sample in order to make more
  general statements about the parent population,
  or about the relationship between different
  samples or populations.
• Measures of differences of the means and measures
  of statistical significance
• For Example "Does a low sodium diet lower the
  diastolic blood pressure of middle-age women?"
  represents a problem definition suitable for
  inferential statistics.

7
ISSUES TO CONSIDER IN DATA ANALYSIS

• There are a number of issues to consider with
  respect to data analysis. These include
• Having the necessary skills to analyze
• Following acceptable norms for data analysis and
  presentation
• Choosing the appropriate statistical software
• Providing honest and accurate analysis
• Manner of presenting data
• Extent of data analysis

8
A Statistical package is a computer programme
that specializes in statistical data analysis.
9
WHAT STATISTICAL SOFTWARES CAN DO IN RELATION TO
DATA ANALYSIS

• Input data into the computer
• Organise data
• Compare data
• Manage data
• Summarise data (transform raw data into
  information)
• Generate tables and graphs
• Facilitate presentation of information and
  preparation of analytical reports

10
SOME OF THE STATISTICAL PACKAGES BY SOURCE
OPEN SOURCE PUBLIC DOMAIN FREEWARE PROPRIETARY ADD-INS
OpenEpi BrightStat BV4.1 SAS ANALYSE-IT
PSPP CSPro GeoDA STATA SIGMA XL
R Epi Info WinBUGS SPSS STATEL
R Commander X-12-ARIMA WINPEPI S-PLUS SUDAAN
Shogun INSTAT WINIDAMS MINITAB TOTAL ACCESS Statistics
Ploticus ZAITUN Time Series GENSTAT SSC-STATA
Simfit E-VIEWS
Statistical Lab STATISTICA
11
MAJOR STATISTICAL DATA ANALYSIS PACKAGES

• In terms of the wide usageare
• STATA
• SAS Statistical Analysis System
• SPSS- Statistical Package for Social Sciences

12
MAJOR STATISTICAL DATA ANALYSIS PACKAGES Cont..

• Licensing policies

STATA SAS SPSS
COST In US Dollars 295 6000 1599
DURATION Purchase and own the version Annual Annual
INSTALLATION Multiple installations allowed One license per CPU One license per CPU
EXTRA COST No extra pay for separate modules No extra cost Extra Modules like Survey data, and time series paid for
13
MAJOR STATISTICAL DATA ANALYSIS PACKAGES Cont..

• Other Issues

STATA SAS SPSS
Installation and Updates Simple Complicated Quick and easy
Availability of Technical support from developer All customers entitled to technical support All customers entitled to technical support Students buying Gradpack are not entitled to technical support
Web Site http/www.stata.com/ http/www.sas.com http//w.w.w.spss.com
Add-on-programs Users permitted to create new commands that integrated in the system Macros are developed but cannot be integrated in the system Little space to accept new macros
14
ISSUES TO CONSIDER WHEN CHOOSING A STATISTICAL
PACKAGE

• Important to know more than one statistical
  software package
• Analyse your needs with respect to data
  management and analysis and choose a package
  that addresses the needs
• Ease of importing and exporting data to other
  computer programmes
• Ease of transferring the output into word
  processing facilities
• Licensing facility-Purchase to own Vs hire
• General Vs Specialized purpose statistical
  software

15
UBOS EXPERIENCE

• UBOS is currently using STATA. Recently STATA
  Ver 10 and Statransfer Ver 8 were procured for
  UBOS, sector ministries and the Higher Local
  Governments
• Why STATA
• It is more sustainable
• Cost
• One time license
• Is it more useful?
• Handling data
• Graphics for exploration and reports
• Capacity for programming
• Latest version is Windows based to a great extent
• Technical capacity already available for the
  stakeholders in the NSS

16
CONCLUSION

• Statistical capacity building is necessary in
  terms of training and mentoring to
• enable countries assist and also learn from each
  other
• enable all stakeholders involved in the
  respective countries NSS to acquire expertise to
  determine their statistical data analysis needs
• enable staff handling statistics in Africa to
  acquire knowledge to use statistical packages to
  process, and analyse data to support planning and
  monitoring of development programmes
• Choosing a statistical package to use requires
  analysis of the cost, data analysis needs and the
  licensing policy.
• The National Statistical Offices should establish
  collaborative arrangements with the Statistical
  Training Institutions to ensure that the
  graduates train in the selected statistical
  packages.

17
THANK YOU