250 likes | 426 Views
Adding a statistics package. Module I3, Session 13. Learning Objectives. Recognise the limitations of any software even Excel! Introduce statistics packages in general using Instat as a simple example Explain the reasons for adding a statistics package at this stage in the training
E N D
Adding a statistics package Module I3, Session 13
Learning Objectives • Recognise the limitations of any software • even Excel! • Introduce statistics packages in general • using Instat as a simple example • Explain the reasons for adding a statistics package • at this stage in the training • Show how easy it is to use a statistics package • hence give confidence to students • that they could easily use another package • later, if needed
Contents • Review of use of Excel • Problems with Excel for statistics • all packages have limitations • Adopting a flexible software strategy • Why do more? • Instat - an example of a statistics package • What are statistics packages? • Why were they scary? • and are not now • Practical, then brief discussion
Use of Excel for statistics • Excel • Used Excel for descriptive statistics • tables and graphs • exploration and summary • Needed an add-in for effective use • Summary of EXCEL for Statistics • Data manipulation very good • Pivot tables very good • Graphics all right • Very “visual” environment
Now need to do more • Process multiple response questions • Produce tables, with weights • Cope with missing values and zeros • Excel is reaching its limits • It is a “jack of all trades” • It is serving us well • We can add other software to our “toolkit” • And still use Excel when appropriate
Another problem with Excel • Software should give the right answer • We show that Excel did not • though SSC-Stat is OK • Give the mean and standard deviation of: 1 2 3 4 5 mean = 3 s.d. = 1.58 • What is the mean and s.d. if we add 10? 11 12 13 14 15 mean = 13 s.d. = 1.58 again • Check you understand that this is true
Conclusions from this problem • Some statisticians argued • If answers were sometimes wrong, how can you trust Excel • therefore use different software for statistical applications • Our view • Data where you see problems are very odd • You should notice when you get such odd data • So continue with Excel for descriptive statistics • But switch to statistical software • for more complex descriptive analyses • and for modelling • So introduce statistics packages at this point • Earlier - SSC-Stat was a first example • of a very, very simple statistics package
Mixing software • This used to be difficult • for statistical applications • because transferring data was messy • and each package had its own commands • Now it is easy • Windows software is all similar • Menus and dialogues replace commands • These dialogues are similar in most packages • And similar to those already in SSC-Stat • An example of a simple statistical package
A familiar dialogue in SSC-Stat There are similar dialogues in any statistics package
So • A statistics package is introduced • We use new software • A simple package called Instat • Designed as a first statistics package • And to assist in teaching concepts • In the next course • We introduce a more powerful statistics package • Probably STATA
Why Instat? • A simple general statistics package • and therefore easy to learn to use • Designed to support teaching statistics • To show the value of using the computer to support the teaching of statistics • Designed as a “stepping stone” • To show how easy it is to use any statistics package • So you can easily add others when you need • It has a special menu • for the analysis of climatic data • which is a component of this training • You can distribute Instat freely • in your country
Why Instat continued • You also learn: • It is easy to move from one Windows package to another • This is useful in the next course, where you will use another statistics package • Once you have used two statistics packages • If later, you need to use further statistics packages • you should find they are simple to learn • hence you no longer need a special course • And you can work with partners • who use different packages
INSTAT Windows Key menus are Manage, Graphics, and Statistics - compare with SSC-Stat Names for columns Data start in row 1 Spreadsheet view of data Results in different type of window
What is a statistics package? • It is essentially a column calculator • like Excel when the data are in “list format” • Some results are on the previous slide • Menus are like those in SSC-Stat • Manage, Graphics, and Statistics in Instat • Manipulation, Visualisation and Analysis in SSC-Stat • (Data, Graphics and Statistics in Stata) • Other aspects of Instat • See details of the previous slide
Practical work • Instat should have been installed for you • It comes with a tutorial • And there is an extensive demonstration • The first practical gives 2 options • The tutorial for those who are more experienced • The demonstration for those with less experience • What you choose is up to you – and your staff! • And then asks for your initial views • Two things you liked • Two things you did not like
Discussion • Evaluate Instat • In its own right • And as an example of a statistics package • Also the demonstration exercise • Consider • Ease of use • Potential for further work • Advantages and limitations • compared to a spreadsheet
What were your likes? What were your dislikes? Initial views on Instat
Why were statistics packages scary? • Instinctive • Statistics is scary, • so statistics packages must be scary too! • They used to be • All different - and users not experienced in computers • Command driven • So you had to learn a language • Then you were tied to one package • Data was difficult to transfer between packages • Now • All packages are in Windows, so easy to use • Data easy to move between packages • All similar, so easy to use any, once one is familiar • Can be used in addition to Excel
Objectives: • Can you now: • Recognise all software has limitations • even Excel!! • Explain the reasons for adding a statistics package • at this stage in the training • See how easy a statistics package is to use • hence have confidence • that you could easily use another statistics package • later, if needed
Further statistics packages • For reference purposes we record • some other statistics packages with their web sites • Simple packages are as follows • CSPRO, EpiInfo – mainly data entry, some statistics • SSC-Stat, Instat used on these courses • Front runners – if you need more • Stata – good for surveys and statistics, reasonably priced • Genstat – comprehensive, good for experiments, reasonably priced and “Discovery edition” free in Africa • SPSS – very popular and excellent for surveys • SAS – the other “industry standard”. • R – very powerful, open source, becoming very popular • There are more – see the slides
The next three sessions will give further practice in using Instat