440 - MAIN (LOG) PAGE  <=you are here

 440- DEFINITIONS 

 440 - PROBLEMS AND SOLUTIONS

 EPI-INFO page

WEEK #

As we move through the materials and examples, the background colour changes from  green

to  yellow. ;  (Note that green - future - is only approximate).

In this way you can see what we have already covered, and what we should be covering next.   

Sept 23  

Displaying data in histogram, normal (Gaussian) distribution. Non-normal distribution examples, calculation of mean, median, mode as indicators of central tendency.

 Confidence limits around the mean, and interpretation. Summary examples.

 Some of the PowerpointŪ slides used during today's lecture are Here  

 Introduction to Null Hypothesis and scientific method of enquiry. Review of sampling theory in epidemiology: target population, random sample, sampling error, systematic error, bias, normal error,

Introduction to hypothesis-testing and comparison of two means, rephrasing as a null hypothesis and concept of testing the null-hypothesis (H₀), rejecting/not rejecting,

 Some of the PowerpointŪ slides used during today's lecture are Here

  Please note a slight change has been made near the end related to the

  standard  deviation of the first group.

Review:.  Clarification of the background to forming a null hypothesis

Part 1: Practice with single group statistics

Part 2: Practice H₀-testing using t-test.  Example from blood-testing for mercury in northern Ont. and relation to fish consumption, rationale, comparison of calculated t vs critical t, probability statement, and conclusion.  Especially note the importance of the denominator of the t-test and how it may be reduced by increasing the "N"

 Practice problems on this material:  2,  13,  16, 17, 18,  21, 24, 25, 26, 27, 31,

  - are being added on the question page.  Answers to these problems are

Special case for t-test used on Paired data; what is pairing?, and why does this offer a better result IF USED PROPERLY! 

Comparison with t-test for UNPAIRED data

  Some of the PowerpointŪ slides used during today's lecture Here 

  General practice problems on this material from the course text:

    Try p.306, Q#1, Q#2,  Q#13  (Answers on page 633)  I also

    recommend using the slightly more simple formulae from the lectures.

    More specific practice problems on this material can be found

    on the Proteus  question page.  

      Try: #  3,   15,   19,   22,   23,   43,   49,   51,   52,   53,

    (Answers to these problems are shown on the solutions page.)

General practice problems on this material from the course text:

    Try p.224, Q#2, Q#6, Q#7, Q#8,  Q#9 a, Q#10 b,c, 

 (Answers on page 627)  You can ignore the calculation of the confidence

limits around the odds ratio, but you should remember that if the CL

include 1.00, you are NOT permitted to claim statistical significance

regardless of the results of the Chi-sq or Fisher's Exact Test. 

Also from Chapter 9, try Q#11, Q#15, Q#16, Q#19, Q#20 b, Q#23

As before, you may find the formulae from the lectures and the PowerPoint*

notes a little easier to use than those included in the text.

More specific practice problems on this material can be found

    on the Proteus  question page.  

      Try: #  1, 6, 9, 20, 28, 29, 35, 38, 41, 50, 54

    (Answers to these problems are shown on the solutions page.)

Some general concept questions on regression can be found on p 498 etc.

of the text: # 1, #4, #5, with answers on page 643.

More specific practice problems on this material can be found

on the Proteus   question page.  : 7, 8, 10, 11, 39, 40,  46, 49,

  (Answers to these problems are shown on the solutions page.).

(An error was found in the regression example shown in class Nov 18. 

Some general concept questions on correlation can be found on p452 etc.

of the text: # 1, #13 (a, b, c), #17 (a, d),  with answers on page 641.

A more specific practice problem can be found on the

Proteus  question page.  # 49.  Also #39 calls for correlation to investigate

the possible relationship between the two independent variables (after

the regression analysis explores the relationship between each of them

and the dependent variable separately. Finally, #48 is solved using correlation

on select sets of data (not yet posted).