Post Reply 
(SR-52) Statistical Tests
10-01-2019, 11:32 AM
Post: #1
(SR-52) Statistical Tests
An extract from Small Programmable Calculators for Computing Statistical Tests, Charles W. Averre, Insecticide and Acaricide Tests, Volume 2, Issue 1, January 1977, Pages 7–10

"Statistical treatment of research data is extremely useful, if not necessary, in interpreting results and drawing valid conclusions from field experiments. The reasons for replicating field experiments in a proper design are twofold: (1) to have greater confidence in the results obtained and (2) to measure the inherent variability in the experiment and, by application of appropriate statistical tests, to determine differences among treatments that are due to treatment effects and not to natural variability. …Probably the principal reasons for the lack of statistical analysis in many reports is the time consuming and complex nature of the computations required and the problems associated with arithmetic errors. …

A large proportion of the field experiments … are simple and involve only a dozen or so treatments with three to six replications (usually four) in a randomized complete blocks design. … For this type of field experiment the "least significant difference" (L.S.D.) and Duncan's New Multiple Range tests are, within certain limitations, well suited and widely accepted as aids in evaluating results and making conclusions. …

In the last few years small, portable, electronic calculators with very sophisticated capabilities have become readily available at very reasonable prices. We have found the TI SR-52 … very useful for analyzing our field plot data. It is a hand held calculator … preprogrammed cards in specialized areas are available … The statistics library contains many useful tests such as various regressions, "t" test, and chi-square; the L.S.D. and Duncan's New Multiple Range tests are not available as of this writing. However, we have developed a program for these tests for a randomized complete blocks designed experiment involving up to eight replications which will display treatment means, grand total, degrees of freedom of error, F (treatments), F (replicates), L.S.D., coefficient of variation (%), and error mean square; the standard error of mean is programmed to be multiplied by the Studentized Ranges to display Rp values for Duncan's New 2 Multiple Range test. Intermediate statistics, such as sum of treatments per replication, c.f., ∑, x², etc., can also be displayed from memory registers by a three key sequence. The complete program involves two cards that are used sequentially. The first card converts the experimental data to required units, such as kilograms per hectare if desired; it also calculates treatment means, calculates sums of squares, and can be modified by the user to perform the following transformations: linear, square root, logarithmic, and arcsine. The second card calculates degrees of freedom of error and displays the values of significance, etc. The practical limiting factor still remains the problem of entering the numbers accurately. …

Fully documented.


ps: use the PDF icon (far left, above ABSTRACT) for active web link or this alternate web link.
Find all posts by this user
Quote this message in a reply
Post Reply 

User(s) browsing this thread: 1 Guest(s)