How to arrange the order of the five factors of Latin and balanced Latin experiments?

First of all, it should be stated that the Latin side in psychology is only a technique to balance the order of experiments in order to reduce the influence of the experimental order on the experiment, so generally speaking, the specific test topic of the experiment is related to the design requirements of the Latin side.

The Latin method is designed to ensure that every possible test sequence appears in the test of the subject in a small experiment, so the principle is the same as that of the random sorting principle of a large-scale experiment, except that manual control is added. For example:

Suppose I want to do an experiment now, the subject has to take a total of 5 tests, and needs to retest many times, so the order of these 5 tests needs to be included in the variable control, it is impossible to have the same order many times, so in order to balance this order effect, the Latin design is adopted, and the 5 tests are named 1, 2, 3, 4, 5. So that's how it goes by:

The first set of test sequences: 1, 2, 5, 3, 4

The second set of test sequences: 2, 3, 1, 4, 5

The third set of test sequences: 3, 4, 2, 5, 1

The fourth set of test sequences: 4, 5, 3, 1, 2

The fifth set of test sequences: 5, 1, 4, 2, 3

The order is determined in such a horizontal order: 1, 2, n, 3, n-1, 4, n-2......(n represents the number of quantities to be sorted) Vertical: 1, 2, 3, 4, 5 and then reincarnated (for example, the following 6, 7, 8, 9, and 10 correspond to the order of 1, 2, 3, 4, and 5 respectively).

By The Way, LZ is going to do a 5-factor experiment ......It's a little scary = =! Generally, a 3-factor 2*2*2 experiment is enough for a headache, but are these 5 factors not interactive? ~~= =~~

hope you well-being！

What kind of mess are you.

It should be a 5×5 grid.

Then, according to the table of random numbers, make row changes first and then column changes.

Guo's book p81 makes it clearer. The rules of the rows and columns are the same, for each row and column in order from small to large, meet n and then start from 1 again in alphabetical order. For example, the first row is 1, 2, 6(n), 3, 5(n (n-2) (small to large) and the second row is 2, 3, 1 (meets n starting from 1, 6(n (n-1)).

In the Latin design, each row or column becomes a complete unit group, and each treatment appears only once in each row or column, i.e., in the Latin design, the number of experimental treatments = the number of horizontal unit groups = the number of inline unit groups = the number of replicates of the experimental treatment. Two factors, three levels, then there are 6 combinations, that is, a 6x6 Latin design (equivalent to six treatments, each treatment 6 repetitions). Of course, the landlord can also carry out orthogonal design.

Reply 2 If you use an orthogonal design, how do you use it? If you use a full-scale experiment, the scale of the experiment is too large.

A6 If there is an interaction between the two additives, can the Latin side still be used?

The Latin square design allows researchers to statistically control two external variables that do not interact with each other and manipulate the independent variables. Each external or partition variable is divided into an equal number of blocks or levels, and the independent variables are similarly divided into the same number of levels.

The Latin side is conceptualized in the form of a **, where rows and columns represent blocks in two external variables, and then the level of the independent variables is assigned to the cells in the table. Specifically, Latin is a technique used to balance the order of experiments in order to reduce the effect of the order of experiments on experiments. It is supplemented by a Latin square, which is a square matrix composed of several variables that need to be sorted.

The specific application process is as follows: Suppose I want to do an experiment, the subject has to take a total of 5 tests, and needs to retest many times, so the order of these 5 tests needs to be included in the variable control, it is impossible to have the same order many times, so in order to balance this order effect, the Latin design is adopted, and the 5 tests are named 1, 2, 3, 4, 5. So that's how it goes by:

The first set of test sequences: 1, 2, 5, 3, 4 The second set of test sequences: 2, 3, 1, 4, 5 The third set of test sequences:

3,4,2,5,1 Fourth set of test sequences: 4,5,3,1,2 Fifth set of test sequences: 5,1,4,2,3 The order is determined in this way, horizontally:

1，2，n，3，n-1，4，n-2……(n represents the number of quantities to be sorted) Vertical: 1, 2, 3, 4, 5 and then reincarnated.

Introduction to Experimental Design Ideas:

In order to explore the factors that affect students' class efficiency, we designed a hybrid design experiment:

The three independent variables were: the gender of the participants, the size of the classroom (size of the space), and the number of participants.

Since it is a hybrid design, we divide students into ...... middle school, high school, and university groups

The Latin square design is the easiest way to design.

For a specific case of the Latin design, see Psychological Research Methods by Huang Xiting of Southwest Normal University.

Adopt me, dear.

Your question is not specific. The Latin formula is a method of experimental design, and the number of factors in the experiment varies according to the number of Latin judgments. How many factors did you consider in your experiment?

I'll bring it up next time. If it has not been tested and dismantled, it can be considered for use on the bridge"Orthogonal design of experiments"Superior to it.

Orthogonal Latin square Zen eggplant table.

Raise the suspicion and attack the void

The Latin square design refers to taking any n numbers, 1, 2, 3, 4, etc. Arrange a matrix of n*n so that each row and column has a 1, 2, 3, 4, etc., for example, the following is a Latin square.

Constructing the Latin side, you can take a look at the relevant content in combinatorics. There's a way. ]

Your question is not specific. The Latin side is a method of experimental design, and the number of factors in the experiment varies depending on the Latin side. How many factors did you consider in your experiment?

I'll bring it up next time. If it has not been tested, it may be considered"Orthogonal design of experiments"Superior to it.