Significance tests

The items on this page are:-

Several significance tests can be carried out using the basic commands, e.g. those involving means, so those included as specific commands are the ones which would be difficult to do via a series of arithmetic commands.

Probability and distribution tables will have to be used as it isn't thought to be efficient to tie up your PC's memory with pages of tables.

The tests will give results for **any** set of data but they will only have significance if the data is of a nature to which the test can be applied.

correlate(*first column number,last column number,label)*)

This finds the product-moment correlation coefficient between the data in the designated columns and stores the result with the specified variable name.

e.g. correlate(3,4,x) will find the product-moment correlation coefficient between the data in columns 3 and 4 and store it as variable x.

chi_test(*observed frequency column number,expected frequency column number,label)*)

This carries out a χ² test on the data in the first column by comparison with that in the second. The statistic, χ², is stored with the specified variable name.

e.g. chi_test(1,2,x) will compare the data in column 1 with the expected frequency of this data in column 2 and store the coefficient as variable x.

t_test(*column number,quoted mean,label)*)

This carries out the 'Student's' T-test on the data in the column, which is compared with the quoted mean. The result is stored with the specified variable name.

e.g. t_test(1,2.3,c) will compare the data in column 1 which may be part of a set which has a mean value of 2.3; storing the result as variable c.

m_w_test(*tails,end,first column number,second column number,label)*)

This carries out a Mann-Whitney test on the data where the acceptable values for **'tails'** and **'end'** are shown below.

Tails | |

Values |
Result |

one |
Applies a one-tailed test. |

two |
Applies a two-tailed test. |

End | |

Values |
Result |

high |
The high end of the distribution is used. |

low |
The low end of the distribution is used. |

e.g. m_w_test(one,high,1,2,a) will carry out a single-tailed test on the high end of the distribution of the data in columns 1 and 2; storing the result as the variable a.

Even though the 'end' parameter has no meaning in a two-tailed test, a value must be entered to avoid a syntax error. In fact, the algorithm assumes a symetrical distribution, carries out the single-tailed test selected and mutiplies the result by 2!