Learn moreLearn moreApplied Statistics Handbook

Table of Contents

 


 

Software Output Example

Raw data from1993 survey of U.S. adults

 

 

Crosstabulation: GUNLAW (Rows)  by   SEX (Columns)

Column Variable Label: Respondent's Sex

Row Variable Label: Gun permits

 

       Count |

       Row % |

       Col % |

     Total % |    Male |   Female |    Total

--------------------------------------------

       Favor |      314|      497|      811

             |    38.72|     61.28|

             |    73.88|     88.91|    82.42

             |    31.91|     50.51|

--------------------------------------------

      Oppose |      111|       62|      173

             |    64.16|     35.84|

             |    26.12|     11.09|    17.58

             |    11.28|      6.30|

--------------------------------------------

             |      425|      559|      984

       Total |    43.19|     56.81|   100.00

 

 

 

Chi-square                     Value         DF         p <

-----------------------------------------------------------

Pearson                       37.622          1      0.0000 A

Likelihood Ratio              37.417          1      0.0000 

Yate's Correction             36.592          1      0.0000 B

                   

 

 

Measures of Association

-------------------------------------

Cramer's V                   .196  C

Pearson C                    .192

Lambda Symmetric             .082  D

Lambda Dependent=Row           .000

Lambda Dependent=Column       .115  E

 

 

Note: 00.00% of the cells have an expected frequency <5

 

Interpretation

A  Statistically significant (most common measure used for significance)

B  When sample sizes are small, the continuous chi-square value tends to be too large.  The Yates continuity correction adjusts for this bias in 2x2 contingency tables. Regardless of sample size, it is a preferred measure for chi-square tests on 2x2 tables.

C  Weak association (both Cramer’s V and Pearson C)

D  A symmetric lambda is used when identification of independent and dependent variables is not useful

E  Knowing a person’s sex can reduce prediction error by 11.5%.


Google

 


Copyright 2015, AcaStat Software. All Rights Reserved.