Here's how to compute your tables with SPSS for class assignments. There are other ways to do it, but this will give you guidance on one good way to do it. |

- Put the
independent variable (Gender) in the
**Columns**, and put the dependent variable ("Afraid to walk alone") in the**Rows**. - In the
"
**cells**" subcommand, tell SPSS that you want Observed Counts and Column Percentages. - In the
"
**statistics**" subcommand, tell SPSS that you want Chi-square; then -- If your data are ordinal and/or dichotomous (2-category, including dummy variables), as they are here, tell SPSS that you want Gamma or another ordinal statistic. This will be most common.
- If your data are nominal, choose Phi/Cramer's V or another nominal statistic. You won't use this as much.

V2R Gender (R) | Total | ||||
---|---|---|---|---|---|

1 Female | 2 Male | ||||

V6XR Afraid to walk alone at night (R) | 1 No | Count | 80 | 119 | 199 |

% within V2R Gender (R) | 36.7% | 63.3% | 49.0% | ||

2 Yes | Count | 138 | 69 | 207 | |

% within V2R Gender (R) | 63.3% | 36.7% | 51.0% | ||

Total | Count | 218 | 188 | 406 | |

% within V2R Gender (R) | 100.0% | 100.0% | 100.0% |

When we set things up this way, we can say that 63.3% of women are afraid to walk alone at night, compared to 36.7% of men. That is, women are much more afraid of this than men (26.6 percentage points more so).

Value | df | Asymp. Sig. (2-sided) | Exact Sig. (2-sided) | Exact Sig. (1-sided) | |
---|---|---|---|---|---|

Pearson Chi-Square | 28.583(b) | 1 | .000 | ||

Continuity Correction(a) | 27.528 | 1 | .000 | ||

Likelihood Ratio | 28.920 | 1 | .000 | ||

Fisher's Exact Test | .000 | .000 | |||

Linear-by-Linear Association | 28.512 | 1 | .000 | ||

N of Valid Cases | 406 | ||||

a Computed only for a 2x2 table | |||||

b 0 cells (.0%) have expected count less than 5. The minimum expected count is 92.15. |

The
Chi-square tests simply measure whether anything is "going on" in
the table. You'll learn more about Chi-square in your statistics classes.
All we need to look at here are the **significance** ("Sig.")
measures. If they are very **low**, as they are here (.000), we say
that Chi-square is significant.

The
usual "cut points" for
This is true for any statistic you use (not just Chi-square). |

Value | Asymp. Std. Error(a) | Approx. T(b) | Approx. Sig. | ||
---|---|---|---|---|---|

Ordinal by Ordinal | Gamma | -.497 | .078 | -5.540 | .000 |

N of Valid Cases | 406 | ||||

a Not assuming the null hypothesis. | |||||

b Using the asymptotic standard error assuming the null hypothesis. |

The Gamma
statistic shows whether there is a relationship between two ordinal variables,
as is the case here. Gamma here is -.497 ("**Value**"), which
is quite strong, and it is highly significant (.000) ("**Approx Sig.**").
Thus, for a strong relationship, the *Value is high* (absolute value),
and the *Significance is low*. Significance levels work the same
here as they do for Chi-square. Gamma ranges between +1.0 and -1.0. Zero
means no relationship, and the higher the absolute value of gamma (the closer
to +1.0 or -1.0), the stronger the relationship.

The **sign**
(+/-) simply shows which way the correlation goes: positive means that the main
diagonal of the table is dominant (upper left cell to lower right), and negative
means that the that the off-diagonal of the table is dominant (upper right cell
to lower left). Since gamma is negative here, it means that women are
more afraid to walk alone at night (you'll see this in the lower left cell of
the table).

The
strength of gamma,
These rules of thumb don't necessarily apply to other statistics. |

V2R Gender (R) | Total | ||||
---|---|---|---|---|---|

1 Female | 2 Male | ||||

V3R Trust people (R) | 1 Can't be too careful | Count | 134 | 90 | 224 |

% within V2R Gender (R) | 62.0% | 47.9% | 55.4% | ||

2 Other, depends (Volunteered) | Count | 17 | 29 | 46 | |

% within V2R Gender (R) | 7.9% | 15.4% | 11.4% | ||

3 Most people can be trusted | Count | 65 | 69 | 134 | |

% within V2R Gender (R) | 30.1% | 36.7% | 33.2% | ||

Total | Count | 216 | 188 | 404 | |

% within V2R Gender (R) | 100.0% | 100.0% | 100.0% |

In this table, notice that the dependent variable ("Trust people") has three categories, and that they are in order (the variable is ordinal). When rows and columns are set up this way, we can say that 62.0% of women say you can't be too careful, compared to 47.9% of men. That is, women are somewhat more distrusting than men (14.1 percentage points more so).

Value | df | Asymp. Sig. (2-sided) | |
---|---|---|---|

Pearson Chi-Square | 10.000(a) | 2 | .007 |

Likelihood Ratio | 10.044 | 2 | .007 |

Linear-by-Linear Association | 5.173 | 1 | .023 |

N of Valid Cases | 404 | ||

a 0 cells (.0%) have expected count less than 5. The minimum expected count is 21.41. |

The Chi-square here is highly significant (though not as much so as in the previous table). Thus, there is some relationship between the variables.

Value | Asymp. Std. Error(a) | Approx. T(b) | Approx. Sig. | ||
---|---|---|---|---|---|

Ordinal by Ordinal | Gamma | .214 | .086 | 2.436 | .015 |

N of Valid Cases | 404 | ||||

a Not assuming the null hypothesis. | |||||

b Using the asymptotic standard error assuming the null hypothesis. |

Gamma is moderate at .214 and is significant at .015. We can conclude that women are somewhat less trusting of other people than are men. Because gamma is positive, we should look in the main diagonal of the table (upper left to lower right cells) to characterize the relationship.