Pages in this section:

4οΈβ£ Section 4: Analysis

π Analysing a file

**π The Library**

**π The Analysis tab**

**π The Report tab**

**π· Screenshotting your maps**

β¨ Filters: Tracing paths

β¨ Filters: Zoom

β¨ Filters: Focus or exclude factors

β¨ Filters: Top factors and links

β¨ Filters: Combine opposites

β¨ Filters: Remove brackets

β¨ Filters: Collapse factors

β¨ Filters: Include or exclude hashtags

β¨ Filters: Autocluster

π¨ Formatters: Translation

π¨ Formatters: Colour links

π¨ Formatters: Link label position

π¨ Formatters: Labels - Surprise

π¨ Formatters: Labels - Tally

π¨ Formatters: Colour factors red

**π The Links Table**

**π₯ The Sources Table**

π The Factors table

**π The Statements Table**

**π¬ The Mentions Table**

**β The Questions Table**

βοΈ

**The Closed Question Blocks Table**π Comparisons

All sections:

**The Analysis tab**

Vignette goes here!!!!!!!!!!!!

### Differences - questions

Shows which links were preferentially mentioned in answer to different questions.

#### Differences - sample

Shows which links were preferentially mentioned according to different groups e.g. women more than men. We ask:

does the proportion of women vs men who mention this link differ from what you would expect (given the total number of mentions of links by both women vs men)?

γ
€ | Women | Men |

... other links ... | γ
€ | γ
€ |

Number of mentions of the link from X to Y | 10 | 9 |

... other links ... | γ
€ | γ
€ |

Total number of mentions of any link | 60 | 10 |

In this case we can see that although women mentioned the link slightly more often than men, women altogether mentioned links twice as often as men. So we can compare the number of mentions of the link with the number of "non-mentions" of the link. So we can work out this table (not shown).

γ
€ | Women | Men |

Number of mentions of the link from X to Y | 10 | 9 |

Number of mentions of any other link | 50 | 1 |

We can do a simple chi-squared test on this table to see if the ratio 10:9 is significantly different from 50:1 (which of course it is) -- this is the same question as to whether 10:50 is significantly different from 9:1 (which of course it is). If this test is significant, the row "Number of mentions of the link from X to Y" is shown in the table, and the intensity of the colouring of each cell reflects its chi-squared residual, i.e. how different is the number it contains from the number you would expect, given the other numbers?

This comparison is agnostic as to whether there are, say, many men or a few men who talk a lot.

At the moment, the tests for this are chi-squared tests which would not give special treatment groups which are actually ordinal e.g. low income, medium income, high income: the chi-squared test is weaker than it should be.

Β

Β