Graphs Flashcards
What type of graphs do we have?
- Pie chart
- Bar graph
- Clustered/stacked bar graph
- Histogram
- Line graph
- Boxplot
- Scatterplot
There are three factors that come into play when choosing a type of visualisation.
- Purpose of the visualisation
- Number of variables
- Measurement level of the variable(s)
Pie chart
- Number of variables?
- Measurement level?
- Purpose?
Extra? When is it preferred over a bar graph? Under what conditions only?
Pie chart
- Number of variables = 1
- Measurement level = nominal/ordinal (categorical)
• Purpose =
display the distribution over different categories. This diagram is often used to clarify the share per category.
Extra? When is it preferred over a bar graph? Under what conditions only?
• There are not too many groups
• The graph stays easy to interpret
Bar graph
How to create?
- Number of variables?
- Measurement level?
- Purpose?
Why choose a Bar graph over a Pie chart?
Bar graph
Create = Put all values in a logical order on the x-axis (very unsatisfied - very satisfied)
- Number of variables = 1
- Measurement level = nominal/ordinal (categorical)
- Purpose = display frequency per category
Wherever you would use a pie chart, you can also use a bar graph. If you want to facilitate the comparison of shares within a category you are better off choosing a bar graph. People are better at comparing vertical bars than pie slices.
Clustered/stacked bar graph
- Number of variables?
- Measurement level?
- Purpose?
Clustered/stacked bar graph
- Number of variables = 2
- Measurement level = nominal/ordinal (categorical)
- Purpose = display frequency per category and per group
Histogram
What 3 things do you do?
- Number of variables?
- Measurement level?
- Purpose?
What is the difference between a histogram and a bar graph?
Histogram
3 things to do =
- Divide your data into fixed group
- You determine how many groups you want
- The bars are stacked against each other to show that the groups are connected
- Number of variables = 1
- Measurement level = Interval/ ratio (continuous)
- Purpose = display the frequency per range
Although a histogram tends to look a little like a bar graph, its makeup is quite different. In a histogram all the data are divided up into classes/ranges. So 1039, 1400 and 1650 are for instance automatically put together into the range labelled 1000-2000. On the y-axis you then find the frequency of observations that fall within a specific range.
Line graph
- Number of variables?
- Measurement level?
- Purpose?
Line graph
- Number of variables = 2
- Measurement level = Interval/ ratio (continuous)
- Purpose = to display a trend. For instance, in the number of violent incidents (per 100.000 inhabitants) during a specific time period.
Boxplot
- Number of variables?
- Measurement level?
- Purpose?
Boxplot
• Number of variables = 1 or 2
• Measurement level =
Variable 1 = interval/ratio (continuous)
Variable 2 = nominal/ordinal (categorical)
• Purpose = to display variation
Note: A boxplot displays the degree of variation within the main concentration of the data, the middle 50% of all values (the box). It also shows between which values the rest of the data are stretched out (lines and dots). Check the separate lesson on the boxplot if you’d like to know.
Scatterplot
- Number of variables?
- Measurement level?
- Purpose?
Scatterplot
- Number of variables = 2
- Measurement level = interval/ratio (continuous)
- Purpose = display the association between two variables
If you have 1 variable on a nominal/ordinal level what graphs can you use?
- Pie chart
* Bar graph
If you have 1 variable on an interval/ratio level what graphs can you use?
- Histogram
* Boxplot (if 1 variable)
If you have 2 variables on a nominal/ordinal level what graphs can you use?
- Clustered/stacked bar graph
* Boxplot (if 2 variables)
If you have 2 variable on a interval/ratio level what graphs can you use?
- Line graph
* Scatterplot
