Cite this page

Doing Statistics

Click on the image for a video guide. An extended text version of the same material, with linked screenshots, can be read below.

Introduction

For the most part, each of the 197,000 trials in the Old Bailey Proceedings contains information about defendants and victims (their gender and sometimes their age), and one or more types of crime, verdict, and, if found guilty, punishment sentence. Because this information has all been tagged and organised into categories, as explained in About this Project, each of these categories of information can be tabulated, and cross-tabulated against each other.

The Statistics Search Home Page.

The statistics search facility on the Old Bailey Online allows users to combine quantitative with qualitative analysis by counting cases according to criteria specified by the user, and then drill down to examine the individual trials which make up any cell. One can even compile a query concerning trials which include one or more specified keywords. Results can be displayed as tables, bar charts, or pie charts. For example, one can compile a table of the distribution of defendants' genders by type of crime in a very short time, something that would have taken a researcher months to achieve before the Proceedings were published online.

The statistics search page enables you to specify the information you would like to count and select the trials you wish to include. The search form may appear daunting, but not every box needs to be filled in and simple searches can be quickly performed. It is helpful to think of defining your search as a four step process.


The Statistics Search Home Page showing the pull down menu associated with 'Row'.

Selecting What to Count

First, you need to decide the category or categories of information you want to count by using the drop down menus in the row and column search boxes. Note that the difference between these two menus is that only the row column includes the choice of year and decade. If you wish to compile a table, your results will display most effectively if you choose the category with the least number of possibilities as column, and the category with the most as row. If you wish to compile a bar chart or pie chart, fill in the row box only.

Using the drop down menus, choose what you want to count, whether it is offences, verdicts, punishments, or defendant or victim genders or ages. If you choose category, this information will be divided into general categories, such as theft and killing for offences. If you choose subcategory, this information will be divided into more specific categories such burglary and infanticide. If you choose year or decade, the category of information chosen in the column box will be distributed chronologically in a table.


A table reflecting Defendant gender.

If you are only interested in determining the frequency of a single category of information, for example defendant gender, then do not fill in the column box. For example, a search for defendant gender under row will produce the following table, which shows that 78.2 per cent of all defendants in the Proceedings were men.


A table cross tabulating the gender of defendants and victims.

If you wish to tabulate one category against another, for example defendant gender against victim gender, then select the second category or subcategory from the drop down menu in the column box. This search will produce the following table, which shows that of the 253,385 defendants in the Proceedings, only 6,603 women were accused of committing a crime against another woman.


The Statistics Search Home Page showing the pull down menu associated with 'Count By'.

Count By

Second, you need to determine the unit of analysis you wish to count by. Do you wish to count defendants, victims, offences, verdicts or punishment sentences? Since some trials have more than one of any of these, results will vary. In deciding what to count by, you need to decide what you are interested in--if you are interested in the characteristics of people, count by defendants or victims; if you are interested in crime, count by offence; and if you are interested in the decisions made by the court, count by verdicts or punishments. If you fail to specify the right type, you may not find the trials you are looking for. For example, if you are interested in murders committed by women, you must count by offence in order to find the 1733 trial of Sarah Malcolm; if you count by defendant the trial will appear under the category multiple offences rather than killing, since she was also accused of burglary.

The default setting will choose the most appropriate category depending on your query and is normally sufficient. The one unit of analysis you cannot count by using this facility is trial. The forthcoming Old Bailey API will allow you to undertake this type of search.

Table, Bar Chart or Pie Chart?

Third, you need to decide how you would like your results displayed.

The Statistics Search Home Page showing the pull down menu associated with 'Display'.

The default setting is table, which is appropriate for all types of query. But if you are only counting a single category of information, such as punishment sentence, you can display the results visually, in a way which makes the results much easier to interpret, using a pie chart or a bar chart. The following bar chart shows the distribution of punishment categories. Note that you need to scroll down the page to match the colours to the labels, and see the numbers of cases in each column.


A bar chart and table reflecting punishment categories.

It can be seen that imprisonment was by far the most frequent type of punishment sentenced over the whole period from 1674 to 1913.


Limiting your Search

Finally, you have the option of limiting the trials to be counted according to criteria you specify.

The Statistics Search Home Page showing the different criteria with which you can limit your search.

Using the drop down menus, you can choose to limit what is counted to any combination of a specific offence, verdict, punishment sentence, defendant and victim age and gender, and time period. For example, if you choose to narrow the previous tabulation of punishment categories to the offence subcategory robbery and the period up to 1800 (by selecting 1800 in the drop down menu for the To (month/year) box, you will generate a bar chart which looks very different.


Bar chart reflecting punishment categories for robbery before 1800.

Now death (the green bar) is the most common punishment sentence, with many convicts also sentenced to transportation (the pink bar), reflecting the frequent use of these two sentences for those convicted of the most serious crimes in the eighteenth century.


The Statistics Search Home Page showing the 'Keyword' search box.

Using Keywords

You can also narrow your search only to trials which include one or more specified keywords. For example, if you specify sword in the keyword search box and repeat the previous search, you will generate a table which once again looks very different.


A bar chart reflecting punishments specified in trials for robbery where the word 'sword' can be found prior to 1800.

This chart displays the punishment sentences for those convicted of robberies, in trials where the accounts include the keyword sword before 1800. Since such trials often involved violence to the person, it is not surprising that almost all convicts were sentenced to death.

Once you have completed these four steps, click generate to complete your search.


Detail of a table of results.

Interpreting Your results

The two most important points to note are: first, that where two percentages are provided in a cell, the lower left percentage is the percentage of cases in that row, and the percentage on the upper right is the percentage of cases in that column. In choosing which percentage is relevant, ask yourself what you need to know: is it the proportion of all cases in the row, or proportion of all cases in the column? Second, each absolute number is a link which takes you to a listing of all the trials which make up that cell. This makes it possible to read the relevant trials and assess the significance of the numbers generated by the statistics search. In this way, users of the Proceedings can combine quantitative and qualitative research.

Table of results with the Export facility indicated.

For further statistical analysis, you can export your results into a database package such as Excel by clicking on the export button at the bottom of the results page.

It is often said, paraphrasing Mark Twain, that is is easy to lie with statistics. Numbers and percentages can certainly mislead, but if you take care to define your search query precisely, and think about the significance of the percentages obtained, you should be able to generate meaningful results. For further information on interpreting statistics, see the Teaching Statistics section of the guide to Using the Proceedings in University Teaching.

Additional information about doing statistical searches is available by clicking on the help icons next to every box on the search page.