Tables
Graham Tall     research@grahamtall.com      September 2003

I.    Introduction

Tables are a dissertation’s work horse because they have the critical advantage of supplying substantial information in compact form.  Unfortunately, they are often inserted with little real thought.  They need to be prepared with as much, if not more, care as a paragraph which has been re-worked a number of times so that its meaning is crystal clear.

a) The title must be accurate and easy to understand.
b) The justification for axes not starting at zero must be clear.
c) The column and row information titles must be easy to interpret.
d) The totals on which percentages have been calculated must be available so that such percentages do not imply
        large differences if, because the totals are small, the differences are small.
e) The relative positions of the columns and rows must be helpful.
f) Whenever necessary, clear explanatory footnotes must be available.

The data in table 1 was collected a few years ago by Birmingham Local Education Authority:

Table 1 Number of Students from Different Ethnic Groups In Birmingham FE Colleges

The initial reason for collecting the ethnic data in the colleges must have been to enable the reader to recognize that
the overall proportions of the ethnic groups varied substantially in the different colleges. But,

i. The size of the F.E. colleges varies enormously. In the original document the LEA document attempted to
        help the reader take account of this by providing the percentage of each ethnic group in each College. But,
         providing twice as many numbers in the same table hardly made it easier to read.

ii. The order of the colleges is alphabetical. Whilst this order enables the reader to rapidly check the information
        in a specific college it does not allow the colleges to be compared easily.

II.    Re-Organising the Table to Make it Easier to Interpret

The logic of doing this is to help the reader understand the researcher’s argument, hence it is necessary decide the purpose of including the table.   Consider a table designed to illustrate the balance of the ethnic groups in the Colleges:
    i)    What categories are really useful?  Is it worth retaining keeping small numbers like - the number of Irish separate?
    ii)    With Colleges differing so much in size it is worth  making comparisons easier by calculating the percentage of 
           each group in each college.
    iii)    The first table ordered the Colleges simply by alphabetical order.  Use data which considers, say,  the
            numbers/proportion of white/UK students. 

Note: In such an analysis the researcher is doing the exact equivalent, though in number rather than prose format, of the writer who carefully re-writes a passage, transposes phrases, selects more descriptive and useful words etc. in order to help the reader understand the underlying message more clearly

In table 2, the number of categories has been reduced to emphasise the three major ethnic groups in Birmingham: the Afro-Caribbean, the Asian and the White/UK (the rather clumsy White/European seemed rather strange when it excluded the Irish). The subsequent decision to order the colleges in the table 2 with respect to the overall number of blacks is, in research terms, completely arbitrary since it could, just as easily, and might more logically, have been done using the number of white students (logically because the great majority of students are white). Certainly, the advantages of retaining an alphabetical order when there are just 8 colleges is minimal. The value of ordering the colleges using a major research characteristic is, however, immediately apparent:

Table 2 NUMBER of Students from Different Ethnic Groups In Birmingham FE Colleges.
                (Combined Categories)   The yellow columns would probably not be included in body of thesis.

Not only is it evident that very large numbers of black students studied at Handsworth, but it that Handsworth and the College of Food are atypical. What the table cannot explain is why the ethnic distribution of the latter two colleges differs. For that level of insight, additional knowledge on ethnic population distribution and on the range and type of courses offered is essential. In neighbourhood terms, Handsworth is the most convenient college for students of Afro-Caribbean and Asian descent. It is worth noting, however, that public transport policies in the West Midlands mean that it is practicable for students to cheaply travel to colleges that are some distance away. The very different ethnic distribution in the College of Food is almost certainly related to the highly specialised courses it offers.

Table 2's major disadvantage is that it is difficult for the reader to calculate the relative proportion of the different ethnic groups. Difficult because the colleges vary substantially in size. To overcome this the information is transformed, in table 4, into the percentage of each ethnic group students attending the particular college. Using the percentage of black students as the criterion for ordering the rows changes slightly the sequence of the colleges from that found in table 2. N.B. An added advantage of using percentages is the fact that the order of the colleges remains the same whether or not the order is based on the overall percentage of black or white students.

Table 3      Percentage of Students in Birmingham FE College in each Ethnic Group.
                (Colleges reorganised according to their overall percentage of Blacks)
                 The yellow columns would probably not be included in body of thesis.

To see this data represented by a series of barcharts click Barcharts

If categories are combined  in your dissertation, do remember to safeguard the right of the reader to re-interpret the data  by including tables of raw information in an appendix.

Home Page    Research Introduction    Quantitative Advice   Index   Statistical Tests    
Research and Statistics Courses