7. Measures 6. Measures

5. Measures

of of

of variation

correlation/ difference/

association significance

Figure 9.2 “ A summary of our overview of ten levels of data analysis.

Step one: assembling and understanding the structure and pattern of

raw data

Table 9.5 shows the types of products used by a sample of businesses. In

Table 9.6 we show the response given by the business sample to certain attitudinal

questions on business strategy.

Step two: ordering the raw data

In Table 9.7 we have ordered the tables on banking services used. For the

importance ratings, we have elected to streamline, by ordering attitudes on

the basis of the total percentage that ¬nds each service ˜at all important™ (see

Table 9.8).

From this reordering, we discern a general ˜inward focus™ amongst businesses:

consolidating and making the most of what they have, before considering

˜outward looking™ strategies.

Step three: introducing cross tabulation to examine sub-group variations

In Table 9.9 we look at the ordered data by the turnover of the business. This

immediately tells us that the general usage of products increases with size (and

implied complexity) of business. Thus, the combination of ordering vertically by

general prevalence, and ordering horizontally via the turnover cross tabulation,

generates a clear storyline along the ˜diagonal™ of the table.

In addition, a cross tabulation of the data by sector reveals a slightly different,

less clear-cut subplot (see Table 9.10).

116 Organizing the quantitative data

Table 9.5 “ Products used by businesses

TOTAL

Base 1,100

Q: Which of these banking services does your business use? (%)

Telephone banking 7

Deposits of cash and cheques 93

Euro accounts 5

BACS credits and debits 73

CHAPS payments 30

Payroll 25

Electronic and PC banking 25

Overdrafts and short-term ¬nance 77

Term loans exceeding three months 52

Asset ¬nance and leasing 15

Factoring/invoice discounting 3

Currency accounts 14

International payments 22

Foreign exchange 19

Deposit and short-term investment accounts 42

Treasury services 15

Investment banking services 5

Vehicle management services 1

Step four: measures of location

We now examine three key elements of the attitudinal data. We see that the

normal measures of location (mean, mode and median) provide us with an option

to summarize the data differently for the ¬nal presentation context. We have

assumed the following scale values: very important = 4; quite important = 3;

not particularly important = 2; not at all important = 1 (see Table 9.11).

Step ¬ve: measures of variation

We now need to comment on the variation in our data set. In cases A and B in

Table 9.11, the normal measures of variation work well, i.e. the standard deviation

(average distance of the population or sample values from the mean) and variance

(the average squared distance of all measurements from the mean). In case C,

though, caution must be exercised in using any measure of location, because

the distribution of responses does not approximate the normal distribution.

Diagnostic statistics, such as standard deviation and variance, supplement our

intuition formally: clearly we can be con¬dent in the location measures on

statement B, moderately so for statement A, but must treat these measures with

great caution for statement C (see Table 9.12).

117

Understanding the hierarchy of data analysis options

Table 9.6 “ Importance ratings

TOTAL

Base 1,100

Q: How important are each of these to your business strategy? (%)

• Diversify into new sectors or markets Very important 23

Quite important 33

Not particularly important 25

Not at all important 19

• Grow the business amongst existing customers Very important 32

Quite important 36

Not particularly important 24

Not at all important 8

• Check that your customers are satis¬ed Very important 30

Quite important 64

Not particularly important 4

Not at all important 2

• Improve internal cost ef¬ciencies Very important 36

Quite important 27

Not particularly important 20

Not at all important 17

• Invest in marketing and promoting the business Very important 16

Quite important 34

Not particularly important 27

Not at all important 23

• Avoid risk Very important 46

Quite important 36

Not particularly important 15

Not at all important 3

• Use the latest technology Very important 21

Quite important 35

Not particularly important 31

Not at all important 13

• Focus on international markets Very important 37

Quite important 13

Not particularly important 15

Not at all important 35

Step six: measures of signi¬cance and difference

We now extend our analysis to apply the standard 95% con¬dence interval

procedure to the difference between two percentages. A rule of thumb will

usually suf¬ce. We revisit the sector data to ¬nd that there are signi¬cant

differences across sectors (see Table 9.13). These are summarized in Table 9.14.

118 Organizing the quantitative data

Table 9.7 “ Services ordered from high to low

TOTAL

Base 1,100

Q: Which of these banking services does your business use? (%)

Deposits of cash and cheques 93

Overdrafts and short-term ¬nance 77

BACS credits and debits 73

Term loans exceeding three months 52

Deposit and short-term investment accounts 42

CHAPS payments 30

Payroll 25