Fault Finding and Fixing' Interpreting and 
Misinterpreting Data Tasks - Set #4 (solutions)


Malcolm Swan
Mathematics Education
University of Nottingham
Malcolm.Swan@nottingham.ac.uk

Jim Ridgway
School of Education
University of Durham
Jim.Ridgway@durham.ac.uk


Each question contains a selection of errors or misleading interpretations of data.


The aim of this assessment is to provide the opportunity for you to: 
· explain clearly the source of each error or misinterpretation. 
· rectify the errors and produce correct interpretations. 

____________________________________________________

1. Equal opportunities?

The table below shows the percentages of public school principals and teachers who were women 
in the period 1984 - 1991.

Women principals
Proportion of all public school principals

1984-85
21.4%

1987-88
24.6%

1990-91
30.0%




Women teachers 
Proportion of public school teachers 

1984-85
68.2%

1987-88
70.5%

1990-91
71.9%


Write a reply to the following statement, explaining carefully why it is misleading. Be concise but 
convincing (give mathematical evidence to support your claims).

"The proportion of women principals in our public schools increased by 40% in the period 
1985 to 1991, while the proportion of women teachers remained relatively stable. If these 
rates continue, it will soon be the case that one half of public school principals will be 
women and we will at last have equity in our school promotion system."



Solution:
The assertion that the proportion of women principals in public schools increased by 40% is based 
on the calculation: 
 

While this is correct, the increase in the proportions of public school principals that were women 
was a mere 8.6% (30% - 21.4%).

The language here is confusing as in the first instance we have a proportional increase expressed as 
a proportion of the original proportion!  It is easy to confuse one result with the other and this is 
potentially misleading.  Of course, these results give no information about the actual increase in 
number of women teachers as we don't know how many principals and teachers were employed in 
these periods.

It seems fair to say that the proportion of women employed as public school teachers remained 
fairly stable. 

It is impossible to extrapolate this information with so few data points. It seems that some progress 
has been made towards equity, but there is a long way to go. Even if the 50% was obtained this 
would not mean equal opportunities for women. For equity, one would expect the proportion of 
women principals to reflect the proportion of women teachers. We would want to see the 
proportion of women principals reach 70% before such a claim could be made.



2. Smoking 

 
The chart below resulted from a study of the smoking habits of men.

It shows data for about 1,000 men in each of four categories:  non-smokers, and those who smoke 
1 to 9, 10 to 39, or more than 40 cigarettes a day. It shows how many men would be expected to 
survive to each age.

For example, of 1,000 men aged 25 who do smoke more than 40 cigarettes per day, about 856 will 
survive to the age of 50.


Number of survivors

Age
Number of cigarettes smoked per day


Zero
1 to 9
10 to 39
More than 40

25
1000
1000
1000
1000

30
994
991
991
988

35
987
981
981
973

40
978
966
965
951

45
964
942
939
910

50
944
906
869
856

55
909
859
831
777

60
855
778
744
671

65
777
673
622
540

70
667
524
468
400

Source: E.C. Hammond, Journal of the National Cancer Institute, 43 (951-962) 1969.

Use the data in the table to write comments on the following four opinions.  You should try to 
reply to each statement as fully and informatively as possible.

a) I am 25 years old. I only smoke 5 cigarettes per day. 
 Smoking isn't going to affect me much at all.
 
b) I am also 25. I am  a heavy smoker (about 50 per day). I reckon that I might reduce my 
lifespan by  two or three years, but its not that much really.
 
c) I am 45 and smoke about 20 per day. I guess I stand about a 70% chance of reaching the 
age of 70. That is little different to a non-smoker.
 
d) This table alone proves that smoking is a cause of early death.



Solution:
Suitable replies to each statement might be:

a) I am 25 years old. I only smoke 5 cigarettes per day.
 Smoking isn't going to affect me much at all.
 
 The data shows that of 100 people like you who smoke 5 cigarettes per day, 524 will live to the 
age of 70. This compares with 667 for non-smokers. You may be reducing your chance of 
living to 70 from about 66% to about 52%, a difference of about 14%. Alternatively, 
inspection of the table shows that the same number of non smokers survive to 65 as light 
smokers to age 60.
 
 
 
b) I am also 25. I am  a heavy smoker (about 50 per day). I reckon that I might reduce my 
lifespan by  two or three years, but its not that much really.
 
 The same number of non smokers survive to 65 as heavy smokers smokers to age 55.  (Some 
students might use the spreadsheet to calculate life expectancy reasonably accurately).
 
 
 
c) I am 45 and smoke about 20 per day. I guess I stand about a 70% chance of reaching the age of 
70. That is little different to a non-smoker.
 
 Given no other information than the table, your chances of survival are 468Ö939 = 0.5 
(approx) or 50%. For a non-smoker, your chances would be 667Ö964 = 0.69 or about 70%. 
Quite a difference!
 
 
 
d) This table alone proves that smoking is a cause of early death.
 
The table alone does not prove that smoking is a cause of early death. There may be other 
associated factors involved. For example, those who smoke may also tend to lead less healthy, 
sedentary lifestyles and it may the lack of exercise that is the cause of early death.






3. College magazine

 

The following headline and chart appeared in the June 14, 1994 issue of USA Today newspaper.

Clinton approval rating up
 
Source: USA TODAY/CNN/Gallup Poll of 756 adults by telephone on June 11-12. Margin of 
error: ±4 percentage points. 


The accompanying story, entitled "With Clinton home, voters lighten up," read in part:

With D-Day observances over and President Clinton back home, voters' attitudes toward the 
president are settling down a bit. Now that attention is back on the economy, health care and 
crises in Bosnia and Haiti, a USA Today / CNN / Gallup Poll taken over the weekend [of June 
12] shows Clinton's job performance rating inching upward to 49%... It's an improvement 
from a poll taken [on June 6] as Clinton was in Europe marking the 50th anniversary of the 
Allied invasion of Normandy, which showed approval dropping to 46%...

Write a letter to the editor of USA Today explaining why the assertion "Clinton approval rating up" 
might be regarded as questionable or misleading. Be concise (editors prefer letters that are brief and 
to the point) but convincing (give mathematical evidence to support your claims).



Solution:
Student responses will vary considerably, but a central point should be a comparison between the 
poll's margin of error (±4%) and the magnitude of the change in popularity between June 6 and 
June 12 (only 3%).

The 4% 'margin of error' is usually a 95% confidence interval - that is to say,  we can be 95% 
certain that Clinton's popularity was between 42% and 50% on June 6, and was between 45% and 
53% on June 12.  Therefore, assertions such as "Clinton approval rating up" and "It's an 
improvement" are not a valid conclusions from the poll data. 

Students might note that it is possible (though not very likely) that Clinton's popularity could have 
actually decreased between the two surveys. In view of the margin of error, Clinton's popularity 
on June 6 could have been 50% instead of the poll's 46%, and his popularity on June 12 could 
have been 45% instead of 49%. While the poll indicates an increase by 3 percentage points, there is 
a small chance that the President's popularity might have actually decreased by 5 percentage points.  
Some students might calculate these probabilities accurately.

Here is a sample letter to the editor:

Dear Editor,

Your report of June 14 on the popularity of the President is inaccurate.  You report his 
popularity has gone up since the 6th of June. However, according to your own data, the 
change in the President's level of popularity is insignificant and should not be reported as 
an increase.

The margin of error is used to indicate that if another poll were taken at the same time as the 
original poll, we can be very certain that the difference between the results of the two polls 
would be less than the margin of error.

In this case the 3% difference between the results of the two polls is less than the margin of 
error. Even if these two polls were taken at the same time we could have expected this kind 
of discrepancy between them.  Therefore, there is a good chance that the difference 
between the polls is due to the nature of random sampling and not to changes in the average 
American's opinion of the President.

Sincerely,

Mary Q. Public






4. Accident Data

The following real data shows how the percentages of cars involved in traffic accidents is related to 
the speed at which they were driving. 

 

Explain why the following claim cannot be made from this data:

"The graph clearly shows that it is safer to drive at over 60 mph than to travel within the 
speed limit."



Solution:
The fact that a smaller proportion of accidents occur at extreme speeds (both fast and slow) is 
partly due to the fact that fewer cars travel at these speeds for any substantial period of time. 

In addition, the graph says nothing about the safety of driving at a particular speed. Two cars are 
rarely traveling at similar speeds when they collide - so how can we say which is 'safer'? 

The data therefore cannot support the claim.