4 Making sense of data

4.6 Line graphs

These are probably the graphs that you will be most used to seeing on an everyday basis. Line graphs are most suitable when you are just comparing one value as it changes with another value. They are less suitable when you want to look at several things at once; for example to study changes in oil prices and supermarket profits on petrol sales, the scales on the left-hand and right-hand sides of a graph would have to be different, and this can be very misleading (there is an example of this in Activity 16). However, you can see this quite often in line graphs in newspapers and magazines.

Activity 10

Look at the line graph entitled ‘Employment prospects’ and the accompanying article, then answer the questions that follow.

Source: Personnel Today 21 January 1999

  1. Overall, what is the graph trying to say?
  2. The graph joins up the lines between the years. How reasonable do you think this is?
  3. Where was the data in the graph derived from?
  4. How does it relate to the article?
  5. What conclusions can you draw about the use of the graph and related text?

Discussion

  1. This graph is quite difficult to interpret. It is a plot of the differences in opinion between employers who feel that the number of jobs they have will increase, minus those who feel that the number of jobs they have will decrease. In other words, it shows the differences between people who feel that their business is doing well enough for them to take on extra staff, and those who don't feel this confident.
  2. The data here should be presented as a series of points. There is no good reason for joining up the points, as has been done here. We can't tell anything from the data about the pattern of employers concerns in between the dates that are mentioned. In other words, there may have been a period of optimism about growth in jobs between 1992 and 1993, or pessimism between 1997 and 1998.
  3. An organisation called Manpower, which specialises in the provision of temporary staff to organisations. The article refers to agencies, and it is possible that Manpower used data from elsewhere as well as their own, but it's hard to tell. The article also mentions the number of people contacted. This seems like a high enough number for us to be able to rely on the conclusions. Organisations such as MORI that conduct opinion polls tend to use a sample of about 1000 people.
  4. It relates well, providing a summary of the information in the graph. The only query we would have is that the article says that a year ago there was a 10-point margin between the two groups, whereas the graph seems to show somewhere between an 8- and 9-point margin. This may be because the graph isn't very accurate, or it may be dramatic license on behalf of the authors.
  5. The 1999 data relies on the first quarter (and only the first part of that, as the article is from the 21 January), while the other years presumably are based on data from all four quarters. The article doesn't mention this at all. The other unknown here is the number of people that Manpower usually contact. It may be that 2195 is unusually high or low number and has distorted the eventual figures. The article also doesn't say how many of these 2195 replied. It could be all, 1000 or just 100. The actual number would probably affect our views of the reliability of the data.