The analysis of quantitative evidence:-The link between smoking and drinking behaviour

University Degree Education and Teaching

Soci215 The analysis of quantitative evidence:-The link between smoking and drinking behaviour

The following analyses aim to investigate the association between smoking and drinking among undergraduate students. It is hypothesised that those students who regularly drink large amounts of alcohol would also be likely to be smokers. Analysis is based on data from the ‘Health Questionnaire’, which was distributed in a second year lecture. The total sample size is 221 students and consists of students taking the ‘Quantitative methods’ module in 2000, 2001 and 2003. The data collected is likely to be reasonably representative of the population of students studying in the Sociology department. However, the sample will be slightly biased towards students who attend lectures.

As part of the Health Questionnaire students were asked ‘How many units of alcohol do you usually drink in a week?’ In the questionnaire a unit of alcohol was defined as equivalent to half a pint of lager or one glass of wine. As can be seen from the histogram displayed as Figure 1, the distribution resembles a slightly positively skewed, gaussian or ‘normal’ distribution. The mean was 23.4 units of alcohol per week and the median was 20 units of alcohol per week. In other words, while half the sample drink 20 units of alcohol or less per week, half of the students in the sample drink more than 20 units of alcohol per week. The range was from 0 units of alcohol per week, (with approximately five percent of students recording that they were non-drinkers) to 100 units of alcohol per week. There were just 4 students (1.8% of the sample) who recorded that they drunk as much as 100 units per week.

The questionnaire also asked students whether they currently smoke. A total of 31.7 students answered this question positively, corresponding to 70 students out of a total sample of 221 students. In other words approximately a third of students currently smoke.

This is a preview of the whole essay

Before examining the possible association between smoking and drinking behaviour, the continuous variable, measuring units of alcohol drunk per week, was recoded into a categorical variable. This makes it possible to construct a simple contingency table to look at the relationship between smoking and drinking among students. Those who drunk 10 units of alcohol or less per week were classified as light drinkers, those who drunk between 11 and 30 units of alcohol per week were classified as ‘medium’ drinkers and those drinking more than 31 units of alcohol per week were classified as heavy drinkers. Clearly, given the skewed nature of the data, the heavy drinker category is relatively heterogeneous as it covers the range from 31 to 100 units of alcohol per week.

In order to examine whether there was any association between smoking and drinking behaviour, the SPSS package was used to construct a contingency table showing the percentage of smokers and non smokers who are light, medium and heavy drinkers.

Table 1 shows the results of this analysis. It can be seen that in this analysis ‘Currently smoke’ was treated as the independent variable with ‘Type of drinker as the dependent variable’. However, given the nature of these two variables it is not possible to make a strong case either that drinking large quantities of alcohol causes people to become smokers or that being a smoker has an effect on alcohol consumption. Rather the interest in this analysis is on the association between the two variables.

From Table 1 it can be seen that there is a link between smoking and drinking in that whereas 30.0% of smokers are heavy drinkers, only 11.3% of non-smokers are heavy drinkers. In other words, smokers are almost three times as likely to be heavy drinkers as non-smokers. This is illustrated in Figure 2.

In order to test whether this association was statistically significant the Chi-square statistic was calculated using the SPSS program. This is the appropriate inferential statistic to use when investigating the association between two categorical variables. Although the sample is relatively small (N=221) the expected frequency in each cell of the table was more than 5 so the sample is large enough for Chi-square to give meaningful results. The Chi-square for this contingency table was 13.7 with two degrees of freedom and an associated probability p of 0.001. As p is clearly less than 0.05 in this case, the Chi-square is said to be significant. This means that the observed association is very unlikely to have occurred by chance i.e. due to sampling error. In other words we can be reasonably confident that the association observed in the sample is a real association that would also be observed across the whole population of second year sociology students.

1000 words