In developing the hypotheses previous research regarding volume and type of noise, as well as type of performance task was taken into account. We have chosen to use loud music for noise as this has been shown previously to adversely affect performance. The type of noise used was a song as this type of noise has a regular repetitive structure similar to much of the noise we hear in real life. The type of task used was a word search as previous research has indicated comprehension to be primarily affected by noise.
The method chosen was a laboratory experiment. One reason for this choice is that it gives us greater control over extraneous such as age, gender etc. A further reason for this choice of method is that it is easier to implement control measures. As well as this a laboratory experiment allows cause and effect to be established and the standardised procedure makes the study replicable, giving higher ecological validity.
A repeated measures design was chosen for the study. An advantage of this design is that accounts for between subjects’ differences as the same subjects are used in each condition. Also fewer subjects are required.
The independent variable was noise or no noise. It was made operational using a stereo playing loud music for the noise condition. The dependent variable was performance. Performance was measured using two word searches (one for each condition); the number of words found in three minutes was recorded. A word search was chosen to measure performance as it tests comprehension, which previous research suggested to be affected by noise and also it does require a degree of thought and concentration.
The overall aim of the current study was to provide support for previous research findings that performance is adversely affected by noise, and to discuss the implications if this is true.
Hypothesis: Noise will have a significant negative effect on performance compared to performance in a quiet environment.
Null Hypothesis: Noise will have no significant effect on performance.
The sample of participants used in the study were college students, from Stockport, aged 16, 17 and 18. In all 22 subjects were used, of which 11 were male and 13 were female. The sampling method used was an opportunity sample.
The study took place in four different college classrooms. The two conditions were ‘noise’ and ‘no noise’. All subjects were asked to complete a word search in each condition. In the ‘noise’ condition each subject was given a word search to complete whilst music was being played at full volume in the background. In the ‘no noise’ condition subjects were asked to complete the questionnaire in silence. In each condition the experimenter handed out a word search to each participant face down. Once each participant had a word search the instruction was given to begin and a timer was started. After three minutes the instruction was given to stop. The word searches were then collected in. A note was also made of the number of males/females partaking. After the experiment all subjects were fully debriefed on what the experiment was investigating and thanked for their help.
- Each word search contained 24 missing words in total so the maximum possible words that could be found were the same.
- Two of the classes took part in the ‘noise’ condition then the ‘no noise’ condition. The other two classes took part in the ‘no noise’ condition first. This was in order to counter-balance to order effects, which may have been present, such as boredom or fatigue.
- The music in the ‘noise’ condition was played at full volume in order to ensure that the condition contained noise and not just music.
- The subjects were not allowed to confer whilst completing the word searches.
The materials used were two word searches and a music CD. The word searches can be seen in appendix A. They consisted of 24 words each. The music used was ‘Everything We Do’ by ‘Ja Rule’.
The statistical test chosen to analyse the data is the ‘Wilcoxon’ matched pairs test. The reasoning for this is that we have collected ordinal data; we are testing to see if there is a difference between two conditions; we are using the same subjects in a repeated measures design and the data collected is most suitable for non-parametric testing.
The main ethical consideration in the study was ‘protection of participants’ as we had to ensure that subjects did not come to any harm such as damaged eardrums from exposure to loud music. We did this by ensuring the music was played at a loud, but safe level. We also had the subjects consent and fully debriefed them.
The results mean that the experimental hypothesis, which stated that ‘noise will have a significant negative effect on performance compared to performance in a quiet environment’, has been supported and the null hypothesis rejected. Support for this has been gained through both descriptive and inferential statistics used to analyse the results. The pie and bar charts (page X) show visually how noise adversely affects performance. The Wilcoxon Statistical Test confirmed our results to be significant at the 5% level allowing us to accept our hypothesis.
Several criticisms can be made of the methodology of the present study. Firstly the repeated measures design meant that different materials had to be used for each condition and it is likely that these won’t have been identical in difficulty. What’s more subjects may not have been naïve for the second condition, which can complicate the results with demand characteristics. A problem with the opportunity sample used is that only college students were used, which ignores everyone else and is therefore not representative, limiting the generalisability of the findings. The procedure could be criticised for creating artificial conditions due to the high level of control in a laboratory experiment. Furthermore it could be said to be low in ecological validity as the nature of the task is unlike performance under noisy conditions in real life and music may not be typical of the type of noise experienced in many environments e.g. the office. Nerves in the laboratory conditions could also have affected the subjects’ behaviour. The location of the study was a classroom in the college, this proved to be suitable but made it difficult to implement the control measure of ‘no conferring’ in some cases.
A confounding variable later identified was the difficulty of each word search. We neglected to vary which was done in each condition. Therefore a variance in difficulty between the two could have affected our results.
A possible change to the method would be to vary which word search was done in the ‘noise’ and ‘no noise’ conditions, as this would improve the validity of the results. A further change could be to use a stratified random sample, which would more representative and generalisable results.
One ethical consideration in the study was to ensure appropriate music was used, that contained no explicit lyrics. Additionally we also had to keep the music at a level that could not damage the subjects’ hearing.
Our results lend support to the theory of von Wright and Nuimi, that noise adversely affects task performance, as we found performance on a word search to be negatively affected by noise. Our findings also lend support to the findings of Hockey, 1979 and Smith and Stansfield, 1986 as we too found comprehension to be adversely affected by noise, namely comprehension of words in a word search in the present study. Our study also suggests that Broadbent’s theory that sudden, loud intermittent noise can harm performance, needs to be modified to state that continuous, loud noise can be just as detrimental.
The current study has implications for the safety of people working in noisy environments. For example people operating complex machinery in noisy factories may not perform adequately and mistakes made could endanger themselves and others. The study also has implications for parents. It is clear that for their child to perform optimally at homework they need to be given a quiet home environment, not a noisy one.
Further study should now be carried out into other aspects of both noise and performance. Performance in real noisy conditions e.g. in a school dining hall, or performance with intermittent noise could be studied. Alternatively different types of performance could be studied, such as performance on more complex tasks e.g. IQ tests. The effects of the volume of the noise on performance would also provide a further useful insight into human behaviour.