Method
Aim
My aim was to find out whether using categorisation and hierarchy improves recall.
Experimental hypothesis
The experimental hypothesis is that the group using categories and hierarchy will recall more, compared to the group with a list of words with no structure.
Design
I have used an independent design for my research. I have also used a matched pairs design.
My participants have been selected using Opportunity sampling.
I have used a laboratory experiment.
Apparatus
- Pens in case participants cannot provide their own
- 24 pieces of blank paper
- 12 copies of organised and categorised words
- 12 copies of disorganised words
- Standardised instructions
- Room to do experiment in
Procedure
- Using opportunity sampling, select 24 participants.
- You will need a room for all of your participants, divide the room so that half is doing one test with structured and categorised words and the other half the test with unstructured words.
- Have the two different sets of words and blank paper set up ready on the tables on each side of the room and ensure all participants have a pen.
- Start by reading the participants your standardised instructions; tell them that there is to be no conferring and no cheating.
- Give the participants 5 minutes to memorise the words.
- After the 5 minutes are up, tell the participants to lay the words face down on the floor.
- Then give 5 minutes for the participants to write down as many words as they can remember.
- Collect in all the sheets of paper and fully debrief all of the participants.
Results
Results
Descriptive statistics
Group 1
Mean: 16.91666r
Median: 18
Mode: 18
Range: 22
Group 2
Mean: 27.5
Median: 14.5
Mode: 16
Range: 7
Graphs
Mann Whitney U test
42, U is not significant
Conclusion from results
The Mann Whitney u test was chosen as the independent measures design and was combined with internal data, treated as ordinal data, since the criteria did not match the requirements for a parametric test. I used a one tailed 0.05 test. In the Mann Whitney U test, U is not significant therefore we cannot reject the null hypothesis. Overall, my results are not significant, therefore it is not true that the group using categories and hierarchy will recall more, compared to the group with a list of words with no structure.
Discussion
By looking at the number of words remembered it is clear to see whether this is an indicator for memory. It is clear that the more words you recall the more words were remembered. In this case the experiment did not work the way expected. My results were not significant in the Mann Whitney U test.
If we look back to Bower, in his study group 1 recalled 65% of the words and group two only recalled 19%. His results were significant. Which therefore implies that hierarchy and structure improves recall, as my experiment does not, there are many reasons why I have different results from Bower.
Construct validity is whether the test can be used to support the variable that is supposed to be measured. If the experiment was replicated, it is likely similar results would appear. Ecological validity is whether the experiment measures a naturally occurring behaviour. Since this was a laboratory experiment it lacks ecological validity, as it is not often that someone would be taken into a room and asked to participate in a test in a daily life setting. Therefore the results may have been affected by demand characteristics. Suggestions for improving validity are, the experiment lacks ecological validity, as it is a laboratory experiment, if the experiment was conducted as a field experiment there will be higher ecological validity which could also lower the possibilities of experimenter expectancy and demand characteristics. Asking the participants to remember a list of ingredients for baking a cake at school could increase the ecological validity. The results would be more accurate although it would be very hard to control any extraneous variables and it is not easy to replicate the study. Another improvement in the study could be to use sentences instead of words for participants to recall. This is because generally people do not memorise lists of words but phrases or sentences. For example, when students study at school they learn material from books, which tend to be in sentences rather than a list of words.
The reliability of the experiment means whether the method of measuring can measure consistently. If the experiment were repeated, similar results would appear. Reliability has increased due to the same words being used in both lists. If there were two different lists of words then there are possibilities that some words are easier to remember than others. Therefore using the same words will reduce this effect. In the list of organised words it was obvious that the experiment was something based on memory therefore there were some demand characteristics. All procedures and lists of words and instructions were standardised therefore the study is easy to replicate. However, with regard to informed consent, the participants were already told that they were taking part in an experiment on memory although not in detail or told that it was a psychology experiment. Therefore participants could have worked out what the experiment was about and tried harder to perform better on the test (demand characteristics).
To improve reliability, if a different sampling method had been chosen, the results would be much more representative. This is because opportunity sampling only considers people who were ‘free’ at that moment. If a stratified sample was used the reliability of the results would increase, since there are different levels of cognitive abilities in students and not only people who were ‘free’ at the time. This method could be done by picking 10 males and females randomly from each year group. This would mean a total of 140 subjects would be used and therefore would be a much more representative sample, however this is not necessarily realistic.
Bousfield found that we have semantic organisation in our long-term memory. Bower found that organising a categorised hierarchy would help to improve recall. In this study it was found that participants showed no significant improvement in recall when the words were organised. This showed that the finding of my experiment do not support Bower’s findings. There wasn’t a significant difference between the experimental and control. This was likely due to small numbers of participants that were conducted on.
Target population is the age and group of people the experimenter plans to generalise the findings on. In this experiment the target population was sixth formers at St. Paul’s school. It was hard to generalise due to the method of opportunity sampling. This method was biased because only students who were available participated in the experiment. It could be improved if a wider range of students and not only people who were ‘free’ to participate therefore this was not a representative sample. The number of participants who took part was only 24. This was too little to generalise to a school of 1080 students. It was hard to generalise beyond the target population, as there are individual differences, psychological differences and cultural differences between much of the population. In addition my sample was too small to generalise beyond target population.
If my results had concluded that an organised list of categorised words would be more efficient to remember than a randomly placed list of categorised words, this could be applied to everyday life. For example, in the education system, when teachers teach children they have to teach in a systematical order so it is easier to recall the mass amount of information. As for high school there is a syllabus. It is organised by categorising the same type of information together. This would be the most efficient way for remembering information and recalling it for the exams.
References
- Journal of verbal learning and verbal behaviour ch8 p 323-343 (1969)
- Journal of experimental psychology human learning and memory 1978 vol 4 no 5 p 486-497
- The journal of general psychology 1953 49 p 229-240
Appendix
Descriptive statistics
Group 1
Mean: 5+7+13+14+15+18+18+19+21+22+24+27=203/12=16.91666r
Median: 5,7,13,14,15,18,18,19,21,22,24,27 = 18
Mode: 5,7,13,14,15,18,18,19,21,22,24,27 = 18
Range: 27-5=22
Group 2
Mean: 9+9+11+14+14+14+15+15+16+16+16+16=330/12=27.5
Median: 9,9,11,14,14,14,15,15,16,16,16,16 =14.5
Mode: 9,9,11,14,14,14,15,15,16,16,16,16 = 16
Range: 16-9=7
Mann Whitney U test
Step 1:
Step 2: Na=12 Nb=12 NaNb= 12x12=144
Step 3: 12+1=13
Step 4: 13x12/2=156/2=78
Step 5:
Step 6: 16+15+14+13+12+11.5+11.5+8+6.5+5+2+1=146.5
Step 7: (144+78)-146.5=75.5
Step 8: 144-75.5=68.5
Step 9: U=68.5 U’=75.5
Step 10: 42, U is not significant