Rationale
This experiment was undertaken to discover whether “chunking” leads to better recall.
Miller suggests that chunking leads to improved recall. This could be tested by using related and unrelated words lists. The related words could be on any subject like clothes but all the words must all be related. The words in the unrelated list could be anything but must be the same in some respects e.g only one syllable.
In the related list participants should recall late words best, early words should be recalled next best and middle words should be recalled least well. However when using the related words list where chunking should hopefully take place, more words should be recalled from the middle of the list than in the unrelated list. This is because words are put into groups or chunks, so their position in the words list does not affect the chance of being recalled as much as in the unrelated list. Another factor which could affect the serial position curve is cue-dependent remembering. This could only have an affect in the related words list as one word could cue another word to be recalled.
Aims
The broad aim of this study is to discover whether the human mind is capable of storing more information if the brain is able to chunk then when the brain is unable to chunk.
The aim of the experiment is to discover whether participants will recall words better from a related word list than from an unrelated word list.
To carry out the experiment the word lists will be either related (in this case the category will be cloths) or unrelated (completely random words). To measure the experiment the number of words recalled will be recorded.
Hypotheses
Experimental hypothesis: There will be a significantly greater recall of words from the related list than those from the unrelated list.
Null Hypothesis: There will be no significant difference in the number of words recalled from the related and unrelated lists; any difference will be due to chance.
Methodology
Method
The method chosen for the investigation was a laboratory experiment.
This method was chosen for many reasons:
The laboratory experiment provides the psychologist with the highest possible level of control. The situation can be created and other methods such as field experiments do not have such high levels of control. As well as this laboratory experiments produce results which are numerical, so results are easy to analyze.
Another reason for choosing the laboratory method is that none of the other methods could be used to acquire the information needed, as questionnaires, case studies and interviews could not reveal such data.
Design
Independent measures design was used to eliminate order affects. Repeated measures design could cause order affects such as boredom this would decrease frequency of recall due to lack of motivation.. More importantly participants could get better because of the practice effect, which would increase frequency of recall. Matched pairs would be a good technique but it is far to time consuming for this experiment as a profile of each person has to be made. Order affects could mean that the experimenter could not be sure whether the change in recall was due to chunking or the order affects.
Variables
The independent variable in the experiment was the related or unrelated word lists. One list of words was related in this case all clothes, and the other word list contained completely unrelated, words with no connection.
The dependent variable in the experiment was the number of words remembered by a participant.
Sampling
Opportunity sampling was used in this study for convenience other samples were not used. Random sampling could involve traveling and is difficult to organise, so this was not an appropriate method. Systemic sampling and quota were other options, but opportunity was more convenient as it much faster and there were time constraints.
Participants
Because opportunity sampling was used, most of the participants were friends and family members. This meant participants were of both sexes and there was a higher percentage of 16 to 18 year olds because much of the sample was taken at school. The age range was from 12 years old to 52 years old, which is a range of 40 years. There were a slightly higher percentage of males, this was purely due to chance.
Apparatus
A computer was used to print of the words onto cards, this ensured the font was the same on each card and they could all be in higher case. Scissors were used to cut the cards so that they were all the exact shape and size. Thick card was used instead of paper so that the cards would be hard wearing and re usable. Pen and paper were also necessary for the participants. It was decided that this apparatus should be used because it filled all of the necessary criteria, such as complete homogeneity for easy replication.
Procedure
To carry out the experiment the experimenter approaches a potential participant and asked whether they would like to take part in an experiment. If the participant agreed the experimenter would lead them to a table and ask them to be seated. The experimenter would then slowly show the participant the list of related or unrelated words, making sure that the participant had 5 seconds to read each word properly. Once the experimenter had shown the participant the whole list, they would place the card out of view, wait for half a minute before asking the participant to write down as many words as they can remember on a sheet of paper. After the experiment is over the participant is debriefed and told the actual aim of the experiment.
Controls
To ensure the unrelated word list was completely unrelated it was repeatedly revise and modified to create a standardized word list. There were also standardized procedures. Each participant was shown the cards in the same order and at the same speed to make sure the experiment was the same each time. Standardized instructions were also given to each experimenter.
Participants are not told what the experiment is actually about, at the beginning they are simply asked if they would like to take part in an experiment. This could be seen as unethical as the participants are not told the true aim, however participants are not harmed emotionally, physically or otherwise as a result of the experiment. Participants were also debriefed.
Results
Summary table of words recalled in each condition
Summary table commentary
The summary table shows that over all there were far more words recalled from the related list, than from the unrelated list. The mean number for the related list is considerably higher than the mean number for unrelated data. This means that on average more related words were remembered. The fact that the mean, mode and median are very similar or the same as the each other suggests that in this case all of these measures are representative of the data collected. The variation ratio is higher for the unrelated data then for the related data. Also it is tri-modal. This suggest that the mode for the related data is more representative of the data that that of the unrelated. The range for both conditions was the same and quite large. This suggests that the median is not very representative. The interquartile range is also very useful, as extreme scores at either end of the data do not affect it. In this case it shows that the data is fairly closely grouped which suggests that it is representative. However when extreme scores are allowed, the median is representative. The range for both conditions was the same.
Additional graphic representation of results (Related Unrelated)
Description statistical commentary
This graph clearly shows that there is a dip in the frequency of recall of words in the middle of the words lists. The dips in the middle of these graphs are probably the result of the serial positioning affect where information is remembered from the beginning and the end but less in the middle due to primacy, recency. The dip in the related list is less distinct then that of the unrelated list, this is probably due to chunking. This was predicted in the hypothesis. This could be due to overall word recall in related list is more accurate. Also when a participant is recalling related words one word may cue another for example: The ‘word’ boxers may cause the participant to remember the word ‘briefs’. This could also have been a factor in this. Both graphs are undulating; this could be a result of specific words being more memorable than others regardless of their position in the word list. The graph for the related condition is far shallower then the graph for the unrelated words list. This is almost certainly due to chunking. There were a few anomalies; one of them was the word ‘bush’. This could have been due to its comic value or even down to current events, such as president Bush visiting during the experiment.
Inferential statistics
For the Mann-Whitney U test the results were:
Na= 20 Nb= 20
U=354.5 U-= 45.5
Therefore the result of our mann-Whitney U test was significant to less than or equal to 0.005, which equates to 0.5%. Therefore the percentage probability of the results being due to chance rather than the independent variable is 0.5%. This means the results are 99.5% due to the investigation. Therefore the experimental hypothesis is accepted and the null hypothesis is rejected.
Relationship of results to hypothesis
The results that were gained from this experiment strongly support the experimental hypothesis stated in the introduction. In the hypothesis it was predicted that ‘There will be a far greater recall of words from the related words list than the unrelated words list.’ The results of the experiment support the experimental hypothesis because there was a significant difference in recall between the unrelated word condition and the related word condition. This can be seen in the mean, mode and median of both sets of results. On every occasion the figure for the related list is greater then that of the unrelated list. The graph shows overall higher memory recall from the related list. The Mann-whitney U test suggests that the results are due to the IV and not to chance. This means that the experimental hypothesis is accepted and the null hypothesis is rejected. Therefore this experiment suggests that chunking leads to better recall.
Discussion
Validity and suggestions for improved validity
The results that were gained from this experiment strongly support the experimental hypothesis stated in the introduction. The results of the experiment support the experimental hypothesis because there is a significant difference in recall between the unrelated word condition and the related word condition.
However the validity of these results can be questioned. The external validity of the experiment is the most obvious downfall. It is not representative of the population as almost all of the participants were 16 or 17 and from the sixth form. This means participants were not really representative of the population. This could have been improved by using systemic sampling. The experiment also lacks ecological validity, as the experiment is not something people do every day.
Over time the experiment is valid, it could easily be repeated in a few years time on a different group. The amount of time each card was shown to the participant could have varied due to the fact that timing was done in manually, instead of electronically. This could have affected the results because if one card was held up longer than the others, then, it could be argued that that particular word would have been remembered more frequently than others. A way to resolve this problem would be to use a stopwatch when timing. Another problem was that all of the participants were wearing clothes during the experiment; this could of cued recall as the related list was based on clothing. This is a major downfall of the experiment and could easily be countered by using a different category for the related words list. Internal validity could also be questioned. Order effects could not have affected the results, as independent measures were used. Experimenter bias could have affected results. When designing this experiment, cue-dependent remembering was not taken into account. It is possible that cue-dependent remembering is responsible for the difference in the number of words recalled from each list. This is possible because in the related list one word could cue another and then even another, this is not possible in the unrelated list.
In some ways the experiment was valid. There was a standard procedure in place, however it is questionable how strictly each interviewer followed these procedures.
To improve validity many things could have been done. Firstly matched pairs could have been used. Also each test could have been done in the same place under the exact same conditions. This would prevent cue-dependent remembering as a result of objects in the room. Also the words lists could be redesigned to ensure that there was not even any kind of distant link between the words on the unrelated words list. The related words list could also be on a different category and revised to make sure that it did not contain words, which would cue other words. Both words lists could also have been redesigned to make sure that neither contained words of comic value or particular significance.
Reliability and suggestions for improving reliability
Having carried out the Mann-Whitney U-test it is fair to say that the experiment was reliable as there was only a 0.5% chance that the significant difference in results was due to chance. But there were also a few problems, which resulted in the actual method of the experiment to be inconsistent. One of the main problems with the experiment was that there were four different experimenters carrying out the study. Despite there being a set of standardized instructions there could have been some fairly large differences in the way each experimenter carried out his experiment and also experimenter bias. There are subtle things like tone of voice or emphasis on certain words, which could affect the recall. Each experimenter was wearing different clothes, and as the related words list was based on clothes this could have greatly affected recall. More importantly the amount of time each word was shown to each person by the experimenter was probably also different for each time. The apparatus was barely adequate for the experiment. Human judgment was too heavily relied on, this could bring in the possibility for human error. For all of these reasons it would be extremely difficult to replicate this experiment. To improve reliability the method would have to be changed and a far more detailed set if standardized instructions would be needed. Firstly the related words list should have been based on a different category for example: Animals. This would be useful as the experiment could have been done in a room where there were no animals or reference to animals, this was of course difficult when using clothing as the topic. Also there should be no chance of one word cueing another. This was often the case in the related condition for example participants would remember boxers, then straight after that remember briefs. This could be seen as chunking, but is more likely to be cue-dependent remembering, as the fraise ‘boxers or briefs?’ is known by almost everyone. The same person should have carried out all the tests in the same room, preferably. Stopwatches could have been used to time the how long each word was shown. The room should be clear of any object, which could cue a word. PowerPoint could even have been used to show the words in sequence at exact time intervals. This would eliminate the possibility for human error.
Implications of study
The results completely support all background research that carried out. The results indicate that chunking leads to better recall as suggested by Miller. The results also support the primacy recency affect and serial position curve theorized by Murdock. In the results late items were recalled best, early items were recalled next best and middle items were recalled least well. The results also support Atkinson and Shiffrin’s multi-store model.
Generalisation of study
The experiment did not really have ecological validity as the experiment was not a real life situation. The experiment has very little population validity as most of the participants were aged 16 to 18. As well as this there were more males than females. The experiment suggests that in general chunking leads to a greatly increased ability to recall. However, this can only really be generalised to 16 to 18 year olds.
Applications of study to real life
In conclusion, it can be said that there are many applications of this experiment to real life. Memory is used everyday from the moment you wake up. Chunking is useful when trying to remember telephone numbers, addresses or codes. Memory is needed for school when revision has to be done for exams, this is when chunking information can be particularly useful, as it is possible to group relative information and remember it as one chunk of information. Without chunking recall would be less useful as far less information could be recalled.
Appendix
Wordlists
Related Condition
- Fleece
- Gloves
- Cardigan
- Thong
- Bra
- Trainers
- Knickers
- Trousers
- Jacket
- Hat
- Shoes
- Jeans
- Scarf
- Jumper
- Briefs
- Boxers
- Coat
- Shirt
- Socks
- Stockings
Unrelated Condition
- Carrot
- Locker
- Microphone
- Airplane
- Bush
- Steel
- Hair
- Elbow
- Poster
- Siren
- Dog
- Cooker
- Curtain
- Rain
- Light
- Bag
- Bike
- Bat
- Chimney
- Telephone
Example Answer Sheet
Unrelated Condition Related condition
Microphone Fleece
Airplane Hat
Bush Gloves
Steel Socks
Poster Jumper
Siren Shoes
Dog Stockings
Cooker Boxers
Briefs
Thong
Knickers
Standardised Instructions to Participants
Participants were invited into an empty quiet room, and sat facing the experimenter. The participant was then briefed that they were going to be shown a series of twenty cards with words written on them. (At this point the true purpose of the experiment was not disclosed). The cards were then held up, one by one in order of the word lists, shown for three seconds at a time, after which the next card was immediately produced. When all the cards were produced, the participant is then asked to free recall as many of the words they had just seen as possible. The participant was then de-briefed as to the true purpose of the experiment.
Raw Data
The Mann – Whitney U test
U=NA NB + NA(NA + 1) - RA
2
U= 20*20 + 20(20+1)/2=610
610-255.5= 354.5
U=354.5
U- = NA NB – U
20*20-354.5= 45.5
U-=45.5
Na= 20 Nb= 20
U=354.5 U-= 45.5
Calculating the mean
Related
=257
257/12= 12.85
Unrelated
=166
160/20= 8
Calculating the variation ration
Related: 100/20*8=40%
Unrelated:100/20*12=60%
References
Atkinson, R.C. and Shiffron, R.M. (1968) Human memory: A proposed system and its control processes. In Spence, K.W. and Spence, J.T. (eds) The psychology of learning and motivation, vol. 2. London: academic press.
Craik, F.I.M. And Lockhart,R.S. (1972) Levels of processing: A framework for memory research. Journal of verbal learning and Verbal behaviour.
Craik, F.I.M. and Tulving, E. (1975) Depth of processing and retention of words in episodic memory. Journal of experimental psychology.
Miller, G.A. 91956) The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological Review, 63, 81-97
Tulving,E. and Pearlstone, Z. (1966) Availability versus accessibility of information in memory for words. Journal of verbal learning and Verbal behaviour, 5, 381-91
References and Appendices