In the classic paper “The Magic number seven, plus or minus two” Miller established that more items were remembered (immediate memory span increased) when information was ‘chunked’ rather than the number of items, when the amount of information averages seven chunks. This is why a familiar date such as 1066 is likely to be a chunk rather than a less meaningful date as 1325. The memory span for letters is about six when in random order and nine when a constant-vowel-constant syllable, increasing the span to approximately 50 when the words made a meaningful sentence, however, the number of chunks remains at about six.
Aim
Focusing on the above research into chunking of digits specifically, the aim of this investigation is to establish the effect of chunking in short-term memory. The above research was conducted over 30 year ago, doing this study will establish if the magic number seven plus or minus two applies to 16-17 psychology students; or maybe with technology we rely more on gadgets to store information for us and the active capacity of information might be decreasing as we could be sub-consciously chunking more nowadays with high usage of telephones and mobile phones. Looking at research mainly by Jacobs (1887) and Miller (1956) the following hypotheses have been drawn up.
Alternate Hypothesis
Significantly more digits will be recalled accurately in sequence in the chunked condition than in the un-chunked condition.
Null Hypothesis
There is no difference in the number of digits recalled in the chunked or un-chunked condition.
Method
Design
The method used in this investigation was a laboratory method with repeated measures design. The independent variable was whether the digit sequence was continuous or chunked and the dependent variable was the number of correctly recalled digits in succession.
Many variables were controlled in this experiment, such as the use of standardised instructions (see appendix). The experiment took place in the same psychology room with all windows shut to minimise the effect of external variables such as noise from traffic and school bell ringing. The study also took place at the same time on consecutive days, to ensure that concentration levels are the same, reducing the variable of tiredness and hunger. As a control from distraction a notice on the door was put up saying ‘Experiment in process, do not enter’. All mobile phones were asked to be switched off at out of hands reach to minimise distraction. Communication between participants was prevented by explaining the importance of independent results at the beginning of the experiment.
To avoid any ethical issues all participants were debriefed at the end of the experiment and given the right to withdraw their results at any moment. All results were anonymous for confidentiality and participants gave consent to participate in the experiment.
Participants
25 participants were used for this investigation, both male and female. The target population were 16-17year old students from Cherwell School. An opportunity sample, two classes of en-rolling psychology students into sixth-form from the target population was used. Using repeated measured design means that participants do both condition. However, this creates a problem with order effect and boredom. To overcome this effect counterbalancing is done. Counterbalancing uses the ABBA procedure; the participants are split into two groups by placing all names into a hat and randomly pulling out the first 12 names to form group one, this reduces the experiment effect. Group 1 do condition A followed by condition B; whilst group 2 do condition B followed by condition A. The researcher was an 18 year old sixth former student at Cherwell School.
Materials Used
(Refer to Appendix)
- Consent Form
- Debriefing Script
- Instructions
- Pen and Paper for every participant
- Silence Sign
- “Experiment in Progress, Do Not Enter” sign
- Stimulus material – PowerPoint presentation
- Computer and Projector
Procedure
- A psychology classroom was booked on two consecutive days for the same time
- All equipment was set prior to the study.
- Computer and Projector
- Pen and Paper on every desk for the participants
- Sign on Door “Experiment in Progress, Do Not Entre”
- Sign “Silence” visible from every participants view
- Got the first class of en-rolling psychology students to attend an extra psychology activity.
- As students entered the room participants were asked to switch off their phones.
- Once participants were seated consent forms should be handed out and collected with a signed signature. (see appendix for consent form)
- The researcher began the presentation reading out loud all instructions and pausing for questions before beginning the study. (see appendix for standardised instructions)
- Group One began with the un-chunked condition followed by three trivia questions to clear their minds. After the trivia questions the instructions were read out again and participants did the second condition, chunked sequence.
- Stopped presentation on slide “Debrief”
- As soon as the time was over results were collected in to minimise the chance of comparison of results and cheating.
- Participants were debriefed and were given the right to withdraw their results.
- The study was repeated on the following day with the second class of en-rolling psychology students. However, the presentation began on the second half, slide “Group 2”. Participants began with the chunked condition, followed by three trivia questions and finished with the un-chunked condition.
- Both groups were treated the same to decrease the effect of extraneous variables and to keep the experiment fair.
Results
Summary Bar Graph:
Summary Tables of the data for Un-chunked and Chunked conditions are show above. The score was the number of consecutively correct digits recalled. If the participant didn’t get the first digit in the sequence correct the participant scored Zero.
The mean was calculated by adding all the scores in a condition e.g. Un-chunked and dividing the sum by the number of pieces of data, 25 pieces of data. Standard Deviation measures the spread of data about the mean, the bigger the value the more spread out the data.
A table to show the mean and the standard deviation values:
I used a Wilcoxon’s Matched Pairs. I chose this statistical test because the two sets of data are related, repeated measures design. This test is also used when they hypothesis predicts a difference between two sets of data. The data collected is ordinal and thus could be ranked.
The critical value of T at p=0.05 (one tailed) for N=21 is 68 (95% due to IV)
The critical value of T at p=0.01 (one tailed) for N=21 is 49 (95% due to IV)
The calculated value from my results was T=26
The calculated value of T must be equal or less than the critical values to reach that level of significance. As 26 was less than both 68 and 49 this shows that my results were significant to a 99% level. I accepted the alternate hypothesis that significantly more digits would be recalled accurately in sequence in the chunked condition than in the un-chunked condition. (For raw data see appendix)
Discussion
Explanation of Findings
In this study, the results support the alternate hypothesis; ‘significantly more digits will be recalled in sequence in the chunked condition than in the un-chunked condition’. Participants performed better and got a higher score in the chunked condition of an average of 6.56 correct digits consecutively. However, looking at the raw data, the difference between the chunked and the un-chunked condition is very small. The average for the un-chunked condition is 4.92 (≈5). Calculating the standard deviation allows one to see the spread of data about the mean. The un-chunked condition was calculated to have a slightly higher spread of data meaning that there was a bigger range of values.
Looking at the bar graph, it is clearly visible that chunking recall is better than un-chunked recall. Nevertheless, all results deviate around the 6 recalled digit mark. There is one anomaly, participant 13, scored highly in the un-chunked condition, scoring 9, whilst only scoring 6 in the chunked condition. This may have been for a number of reasons. If the participant was bored, the concentration levels may have dropped and no effort was put in; or, the participant found the task hard and needed a practise round, scoring highly the second time but with the different condition. As the results were anonymous it is impossible to identify the reason.
Relationship to background research
These findings relate to previous studies. Jacobs (1887) did a study with a serial digit span and he found that on average the digit span was 9.3 digits. Jacobs conducted his study on university students aged 20+; this supports Jacob’s theory that age plays a big role in recall. At the age of 16 people are still increasing their memory capacity and the means of organising information, chunking.
These results do not fit Miller’s 7±2 theory. The 15 digit serial digit span was split into five chunks and had no semantic meaning. Participant on average recalled two chunks. Could it be that the time given for learning the sequence was too short, or having mobile phones to hold 11 digit numbers mean that we do not need to develop the skill of recalling a sequence of digits?
Wickelgren’s observation is valid and seen in this study. Most participants in the chunked condition recalled six digits, which is equivalent to two chunks; the second highest score was nine digits, three chunks, followed by five and seven digits. When recalling mobile phone numbers the 11 digits are split into two parts, the code and the number. The code always begins with 07 and is four digits long; the number is six digits long and usually split into two chunks. If the 15 digit sequence was split into two sequences of 6 digits, could participates recall more digits because they would see it as two separate numbers?
This shows that chunking does improve recall slightly but it is difficult to determine the true capacity of short-term memory.
Limitations and modifications
One limitation of the study was the participant sample used. Opportunity sampling from a sixth form target population was used and so the sample contained many biases like age and level of intelligence. Using participants from other schools would make the results more reliable as there would be a greater variety of intelligence and the population sample would be closer to the target population.
Another flaw is the participants might not have taken the experiment seriously since they were tested in a relaxed setting. Discussion was hard to prevent between the slides shown, meaning concentration levels were low. A relaxed setting might show the true capacity of the active memory rather than forced when in exam conditions for instance. To prevent discussion, participants should sit facing one direction and as spread out from each other as possible.
An additional drawback was the attitude towards the researchers, some of the participants knew the researchers and therefore answers would not be true, but skewed to please the researcher. This can be avoided by asking a person of authority, such as a teacher, reading the instructions and explaining the importance of concentration and silence. This would also avoid the experimenter effect.
The time given for participants to learn the digit sequence was 15 seconds which was not long enough because the results between the two conditions are too similar. A distraction task was put in between conditions to clear the memory. Task 1 may have interfered with Task 2. To truly see if chunking increases the active memory capacity the time to learn the sequence should be increased to at least 30 seconds and the time to clear the memory with trivia questions should also be increased.
Further Studies
There are many follow up studies that could be conducted. By taking results from different age groups, 6-8, 16-18 and 26-28 year olds, would expand on Jacob’s theory that the chunking capacity of short-term memory increases with age due to an increase in learnt material and developed organisation skills. Participants aged 6-8 would require parental consent and it is predicted that the recall score would be lower than six digits in consecutive order. Participants aged 26-28 would be predicted to recall the whole digit sequence, up to 15 digits correctly consecutively to support Jacobs theory that active capacity increases with age due to developing means of chunking information.
Another study that could be done is a follow up study on Wickelgren; by having a sequence of digits in different size chunks of two digits, three digits or four digits per chunks. Results should show that having three digits per chunk produce the best recall.
A different study that would be interesting to follow up would be a study with mnemonics. Slak’s study showed that by having a mnemonic which could be chunked, improved the digit recall. The study would be to see if digit recall would be as good when the number of digits in the sequence increased, consequently the number of chunks in the mnemonic. The participants IQ might have an impact on sequence recall.
In conclusion, this investigation has found that 16-18 year olds on average recall six digits in the chunked condition, equivalent of two chunks. This research has proven that chunking does increase the capacity of short-term memory, even if it is only by a few more digits. In the real world this information could imply that 16-18 year olds are not very good at holding a vast amount of information in short-term memory, which could influence their learning skills if new material is not rehearsed. Another place where digit chunks are used is in restaurants – taking this investigations results suggests that a 16-18 year old would only be able to deal with two tables (two chunks) at a time (every table has a number).