Hypothesis
The presence or absence of delay in recall will affect which words (position) are recalled from a list.
Method
Method and Design
The method used for this experiment was a laboratory experiment. It is the appropriate choice as this investigation looks at the manipulation of the independent variable (IV), time delay in recall and sees how it affects the dependent variable (DV), position of the list for the word(s) recalled. This demonstrates the ‘cause and effect’ relationship between the two variables, thus making it experimental.
The design used was independent measures. Using different subjects for each condition ensured there were no practice effects because if the same subjects were used in both conditions, their results would inevitably get better through practice.
Variables
Independent variable: the delay in recall (seconds)
Dependent variable: the position of the list for the word(s) recalled. (It can be measured by converting results into a ‘percentage recall’)
Participants
The target population was year 10 and 11 students from KGV, an international school in Hong Kong, ranging from 13-16 years old. A total of 38 males and 38 females were chosen through random sampling and those who were able to attend were used in the experiment. Random selection of sampling ensured no bias from experimenters, but also meant ethnicity of sample is random. However, being an international school represented by over 40 nations, ethnicity should be widely varied.
Apparatus
-
- Sheets of A4-sized paper- for participants to write their words recalled
-
- Pens- for writing on the paper
-
- 2 sets of 20 word list (See appendix 3), created by one of the researchers where words are all of one-syllable and similar difficulty. A list of 20 words was believed be appropriate because in order the show the separate existence of STM and LTM it would require more than 5-9 words. Therefore 20 words suited well as it met the criteria but would not be as time-consuming as say, a 30-word list.
Procedure
Available participants (experiments for different conditions were performed at different times) were sat down in a classroom (lecture theatre of school) and were tested as a whole group. There were 2 main steps involved in each condition. (Refer to appendix 2)
Subjects are briefed- includes notifying participants that they have the right to withdraw from the experiment at any time.
- A 20-word list was read out one-by-one with a second interval between each word by one of the researchers. Subjects were asked to keep silent and remember as many words as possible.
-
This next step was carried out according to its particular condition.
Condition 1: Subjects were given nothing to do (for 0 seconds)
Condition 2: Subjects were left to do nothing in silence for 30 seconds.
- Subjects were then given 2 minutes to remember as many words as they possibly could and write them down on the paper provided.
Subjects were debriefed and assured of confidentiality of statistics obtained.
Controls
Measures were taken when samples were chosen to control extraneous variables. The same number of males and females were chosen through random selection for each condition.
Independent measures did eliminate some extraneous variables, although not all. Order effects can also be minimised as participants only took part in one condition, so factors such as fatigue and boredom was not a problem. In addition, because subjects were not exposed to both conditions and hence reduced the chances of them behaving differently. Therefore, chances of demand characteristics were also reduced to an extent. Confusion was avoided through clear, concise, standardized instructions that guaranteed everyone had equal information about the experiment. Lastly, there were no distractions with the procedure.
Ethical issues were resolved as consent forms were given out for parents of participants to sign as they have yet to reach the age of 18 (see appendix 1). It gave parents a brief outline of the procedures in the experiment. In addition, participants were informed they had the liberty to withdraw from the experiment at any given time. Results were also kept confidential.
Results
Summary table to show the percentage recall of words (position) in a list.
Summary table commentary
Statistics such as the median, mode and range are unnecessary for this investigation because each value in the table has no relationship to any other value so therefore a value that distinguishes all values in the table is meaningless. The trend in condition 1 begins high (16) but then drops towards the middle of the list (3) before increasing again at the end (14). In condition 2, it showed a similar trend in the beginning with (15) and again descending towards the middle of the list (2). However, there was a less significant increase towards in the end in condition 2 (6).
Graph to show the percentage recall of words (position) in a list.
Commentary on graphical display of results
As shown in the graph above, condition 1 first shows a decreasing trend (from 89% to 11%) before it increases to 78% at the end. Condition 2 also shows a decreasing trend to begin with (from 83% to 11%). However, from that point it does not show an apparent increase and concludes at 33%.
Relationship to the hypothesis
The hypothesis states, “The presence or absence of delay in recall in a memory experiment will affect which words (position) are recalled from a list. People have better recall for words at the beginning and end of a list due to the LTM and STM respectively. Lastly, if recall is delayed by 30 seconds the recency effect will be destroyed.”
The results support the hypothesis as can be seen in the graph above, the absence of delay (condition 1) gives different results than when delay of recall is present (condition 2) as condition 2 demonstrates a reduced, almost destroyed (33%) recency effect whereas the recency effect is very much present (78%) in condition 1. Results from condition 1 (control) also illustrate the serial position curve, which highlights much better recall at the beginning and ends of a list.
However, a slight difference between the hypothesis and results is a higher recall in the recency effect than predicted (33%), which if followed the hypothesis would’ve been close to its mid-list value (11%). An explanation for this is that subjects is condition 2 had experimenter demands, being aware of the experiment being a memory test and secretly rehearsing the words rather than sitting in silence doing nothing like they were told to in order to achieve a higher score. Hence, there is a slight sign of the recency effect.
A relationship shows an increase in IV (more delay) would decrease the recall towards the end of the list, which is part of the DV.
Discussion
Validity
Validity refers the ‘trueness’ of an experiment and how well it can be generalised. The variables were operationalised by giving them a standard in which it all can be measured against.
For the IV, time for delay was measured in seconds. This is the most appropriate measure as it is the most common measure for time. The DV was measured by a percentage. The total number for each word in the list was divided by the total participants and multiplied by 100 to give a percentage of recall. This is again an appropriate method for measure as it gives a standard to which all values are measured against (i.e. 100%).
Ecological validity is how true and similar a study is to real life. This study does not have very good ecological validity, as it is very uncommon for people to have to remember a set of single-syllable words.
Construct validity is the comparison of test scores with the expected results. As seen from the graph, this demonstrates a similar trend to that of the expected results. However, the recency effect of the experimental condition did not exactly follow its expected trend. This may be due to experimenter demands.
Suggestions for improving validity
An improvement for the ecological validity of the study is to recall a sentence or paragraph of words rather than a list of words. It is common for people to remember a sentence of words possibly for an exam or a presentation speech.
Another improvement for validity could be to improve construct validity. To do so, experimenter demands must be reduced. Giving subjects an article to read during the 30-second delay may do this. However, it can be argued that this may link in with interference rather than just delay
Reliability
Reliability is a measure of how dependable or consistent a study is. It does not define its accuracy, as an inaccurate study can still be reliable. A reliable study would achieve very similar results if repeated.
This study may not achieve the same results because many factors can change the process of the study, thus probably changing the results. Using independent measures would prevent practice effects every time. However this means the two conditions can have very different subjects hence results can vary greatly each time. Moreover, if the same wordlist were used again then this would increase practice effects even more. On the contrary, if a different wordlist was used, then this can produce completely different results, as some words may be easier for some subjects to remember than others possibly due to the familiarity of them. There is no way of identifying how easy/hard a word is to remember due to its subjectivity.
Sampling can also change a study’s reliability as all subjects have the liberty to choose whether they attend the study or not hence the number of participants can vary greatly between studies.
Controls and apparatus can be kept fixed with relative ease and standardised procedures and instructions can easily be reused to ensure maximum reliability.
Improving reliability
An alternative to independent measures is matched pairs design. This method has subjects of similar qualities taking part in each of the two conditions thus eliminating practice effects. This would bring much similarity between the beginning and middle of the SPC and have a significant change on the recency effect, as participants’ properties are alike. It would be important to match age, gender, ethnicity and intelligence. Pairing people of similar intelligence would give an indication of their memory capabilities. However, this method is very time-consuming and not always effective.
Another improvement that can be made is to have a larger sample of participants in each condition. It will increase reliability significantly and give more generalisability to the target population. Then again, the number of subjects participating is and extraneous variable as it is difficult to determine how many will show voluntarily. Having a larger sample would inevitably lead to more participants, which would help improve accuracy.
Implications of the study
As seen from the graph, results of this study are very similar to Glanzer and Cunitz (1966). They found that a 30 second delay in recall destroyed the recency effect whereas this study showed a significant decrease in the recency effect but not completely destroyed. An explanation is that subjects rehearsed words out of boredom or experimenter demands during the 30-second delay.
The results also agree well with Miller (1956) and Atkinson and Shiffrin (1968, 1971). The SPC suggests that STM and LTM do exist separately and that STM has 5-9 chunks of storage. More evidence is shown if both ideas are put into perspective because the delay in recall destroyed the last 5-9 words of the list held in STM, which only has a storage time of approximately 30 seconds. Therefore, it explains the more evident recency effect in condition 1(no delay) compared to condition 2 (with delay) where the recency effect was almost completely destroyed.
To sum up, the recency effect decreases as the time in delay before recall increases. In order to determine its time for complete termination further research would be needed.
Generalisation of findings
It is possible to generalise for the target population. The target population are year 10 and 11 students from an international secondary school in Hong Kong (KGV).
There were 36 participants from a total of approximately 550 students, which means 7% of the target population was tested. However, ethnicities of samples are completely random hence its range can be vast or very limited. The participants, depending on who turns up the ethnic range can be widely varied and include every ethnicity form our school, or be entirely Chinese. It is completely random.
It is probably possible to generalise beyond the target population to an extent. It is probably possible to generalise the ethnicities of the sample to the whole of Hong Kong as there are over 40 nationalities represented in KGV. Moreover, there was a 1:1 ratio of boys to girls so it can be generalised for both sexes. However, it is not possible to generalise in terms of age as participants were between 13-16 years old from an international school in Hong Kong.
Therefore, it can be concluded that it can be generalised beyond the target population to all 13-16 year olds in Hong Kong.
Application to everyday life
The results show that the findings agree with the majority of Glazer and Cunitz’s work. Applying this to everyday life would indicate learning great amounts of information, for example, the night or morning before an exam is unadvised as much of this would be lost and not properly rehearsed into LTM. Moreover, the loss of the recency effect due to delay encourages people not to revise for the exam as much of the words they place in their STM would be lost by the time they get into the exam hall due to delay in recall.
This encourages students to revise thoroughly throughout the year and be well prepared rather than rushing revision the night before.
References
-
- Atkinson, R.C. and Shiffrin, R.M. (1968, 1971) 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
-
- Glanzer, M. and Cunitz, A.R. (1966) Two storage mechanisms in free recall. Journal of Verbal Learning and Verbal Behaviour, 5, 351-60
-
- Miller, G.A. (1956) The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological Review, 63, 81-97.
-
- Sperling, G. (1960) The information available in brief visual presentations. Psychological Monographs: General and Applied, 74, 1-29.
Bibliography
-
- Jarvis, M., Russell, J., Flanagan, C., Dolan, L., Angles on Psychology, 2000, Stanley Thorns Ltd.
-
- Cognitive Psychology Lecture #6,
-
- Exam 2 -- Sample Answers to Short Answer Questions,
-
- Graph of the serial position curve,
-
- Banyard, P. and Graysen, A., Introducing Psychological Research, 1996, Macmillan Press LTD.