Chapter 4 Experiment 2
Manipulating the Content of Preceding Associates Tasks and the Hint Salience The first goal of Experiment 2 was to replicate the findings of Experiment 1. The second goal was to determine more precisely the interaction between the construction of representations and the operation of hints in verbal insight problem solving. The results of Experiment 1 suggested that HT problems designed to activate the construction of inappropriate representations are more likely to benefit significantly from the presence of hints while LT problems designed to activate the construction of appropriate representations are less likely to benefit significantly from the presence of hints. Nevertheless, the manipulation of the typicality of critical components in Experiment 1 inevitably altered the problems. It raised the question of whether the interaction between the construction of representations and the operation of hints remains the same when the manipulation is conducted without altering the problems.
Hence, we manipulated simultaneously two variables in Experiment 2: The content of preceding associates tasks and the hint salience.
The manipulation of the content of preceding associates tasks aimed at testing the role the construction of representations plays in verbal insight problem solving. The associates task was constructed with two variants, including the experimental version consisting of items identical to the critical components in the twelve insight problems (II) and the control version consisting of items non-identical to the critical components in the twelve insight problems (NI). II task consisting of items identical to the critical components which facilitate the occurrence of transfer was designed to
of items non-identical to the critical components, on the other hand, was designed to serve as a reference state. One of the tasks was administered before participants began to attempt the twelve insight problems, requiring them to generate as many examples as possible for each item in the task. The associates task also required participants to apply this strategy of generating examples in the associates task to the following insight problem solving activity (Table 4-1).
Table 4-1 Manipulation of the Content of Preceding Associates Tasks
! " # $ % &' ( &) * + ' , - &. / 0 1 2 3 4 5 6
7 1 8 9 : ; < = &> / ? @ A B C D E F < G / H I J 1 8 K L
M N O P Q
Item in II task Item in NI task
R
The manipulation of the hint salience aimed at testing the role the operation of hints plays in insight problem solving. The manipulation, which was identical to that in Experiment 1, was conducted by constructing each insight problem with two variants, including the one with high hint salience (HHS) and the other with low hint salience (LHS). HHS problems with the critical components being highlighted instructed participants that the key to solve these problems lies in the highlighted parts. LHS problems, on the other hand, merely instructed participants to solve these problems without highlighting any components.
Since the manipulation of the content of preceding associates tasks aiming at testing the role the construction of representations plays was conducted without altering the problems, whether the construction of representations is influenced in the former
stage or the later stage of verbal insight problem solving remained an open question.
On the basis of the findings in Experiment 1, we generated the following predictions.
1. If the manipulation of the content of preceding associates tasks influences the construction of representations in the former stage of verbal insight problem solving, an interaction between the content of preceding associates tasks and the hint salience should be observed. In NI condition, there should be a significant difference between participants’ performance on HHS problems and LHS problems. Moreover, their performance on HHS problems should be significantly higher than that on LHS problems. In II condition, however, there should be no difference between participants’ performance on HHS problems and LHS problems. The difference between participants’ performance on HHS problems and LHS problems is expected to be restricted to NI condition because participants in NI condition are more likely to benefit significantly from the presence of hints while those in II condition are less likely to benefit significantly from the presence of hints.
2. If the manipulation of the content of preceding associates tasks influences the construction of representations in the later stage of verbal insight problem solving, an interaction between the content of preceding associates tasks and the hint salience should not be observed. The manipulation of the hint salience is expected to significantly influence participants’ performance no matter whether they are in II condition or NI condition while the manipulation of the content of preceding associates tasks is expected to significantly influence participants’ performance no matter whether they attempt HHS problems or LHS problems. A lack of interaction is expected because the manipulation of the content of preceding associates tasks influences the construction of representations in the later stage
hint salience influences verbal insight problem solving when the construction of representations is absent.
4.1 Method
4.1.1 Participants
180 undergraduate and graduate students at National Taiwan Normal University participated in Experiment 2 as partial fulfillment of a requirement for Educational Psychology, Sociology of Education, or Instructional Medium. Participants were quasi-randomly assigned to one of the conditions, with 45 participants in each condition (Table 4-2).
Table 4-2 Participants in Experiment 2
Low hint salience, LHS High hint salience, HHS
Identical items, II 45 45
Non-identical items, NI 45 45
4.1.2 Materials
4.1.2.1 The Verbal Insight Problems
Twelve verbal insight problems, each of which was constructed with two variants based on the manipulation of the hint salience (Appendix C), were presented in packets (Appendix D).
The order of the problems was counterbalanced, resulting in four problem versions.
Each packet began with a direction page which explained the procedure. Following the direction page, each of the twelve insight problems was presented with two
familiarity questions to check whether participants had ever seen the problem or known the solution to the problem.
4.1.2.2 The Preceding Associates Tasks
The preceding associates tasks, including the II associates task (Appendix E) and the NI associates task (Appendix F), were presented in packets. II associates task consists of twelve items identical to the critical components in the twelve insight problems while NI associates task consists of twelve items non-identical to the critical components in the twelve insight problems. The order of the items was counterbalanced, resulting in four task versions. Each packet began with a direction page which explained the procedure requiring participants to generate as many examples as possible for each item in the task.
4.1.3 Design
The design of Experiment 2 was a 2*2 between-subject design. The two independent variables were the content of preceding associates tasks (identical items, II/
non-identical items, NI; between-subject variable) and the hint salience (high hint salience, HHS/ low hint salience, LHS; between-subjects variable) of the twelve verbal insight problems. The dependent measure was the solution rate.
4.1.4 Procedure
The participants were quasi-randomly assigned to a one of the conditions with all aspects of the procedure being held constant. They were informed that they would be performing (1) an association task in 12 minutes and (2) twelve insight problem
Solutions to the problems were coded as correct. Any other solutions or failures to answer were coded as incorrect. Solutions to the problems that were indicated as having been known before were coded as missing.
4.2 Results
4.2.1 The Preceding Associates Tasks
Table 4-3 presents the mean associates number and standard deviation on the preceding associates tasks for the four conditions. Mean associates number for the four conditions was 5.557 (II and LHS), 5.393 (II and HHS), 5.345 (NI and LHS), and 5.452 (NI and HHS), respectively.
Table 4-3 Mean Associates Number (Standard Deviation) in Experiment 2
Low hint salience, LHS High hint salience, HHS Identical items, II 5.557 (1.260) 5.393 (1.043)
Non-identical items, NI 5.345 (1.333) 5.452 (1.431)
An analysis of variance (ANOVA) shown in Table 4-4 demonstrated that the interaction between the manipulations of the content of preceding associates tasks and the hint salience was not observed (F(1, 176)=.510, MSE=1.625, p=.476). Neither the main effect for the manipulation of the content of preceding associates tasks (F(1,176)=.164, MSE=1.625, p=.686) nor the main effect for the manipulation of the hint salience (F(1,176)=.023, MSE=1.625, p=.880) was observed.
Table 4-4 Two-way ANOVA for Mean Associates Number in Experiment 2
SV SS df MS F p Partial
Eta Squared Between subjects
Associates .267 1 .267 .164 .686 .001
Hint salience .037 1 .037 .023 .880 .000
A*H .828 1 .828 .510 .476 .003
Error 286.062 176 1.625
Total 287.194 179
*p<.05
4.2.2 The Verbal Insight Problems
Table 4-5 presents the frequency and percent of solutions which were indicated as having been known before. Missing rate for each insight problem ranges from 0.00 (item 12) to 12.80 (item 09).
Table 4-5 Missing Rate in Experiment 2
item 01 item 02 item 03 item 04 item 05 item 06
Frequency 1 1 5 14 4 9
Percent 0.60 0.60 2.80 7.80 2.20 5.00
item 07 item 08 item 09 item 10 item 11 item 12
Frequency 8 6 23 14 1 0
Percent 4.40 3.30 12.80 7.80 0.60 0.00
Table 4-6 presents the mean solution rate and standard deviation for the four conditions. Mean solution rate for the four conditions was .540 (II and LHS), .579 (II and HHS), .483 (NI and LHS), and .537 (NI and HHS), respectively.
Table 4-6 Mean Solution Rate (Standard Deviation) in Experiment 2
Low hint salience, LHS High hint salience, HHS Identical items, II .540 (.138) .579 (.149)
Non-identical items, NI .483 (.146) .537 (.164)
An analysis of variance (ANOVA) shown in Table 4-7 revealed a main effect for the manipulation of the content of preceding associates tasks (F(1,176)=4.849, MSE=.022, p=.029), with participants in II condition outperforming those in NI condition. It also
revealed a main effect for the manipulation of the hint salience (F(1,176)=4.379, MSE=.022, p=.038), with participants in HHS condition performing significantly
better than those in LHS condition. However, the interaction between the manipulations of the content of preceding associates tasks and the hint salience was not observed (F(1, 176)=.108, MSE=.022, p=.743).
Table 4-7 Two-way ANOVA in Experiment 2
SV SS df MS F p Partial
Eta Squared Between subjects
Associates .109 1 .109 4.849* .029 .027
Hint salience .098 1 .098 4.379* .038 .024
A*H .002 1 .002 .108 .743 .001
Error 3.939 176 .022
Total 4.148 179
*p<.05
To summarize, the results revealed that participants in II condition significantly outperformed those in NI condition. It is also demonstrated that participants in HHS condition performed significantly better than those in LHS condition (Figure 4-1).
Figure 4-1 Mean Solution Rate in Experiment 2
4.3 Discussion
It is observed that participants’ mean solution rate in II condition is significantly higher than that in NI condition either when they attempted LHS problems or HHS problems. According to our manipulation, II task was designed to facilitate participants’ construction of appropriate representations while NI task was designed to serve as a reference state. The result that participants performed better in II condition than in NI condition suggested that participants benefited significantly from the design of II task.
It is also observed that there was a significant difference between participants’
performance on HHS problems and LHS problems either when they were in II condition or NI condition. According to our manipulation, HHS problems were designed to facilitate participants’ identification of critical components while LHS
performed better on HHS problems than on LHS problems suggested that participants benefited significantly from the presence of hints.
In sum, the result that there was a lack of interaction between the manipulation of the content of preceding associates tasks and the hint salience supported our prediction that the manipulation of the content of preceding associates tasks influences the construction of representations in the later stage of verbal insight problem solving. On one hand, since the operation of hints influence verbal insight problem solving when the construction of appropriate representations is absent, their different performance on HHS problems and LHS problems should not be influenced by the manipulation of the content of preceding associates tasks. On the other hand, since the construction of representations is conducted when the operation of hints is accomplished, their different performance in II condition and NI condition should not be influenced by the manipulation of the hint salience.