This study examined the effect of time pressure on solving optimal stopping problems. 71 participants completed an optimal stopping task with numbers under short (3 numbers) or long (7 numbers) conditions. Performance was analyzed across blocks of trials. Results showed that the short condition led to significantly better performance than the long condition. While no significant learning effects were found overall, subjects in the long condition showed a slight improvement in the final block, indicating possible learning with more alternatives. Thus, increasing the number of choices can hinder optimal decision making due to added complexity and pressure.
Invited talk at the Focus Fortnight 8: ""The analysis of discrete choice experiments", organized by the Centre for Bayesian Statistics in Health Economics, University of Sheffield (UK), September, 2007.
Fuzzy soft set approach in decision making plays a crucial role by using Dempster–Shafer theory of evidence. First, the uncertain degrees of several parameters are obtained via grey relational analysis that apply to calculate the grey mean relational degree. Secondly, a mass functions of different independent choices with several parameters have given according to the uncertain degree. Lastly, aggregate the choices into a collective choices, Dempster’s rule of evidence combination have been utilized. The aforesaid soft computing based method have been applied on decision making problem.
Invited talk at the Focus Fortnight 8: ""The analysis of discrete choice experiments", organized by the Centre for Bayesian Statistics in Health Economics, University of Sheffield (UK), September, 2007.
Fuzzy soft set approach in decision making plays a crucial role by using Dempster–Shafer theory of evidence. First, the uncertain degrees of several parameters are obtained via grey relational analysis that apply to calculate the grey mean relational degree. Secondly, a mass functions of different independent choices with several parameters have given according to the uncertain degree. Lastly, aggregate the choices into a collective choices, Dempster’s rule of evidence combination have been utilized. The aforesaid soft computing based method have been applied on decision making problem.
Computational Pool-Testing with Retesting StrategyWaqas Tariq
Pool testing is a cost effective procedure for identifying defective items in a large population. It also improves the efficiency of the testing procedure when imperfect tests are employed. This study develops computational pool-testing strategy based on a proposed pool testing with re-testing strategy. Statistical moments based on this applied design have been generated. With advent of computers in 1980‘s, pool-testing with re-testing strategy under discussion is handled in the context of computational statistics. From this study, it has been established that re-testing reduces misclassifications significantly as compared to Dorfman procedure although re-testing comes with a cost i.e. increase in the number of tests. Re-testing considered improves the sensitivity and specificity of the testing scheme.
This presentation is on using repeated measures design in the area of social sciences, behavioural sciences, management, sports, physical education etc.
Group Testing with Test Errors Made EasierWaqas Tariq
Group testing is a cost effective procedure for identifying defective items in a large population. It also improves the efficiency of the testing procedure when imperfect tests are employed.This study develops computational group-testing strategy based on Kline et. al.,(1989) testing strategy. Statistical moments based on this applied design have been generated. With advent of digital computers in 1980`s, group-testing strategy under discussion is handled in the context of computationa statistics.
August 1, 2010. Design of Non-Randomized Medical Device Trials Based on Sub-Classification Using Propensity Score Quintiles, Topic Contributed Session on Medical Devices, (Greg Maislin and Donald B Rubin). Joint Statistical Meetings 2010, Vancouver Canada.
A growing literature in experimental economics examines the conditions under which cooperation can be sustained in social dilemma settings. In particular, several recent studies contrast cooperation levels in games in which the number of decision rounds is probabilistic to games in which the number of decision rounds is finite. We contribute to this literature by contrasting the evolution of cooperation in probabilistically and finitely repeated linear voluntary contribution public goods games (VCM). Consistent with past results, cooperation increases in MPCR, and in group size, holding MPCR constant. We also find, as the number of decision sequences increase, there is a pronounced decrease in cooperation in the final round of finite sequences compared to those with a probabilistic end round. We do not, however, find strong evidence that overall cooperation rates are affected by whether the number of decision rounds is finite or determined probabilistically.
There are many valid criticisms of P-values but the criticism that they are largely responsible for the reproducibility crisis has been accepted rather lightly in some quarters. Whatever the inferential statistic that is used, it is quite illogical to assume that as the sample size increases it will tend to show more evidence against the null hypothesis. This applies to Bayesian posterior probabilities as much as it does to P-values. In the context of P-values it can be referred to as the trend towards significance fallacy but more generally, for reasons I shall explain, it could be referred to as the anticipated evidence fallacy.
The anticipated evidence fallacy is itself an example of the overstated evidence fallacy. I shall also discuss this fallacy and other relevant matters affecting reproducible science including the problem of false negatives.
Webinar slides- alternatives to the p-value and power nQuery
What are the alternatives to the p-value & power? What is the next step for sample size determination? We will explore these issues in this free webinar presented by nQuery
Talk given at ISCB 2016 Birmingham
For indications and treatments where their use is possible, n-of-1 trials represent a promising means of investigating potential treatments for rare diseases. Each patient permits repeated comparison of the treatments being investigated and this both increases the number of observations and reduces their variability compared to conventional parallel group trials.
However, depending on whether the framework for analysis used is randomisation-based or model- based produces puzzling difference in inferences. This can easily be shown by starting on the one hand with the randomisation philosophy associated with the Rothamsted school of inference and building up the analysis through the block + treatment structure approach associated with John Nelder’s theory of general balance (as implemented in GenStat®) or starting on the other hand with a plausible variance component approach through a mixed model. However, it can be shown that these differences are related not so much to modelling approach per se but to the questions one attempts to answer: ranging from testing whether there was a difference between treatments in the patients studied, to predicting the true difference for a future patient, via making inferences about the effect in the average patient.
This in turn yields interesting insight into the long-run debate over the use of fixed or random effect meta-analysis.
Some practical issues of analysis will also be covered in R and SAS®, in which languages some functions and macros to facilitate analysis have been written. It is concluded that n-of-1 hold great promise in investigating chronic rare diseases but that careful consideration of matters of purpose, design and analysis is necessary to make best use of them.
Acknowledgement
This work is partly supported by the European Union’s 7th Framework Programme for research, technological development and demonstration under grant agreement no. 602552. “IDEAL”
There are many questions one might ask of a clinical trial, ranging from what was the effect in the patients studied to what might the effect be in future patients via what was the effect in individual patients? The extent to which the answer to these questions is similar depends on various assumptions made and in some cases the design used may not permit any meaningful answer to be given at all.
A related issue is confusion between randomisation, random sampling, linear model and true multivariate based modelling. These distinctions don’t matter much for some purposes and under some circumstances but for others they do.
Researchers use several tools and procedures for analyzing quantitative data obtained from different types of experimental designs. Different designs call for different methods of analysis. This presentation focuses on:
T-test
Analysis of variance (F-test), and
Chi-square test
Computational Pool-Testing with Retesting StrategyWaqas Tariq
Pool testing is a cost effective procedure for identifying defective items in a large population. It also improves the efficiency of the testing procedure when imperfect tests are employed. This study develops computational pool-testing strategy based on a proposed pool testing with re-testing strategy. Statistical moments based on this applied design have been generated. With advent of computers in 1980‘s, pool-testing with re-testing strategy under discussion is handled in the context of computational statistics. From this study, it has been established that re-testing reduces misclassifications significantly as compared to Dorfman procedure although re-testing comes with a cost i.e. increase in the number of tests. Re-testing considered improves the sensitivity and specificity of the testing scheme.
This presentation is on using repeated measures design in the area of social sciences, behavioural sciences, management, sports, physical education etc.
Group Testing with Test Errors Made EasierWaqas Tariq
Group testing is a cost effective procedure for identifying defective items in a large population. It also improves the efficiency of the testing procedure when imperfect tests are employed.This study develops computational group-testing strategy based on Kline et. al.,(1989) testing strategy. Statistical moments based on this applied design have been generated. With advent of digital computers in 1980`s, group-testing strategy under discussion is handled in the context of computationa statistics.
August 1, 2010. Design of Non-Randomized Medical Device Trials Based on Sub-Classification Using Propensity Score Quintiles, Topic Contributed Session on Medical Devices, (Greg Maislin and Donald B Rubin). Joint Statistical Meetings 2010, Vancouver Canada.
A growing literature in experimental economics examines the conditions under which cooperation can be sustained in social dilemma settings. In particular, several recent studies contrast cooperation levels in games in which the number of decision rounds is probabilistic to games in which the number of decision rounds is finite. We contribute to this literature by contrasting the evolution of cooperation in probabilistically and finitely repeated linear voluntary contribution public goods games (VCM). Consistent with past results, cooperation increases in MPCR, and in group size, holding MPCR constant. We also find, as the number of decision sequences increase, there is a pronounced decrease in cooperation in the final round of finite sequences compared to those with a probabilistic end round. We do not, however, find strong evidence that overall cooperation rates are affected by whether the number of decision rounds is finite or determined probabilistically.
There are many valid criticisms of P-values but the criticism that they are largely responsible for the reproducibility crisis has been accepted rather lightly in some quarters. Whatever the inferential statistic that is used, it is quite illogical to assume that as the sample size increases it will tend to show more evidence against the null hypothesis. This applies to Bayesian posterior probabilities as much as it does to P-values. In the context of P-values it can be referred to as the trend towards significance fallacy but more generally, for reasons I shall explain, it could be referred to as the anticipated evidence fallacy.
The anticipated evidence fallacy is itself an example of the overstated evidence fallacy. I shall also discuss this fallacy and other relevant matters affecting reproducible science including the problem of false negatives.
Webinar slides- alternatives to the p-value and power nQuery
What are the alternatives to the p-value & power? What is the next step for sample size determination? We will explore these issues in this free webinar presented by nQuery
Talk given at ISCB 2016 Birmingham
For indications and treatments where their use is possible, n-of-1 trials represent a promising means of investigating potential treatments for rare diseases. Each patient permits repeated comparison of the treatments being investigated and this both increases the number of observations and reduces their variability compared to conventional parallel group trials.
However, depending on whether the framework for analysis used is randomisation-based or model- based produces puzzling difference in inferences. This can easily be shown by starting on the one hand with the randomisation philosophy associated with the Rothamsted school of inference and building up the analysis through the block + treatment structure approach associated with John Nelder’s theory of general balance (as implemented in GenStat®) or starting on the other hand with a plausible variance component approach through a mixed model. However, it can be shown that these differences are related not so much to modelling approach per se but to the questions one attempts to answer: ranging from testing whether there was a difference between treatments in the patients studied, to predicting the true difference for a future patient, via making inferences about the effect in the average patient.
This in turn yields interesting insight into the long-run debate over the use of fixed or random effect meta-analysis.
Some practical issues of analysis will also be covered in R and SAS®, in which languages some functions and macros to facilitate analysis have been written. It is concluded that n-of-1 hold great promise in investigating chronic rare diseases but that careful consideration of matters of purpose, design and analysis is necessary to make best use of them.
Acknowledgement
This work is partly supported by the European Union’s 7th Framework Programme for research, technological development and demonstration under grant agreement no. 602552. “IDEAL”
There are many questions one might ask of a clinical trial, ranging from what was the effect in the patients studied to what might the effect be in future patients via what was the effect in individual patients? The extent to which the answer to these questions is similar depends on various assumptions made and in some cases the design used may not permit any meaningful answer to be given at all.
A related issue is confusion between randomisation, random sampling, linear model and true multivariate based modelling. These distinctions don’t matter much for some purposes and under some circumstances but for others they do.
Researchers use several tools and procedures for analyzing quantitative data obtained from different types of experimental designs. Different designs call for different methods of analysis. This presentation focuses on:
T-test
Analysis of variance (F-test), and
Chi-square test
This presentation deals with the basics of design of experiments and discusses all the three basic statistical designs i.e. CRD, RBD and LSD. Further it explains the guidelines for developing experimental research.
11 T(EA) FOR TWO TESTS BETWEEN THE MEANS OF DIFFERENT GROUPS11 .docxnovabroom
11 T(EA) FOR TWO TESTS BETWEEN THE MEANS OF DIFFERENT GROUPS
11: MEDIA LIBRARY
Premium Videos
Core Concepts in Stats Video
· Testing the Difference Between Two Sample Means
Lightboard Lecture Video
· Independent t Tests
Time to Practice Video
· Chapter 11: Problem 5
Difficulty Scale
(A little longer than the previous chapter but basically the same kind of procedures and very similar questions. Not too hard, but you have to pay attention.)
WHAT YOU WILL LEARN IN THIS CHAPTER
· Using the t test for independent means when appropriate
· Computing the observed t value
· Interpreting the t value and understanding what it means
· Computing the effect size for a t test for independent means
INTRODUCTION TO THE T TEST FOR INDEPENDENT SAMPLES
Even though eating disorders are recognized for their seriousness, little research has been done that compares the prevalence and intensity of symptoms across different cultures. John P. Sjostedt, John F. Schumaker, and S. S. Nathawat undertook this comparison with groups of 297 Australian and 249 Indian university students. Each student was measured on the Eating Attitudes Test and the Goldfarb Fear of Fat Scale. High scores on both measures indicate the presence of an eating disorder. The groups’ scores were compared with one another. On a comparison of means between the Indian and the Australian participants, Indian students scored higher on both of the tests, and this was due mainly to the scores of women. The results for the Eating Attitudes Test were t(544) = −4.19, p < .0001, and the results for the Goldfarb Fear of Fat Scale were t(544) = −7.64, p < .0001.
Now just what does all this mean? Read on.
Why was the t test for independent means used? Sjostedt and his colleagues were interested in finding out whether there was a difference in the average scores of one (or more) variable(s) between the two groups. The t test is called independent because the two groups were not related in any way. Each participant in the study was tested only once. The researchers applied a t test for independent means, arriving at the conclusion that for each of the outcome variables, the differences between the two groups were significant at or beyond the .0001 level. Such a small chance of a Type I error means that there is very little probability that the difference in scores between the two groups was due to chance and not something like group membership, in this case representing nationality, culture, or ethnicity.
Want to know more? Go online or to the library and find …
Sjostedt, J. P., Schumaker, J. F., & Nathawat, S. S. (1998). Eating disorders among Indian and Australian university students. Journal of Social Psychology, 138(3), 351–357.
LIGHTBOARD LECTURE VIDEO
Independent t Tests
THE PATH TO WISDOM AND KNOWLEDGE
Here’s how you can use Figure 11.1, the flowchart introduced in Chapter 9, to select the appropriate test statistic, the t test for independent means. Follow along the highlighted sequence of steps in Figure 1.
11 T(EA) FOR TWO TESTS BETWEEN THE MEANS OF DIFFERENT GROUPS11 .docxhyacinthshackley2629
11 T(EA) FOR TWO TESTS BETWEEN THE MEANS OF DIFFERENT GROUPS
11: MEDIA LIBRARY
Premium Videos
Core Concepts in Stats Video
· Testing the Difference Between Two Sample Means
Lightboard Lecture Video
· Independent t Tests
Time to Practice Video
· Chapter 11: Problem 5
Difficulty Scale
(A little longer than the previous chapter but basically the same kind of procedures and very similar questions. Not too hard, but you have to pay attention.)
WHAT YOU WILL LEARN IN THIS CHAPTER
· Using the t test for independent means when appropriate
· Computing the observed t value
· Interpreting the t value and understanding what it means
· Computing the effect size for a t test for independent means
INTRODUCTION TO THE T TEST FOR INDEPENDENT SAMPLES
Even though eating disorders are recognized for their seriousness, little research has been done that compares the prevalence and intensity of symptoms across different cultures. John P. Sjostedt, John F. Schumaker, and S. S. Nathawat undertook this comparison with groups of 297 Australian and 249 Indian university students. Each student was measured on the Eating Attitudes Test and the Goldfarb Fear of Fat Scale. High scores on both measures indicate the presence of an eating disorder. The groups’ scores were compared with one another. On a comparison of means between the Indian and the Australian participants, Indian students scored higher on both of the tests, and this was due mainly to the scores of women. The results for the Eating Attitudes Test were t(544) = −4.19, p < .0001, and the results for the Goldfarb Fear of Fat Scale were t(544) = −7.64, p < .0001.
Now just what does all this mean? Read on.
Why was the t test for independent means used? Sjostedt and his colleagues were interested in finding out whether there was a difference in the average scores of one (or more) variable(s) between the two groups. The t test is called independent because the two groups were not related in any way. Each participant in the study was tested only once. The researchers applied a t test for independent means, arriving at the conclusion that for each of the outcome variables, the differences between the two groups were significant at or beyond the .0001 level. Such a small chance of a Type I error means that there is very little probability that the difference in scores between the two groups was due to chance and not something like group membership, in this case representing nationality, culture, or ethnicity.
Want to know more? Go online or to the library and find …
Sjostedt, J. P., Schumaker, J. F., & Nathawat, S. S. (1998). Eating disorders among Indian and Australian university students. Journal of Social Psychology, 138(3), 351–357.
LIGHTBOARD LECTURE VIDEO
Independent t Tests
THE PATH TO WISDOM AND KNOWLEDGE
Here’s how you can use Figure 11.1, the flowchart introduced in Chapter 9, to select the appropriate test statistic, the t test for independent means. Follow along the highlighted sequence of steps in Figure 1.
Directions The purpose of Project 8 is to prepare you for the final.docxeve2xjazwa
Directions: The purpose of Project 8 is to prepare you for the final, comprehensive exam and is set up EXACTLY the same. Questions 1 and 2 are not graded in this exercise, but are on the final. Be sure to answer them still so you can receive feedback. Once done with these, move into the calculation questions.
Be advised that you will need to decide which type of test to use in most of the problems. Please write out all pertinent information for each of the 4 steps of hypothesis testing. For the calculations, you only need to provide the values of all statistics for that test. There is no need to show work.
List the four Steps of the Hypothesis test:
Step 1 –
Step 2 –
Step 3 –
Step 4 –
This semester we have discussed the following statistical analyses.
Z-test
One-Sample
t
-test
Independent Groups
t
-test
Repeated Measures
t
-test
One-Way ANOVA
Repeated Measures ANOVA
Correlation
When do you use them? Please type your answer in the Test Used column.
ơ is given
µ is given
Groups Compared
Test Used
No
No
Looks at the same group at 2 different times or across two different conditions
Yes
Yes
Sample against population
Examines the degree to which two variables relate to one another
No
No
Looks at the same group at 2 or more times or across 2 or more conditions
No
No
Examines mean differences between two different groups
No
Yes
Sample against population
No
No
Examines mean differences between 2 or more groups
1. A researcher for a cereal company wanted to demonstrate the health benefits of eating oatmeal. A sample of 9 volunteers was obtained and each participant ate a fixed diet without any oatmeal for 30 days. At the end of the 30-day period, cholesterol was measured for each individual. Then the participants began a second 30-day period in which they repeated exactly the same diet except that they added 2 cups of oatmeal each day. After the second 30-day period, cholesterol levels were measured again and the researcher recorded the difference between the two scores for each participant. For this sample, cholesterol scores average M = 16 points lower with the oatmeal diet with SS = 538 for the difference scores.
10 points
·
Are the data sufficient to indicate a significant change in cholesterol level? Use a two-tailed test with α = .01.
·
Compute r
2
to measure the size of the treatment effect.
2. One possible explanation for why some birds migrate and others maintain year round residency in a single location is intelligence. Specifically, birds with smaller brain, relative to their body size, are not simply smart enough to find food during the winter and must migrate to warmer climates where food is easily available. Birds with bigger brains, on the other hand, are more creative and can find food even when the weather turns harsh. Following are hypothetical data similar to the actual results. The numbers represent relative brain size for the individual birds in each sample.
10 points
Non-Migrating
S.
Question 1.A group of researchers is replicating an earlier .docxIRESH3
Question 1.
A group of researchers is replicating an earlier experiment that indicated that participants who received task-specific feedback were more likely to persist at a task than participants who received more general, encouraging feedback. In an effort to ensure that participants are not treated differently based on the condition that they are in, the researchers automate all of the procedures and follow a written protocol when interacting with the participants. The researchers are trying to minimize:
placebo effects.
demand characteristics.
experimenter expectancy effects.
participant suspicion effects.
Question 2.
In a study examining the effects of heredity on intelligence, researchers compare the correlation of intelligence test scores of identical twins with the correlation of intelligence test scores for fraternal twins. In this experiment, the researcher is assuming that the comparison of identical and fraternal twins is a measure of heredity. This is an example of a ________________ inference.
construct
statistical
generalizability
Causal
Question 3
Researchers interested in studying the effect of happiness on various health outcomes randomly assign each person who comes in to the laboratory to one of two study conditions. However, several people in the study are friends and drove to the study together. The group of friends indicates that they need to be in the same condition of the study so that they can all leave at the together to get home. Accommodating the subjects' request might threaten validity because of the effect of:
regression to the mean.
attrition.
maturation.
selection.
Question 4
In an experiment on the effects of everyday stress on memory, a researcher has participants record every hour how much stress they are feeling and then complete a short-term memory task. The results of the study reveal that everyday stress may affect short-term memory. After evaluating the results of the study, however, the researcher is concerned that people who have high scores on neuroticism questionnaires are more likely to report stress and exhibit memory problems than people who have low scores. The researcher is worried about __________ validity.
construct
internal
statistical conclusion
external
Question 5
__________ validity concerns the generalizability of findings beyond the present study.
Ecological
Construct
Statistical conclusion
External
Question 6
A researcher is investigating the ability of aversive punishment to decrease students' disruptive behaviors in class. She is worried that the number of punishments will vary from student to student and thus will bias the results of the study. The researcher would do well to:
run a pilot test before conducting the study.
manipulate participants' knowledge about the study.
use a yoked control-group.
use a red herring technique.
Question 7
A psychologist is examini ...
1. The Effect of Time Pressure on Solving
Optimal Stopping Problems
William Teng
94396558/Michael Dawson Nunez
wteng1@uci.edu
Abstract
Optimal decision process has repeatedly been explored within
experimental psychology in order to simulate and understand real-
world decision-making. This concept is a highly relevant idea used
in fields spanning from psychology to finance. The purpose of our
current study is to bring about learning of optimal decision-making
within subjects, and to examine differences in performance
between conditions varying in amount of alternatives presented
through the use of an optimal stopping task. Our results failed to
yield any statistically significant evidence indicating learning.
However, subjects belonging to the shorter condition were found to
demonstrate significantly higher performance than subjects
participating in the longer condition.
1 Introduction
Nearly every instance of human behavior and interaction is narrated by a series of
decisions. Given the complex and competitive nature of modern societies, learning
to make sound and efficient decisions has become a crucial and necessary trait for
attaining successes and rewards. However, one is often faced with the difficult task
of making decisions without complete knowledge of all relevant information and
alternatives. As Campbell and Lee (2006) states, “Many real world decision-making
problems are sequential in nature. A series of choices is made available over time,
and it is often efficient (and sometimes even necessary) to make a selection without
waiting to be presented with all of the alternatives” (p. 1068). This process of
decision-making can be studied in an experimental setting using what is known as
an “optimal stopping” task (otherwise referred to as the “secretary problem”).
An optimal stopping experiment simulates real world decision-making by
presenting an observer with a series of possible choices, at which point the observer
must decide whether to accept or reject the current choice as the best alternative.
Each alternative is only presented once and can only be chosen during the time of
presentation. This task urges participants to utilize an “optimal (rational) decision
process” which involves choosing the first alternative containing a value that is both
greater than the optimal threshold (a decreasing function of optimal points
dependent on a choice’s serial positioning within a series) as well as all previously
presented alternatives (Vandekerckhove, 2014, p. 12).
2. Campbell and Lee (2006) studied learning in optimal decision making
through the use of a five-point problem optimal stopping task, in which each
problem series contained a set of five choices. The choices were comprised of two-
decimal point numbers uniformly distributed across a scale from 0 – 100. They also
studied the effects of feedback on learning by manipulating for conditions of no
feedback, outcome feedback, and full feedback. Financial rewards were also
controlled for by providing one group of subjects with no financial incentive,
whereas, the other was paid five dollars for participating with room for additional
monetary rewards if predetermined quotas of correct responses were met.
The results of their experiment yielded no significant evidence that would
suggest learning during the task. Feedback and financial rewards also failed to
demonstrate any significant effects towards learning. Nonetheless, when overall
performance was analyzed, an interaction between feedback and reward was found
to increase optimal decision-making but still showed no signs of learning (Campbell
& Lee, 2006, p. 1073).
The purpose of our current study is to bring about learning of optimal
decision-making within subjects, as well as to examine differences in performance
between conditions varying in numbers of alternatives presented through the use of
an optimal stopping task. To do so, we will utilize a 2X3 mixed ANOVA model and
a 2X4 mixed ANOVA model using series of numbers as our stimuli. The
manipulated variables are length of series and size of blocks. The variable “length”
will be controlled to be either series of three number choices (short) or of seven
number choices (long). Furthermore, data from 120 trials will then be divided into
three blocks consisting of 40 trials for model 1; and four 30 trials blocks for model
2. We hypothesize that rational decision-making performance will improve between
the first block and the last indicating an occurrence of learning. Secondly, we
predict that overall performance of correct rational decisions will be higher for the
short condition than the long condition.
2 Experiment
2.1 Participants
For this optimal stopping experiment, we enlisted a total of 71 participants. All
participants were recruited from the same population of only University of
California, Irvine undergraduate students belonging to the Psychology subject’s
pool. 45 subjects completed the task under the long condition, whereas, the
remaining 26 were tested in the short condition.
2.2 Stimuli
The experiment was performed through the use of 120 generated series of numbers
as its stimuli. There existed two experimental conditions defined as either short or
long. The short conditions generated series of three numbers as opposed to the long
conditions, which utilized series of seven numbers. All numbers generated were
two-decimal point values and fell between the range of 0 – 100.
2.3 Procedures
Each subject was first presented a demonstration trial by his or her experimenter in
order to ensure full understanding of the task presented. For all subsequent trials,
the participants were presented series of numbers and they were asked to determine
the maximum value of each series. Each number within a series was only presented
once and could only be accepted or rejected as the maximum value while being
shown. Depending on the assigned condition, either the third or seventh numbers
were forced selections since they were the last of each respective series. All
3. participants repeated the task for a total of 120 trials, which were then divided into
three blocks of 40 and four blocks of 30. Every participant completed the blocks in
the same order, however, the series within each block were randomized. After each
trial, immediate outcome feedback was given informing participants whether they
were correct or incorrect in choosing the maximum value.
3 Results
Figure 1:
Figure 1: A plotted graph displaying mean accuracy of correct rational decision for
the “long” and “short” conditions in model 1. Both conditions were separated into
three different blocks. A significant difference in performance was found between
the three blocks [F (2,138) = 30.461, p < 0.01]. Contrast revealed that subjects
made significantly more rational decisions in block 1 than in block 2 [F (1,69) =
50.780, p < 0.01, r² = 0.4239] and block 3 produced more rational decisions than
that of block 2 [F (1,69) = 16.115, p < 0.01, r² = 0.1893]. Length also displayed a
significant main effect [F (1,69) = 68.377, p < 0.01, r² = 0.4977]. An interaction
effect was also found between length of condition and blocks [F (2,138) = 40.039, p
< 0.01]. Contrast revealed that the interaction effect of condition length and blocks
differed significantly between block 1 and block 2 [F (1,69) = 5.459, p = 0.022, r² =
0.0733] as well as between block 2 and block 3 [F (1,69) = 96.888, p < 0.01, r² =
0.5841].
4. A mixed-design ANOVA for model 1 revealed a significant main effect of blocks on
mean rational decision correctness [F (2,138) = 30.461, p < 0.01]. Contrast of the
three blocks displayed significantly better performances in rational decision-making
during block 1 than block 2 [F (1,69) = 50.780, p < 0.01, r² = 0.4239]. Block 3 was
also found to produce more rational decisions than block 2 [F (1,69) = 16.115, p <
0.01, r² = 0.1893] but performance in block 1 was higher than block 3 [F (1,70) =
4.705, p = 0.033, r² = 0.063]. A significant main effect of condition length was
found [F (1,69) = 68.377, p < 0.01, r² = 0.4977]. Furthermore, an interaction effect
was discovered between condition length and blocks [F (2,138) = 40.039, p < 0.01].
Additional contrast demonstrated that the effects of condition length on blocks
differed significantly between block 1 and block 2 [F (1,69) = 5.459, p = 0.022, r² =
0.0733] as well as between block 2 and block 3 [F (1,69) = 96.888, p < 0.01, r² =
0.5841].
Figure 2:
Figure 2: A Scatter-plot displaying a positive correlation between the proportion of
rational decisions for each subject in block 1 and block 2 [r = 0.679, p < 0.01].
Correlation analysis of block 1 vs. block 2 revealed a positive correlation
between the proportion of rational decision making of the two blocks [r = 0.679, p <
0.01].
5. Figure 3:
Figure 3: A Scatter-plot displaying the positive correlation between the proportion
of rational decisions for each subject in block 2 and block 3 [r = 0.391, p = 0.001].
Correlation analysis of block 2 vs. block 3 revealed a positive correlation
between the proportion of rational decision making of the two blocks [r = 0.391, p =
0.001].
Figure 4:
Figure 4: A Scatter-plot displaying the positive correlation between the proportion
of rational decisions for each subject in block 1 and block 3 [r = 0.391, p = 0.001].
6. Correlation analysis of block 1 vs. block 3 revealed a positive correlation
between the proportion of rational decision making of the two blocks [r = 0.391, p =
0.001].
Figure 5:
Figure 5: Figure 1: A plotted graph displaying mean accuracy of correct rational
decision for the “long” and “short” conditions in model 2. Both conditions were
separated into four different blocks. A significant difference in performance was
found between the four blocks [F (3,207) = 15.888, p < 0.01]. Contrast revealed
that subjects made significantly more rational decisions in block 1 than in block 2
[F (1,69) = 4.557, p = 0.036, r² = 0.062] and block 3 produced less rational
decisions than that of block 2 [F (1,69) = 22.323, p < 0.01, r² = 0.2444] while block
4 yielded more rational decisions than block 3 [F (1,69) = 24.166, p < 0.01, r² =
0.2594]. Length also displayed a significant main effect [F (1,69) = 68.377, p <
0.01, r² = 0.4977]. An interaction effect was also found between length of condition
and blocks [F (3,207) = 21.120, p < 0.01]. Contrast revealed that the effect of
condition length differed significantly between block 1 and block 2 [F (1,69) =
6.163, p = 0.015, r² = 0.082] and between block 3 and block 4 [F (1,69) = 33.125, p
< 0.01, r² = 0.3244] but not between block 2 and block 3 [F (1,69) = 2.981, p =
0.089].
A repeated measures ANOVA for model 2 yielded a significant main effect
of blocks on optimal decision making [F (3,207) = 15.888, p < 0.01]. A significant
effect of length was also discovered [F (1,69) = 68.377, p < 0.01, r² = 0.4977].
Interaction between blocks and length once again proved to be significant [F (3,207)
= 21.120, p < 0.01]. Contrast for blocks revealed that subjects made significantly
7. more rational decisions in block 1 than in block 2 [F (1,69) = 4.557, p = 0.036, r² =
0.062] and block 3 produced less rational decisions than that of block 2 [F (1,69) =
22.323, p < 0.01, r² = 0.2444] while block 4 yielded more rational decisions than
block 3 [F (1,69) = 24.166, p < 0.01, r² = 0.2594]. Additional contrast of interaction
demonstrated that the effect of condition length differed significantly between block
1 and block 2 [F (1,69) = 6.163, p = 0.015, r² = 0.082] and between block 3 and
block 4 [F (1,69) = 33.125, p < 0.01, r² = 0.3244] but not between block 2 and block
3 [F (1,69) = 2.981, p = 0.089].
Figure 6:
Figure 6: A Scatter-plot displaying the positive correlation between the proportion
of rational decisions for each subject in block 1 and block 2 of model 2 [r = 0.651,
p < 0.01].
Correlation analysis of block 1 vs. block 2 of model 2 revealed a positive
correlation between the proportion of rational decision making of the two blocks [r
= 0.0.651, p < 0.01].
4 Discussion
For the current study, participants were tested using an optimal stopping task in
order to explore the optimal decision process. The experimental design was
manipulated to have both a short and long condition for the sake of illustrating the
effects condition length has on optimal decision performance. Furthermore, all data
were separated into blocks to determine whether learning occurred across trials
The results successfully showed a significant main effect of condition
length on subsequent optimal decision performance. The short condition was proven
to elicit better performance than the long condition, thus, confirming our first
hypothesis. This is an interesting discovery because it has direct applications to
everyday situations. Perhaps, as common practices often demonstrate, an increase in
8. choices causes complexity and promotes indecisiveness; consequently, the ability to
make proper and efficient decisions begins to decline. Another fascinating point to
consider is the emotional impact an individual may experience when faced with too
many alternatives. It may be that high number of choices can prove overwhelming;
resulting in apprehension and anxiety, which then affects decision-making. An
alternative optimal stopping study could manipulate emotion and mood through
some version of priming and analyze whether positive vs. negative moods have a
significant impact on ability to follow the rational rule.
Our second hypothesis, however, failed to generate any statistical support.
Initially, we predicted an increase in optimal decision performance from early
blocks to later blocks due to learning; unfortunately, results showed statistically
significant decreases in performance across trials. For both models used, decision-
making using the rational rule decreased from blocks 1 to blocks 2, however,
performance began to slightly increase from blocks 2 towards blocks 3 or 4.
Correlation analysis would reveal a positive linear relationship in proportion
rationality between blocks 1 and blocks 2 in both models. This suggests that a
significant amount of variability can be accounted for by individual differences in
decision-making abilities.
In contrast, the correlational value between blocks 2 and blocks 3, or blocks
4 for model 2, is low so it would be dangerous to explain variation under the
assumption of individual differences (despite a statistically significant positive
correlation). Instead, what may have caused the slight increase in performance could
be attributed to the interaction effect found between length and blocks. Upon further
examination of Figure 1, it is worth noting that the short condition followed a trend
of constant decrease in optimal decisions, whereas, the long condition seemed to
increase during the last block of trials. Perhaps this is an indication of learning
under longer condition. This would be an interesting implication and is worth
further studies using even larger series of choices. Another factor that may have
affected learning was the feedback given at each trial. The feedbacks were designed
to inform subjects of correct acceptance of the maximum value as opposed to the
optimal value. It is possible that learning was inhibited because participants
received negative feedback despite correctly identifying an optimal value.
In conclusion, the current study was able to successfully demonstrate that
shorter conditions yield higher levels of optimal decision performance. However,
like Campbell and Lee’s (2006) experiment, significant evidence of learning failed
to be collected. Nonetheless, this study has raised an interesting question regarding
the effects of emotion on optimal decision-making. Furthermore, it may have also
shown slight implications of learning. Additional research on optimal decision
process should be considered and will be crucial in our quest to understanding the
fundamental luxury and right of human consciousness, freedom of choice.
References
Campbell, J., & Lee, M. (2006). The effect of feedback and financial reward on
human performance solving ‘secretary’ problems. 1068-1073. Retrieved
from https://eee.uci.edu/14w/68200/psych112bw/p1068 1 .pdf
Vandekerckhove, J. (2014). Lecture 8 [PowerPoint slides]. Retrieved from
https://eee.uci.edu/14w/68200/psych112bw/lec8_p112bw.pdf`