Original Research Dynamic MouselabWEB: Individualized information display matrixes Zeliha Yıldırım1 and Semra Erpolat Taşabat2 1 Middle East Technical University, Ankara, Turkey 2 Mimar Sinan University, Istanbul, Turkey This paper introduces Dynamic MouselabWEB, a com- puterized process tracing tool that was designed to cre- ate flexible decision-making settings that are similar to real life. While Dynamic MouselabWEB is an extension of MouselabWEB, it differs in that it creates individual- ized information display matrixes (IDMs) rather than pre- senting predetermined IDMs, so participants decide on the attributes and alternatives before the decision task. This structure can improve the involvement of decision- makers in the decision process and it gives researchers a chance to observe decision-making behaviors and to ex- plore new decision-making strategies when the decision task has only the decision-makers’ important attributes and appealing alternatives. In order to measure the effect of this dynamic structure, two groups of students worked on job selection tasks, one in the Dynamic MouselabWEB program (n = 32) and one in the traditional Mouselab- WEB program (n = 20). Results indicate a significant dif- ference between the decision-making behaviors of these two groups. The students who chose a job on Dynamic MouselabWEB were informed more and they spent more time on task than the other group; in other words, they were involved more in the decision process. Keywords: decision-making, process tracing, information display matrix, MouselabWEB This article consists of three main sections; the first sec-tion is dedicated to a literature review in order to make the readers familiar with the background informa- tion, the second is created for technical aspects of the soft- ware, and the last section explains the experiment. Literature review Decision models There are mainly two approaches to studying decision- making behavior: 1) the outcome-based approach and 2) the process tracing approach (Harte & Koele, 2001). The outcome-based approach attempts to construct mathematical models for the relationship between in- put (information) and output (decisions) to reveal the cognitive patterns underlying decision-making pro- cesses. This approach has been applied for over two centuries (Abelson & Levi, 1985; Brehmer, 1994; Dawes, 1979; Einhorn, Kleinmuntz, & Kleinmuntz, 1979; Ford, Schmitt, Schechtmann, Hults, & Doherty, 1989; Westenberg & Koele, 1994), but it has a large number of limitations. For example, the approach im- plies fitting mathematical models to data on decisions (output) in various situations (input) in order to infer the underlying decision-making processes without tak- ing process data into account. According to Svenson (1979), these models provide surface descriptions of the processes rather than a detailed information con- cerning stages in actual decision processes. As a result, researchers have suggested that cognitive processes cannot be sufficiently understood by solely studying input-output relationships and other research methods need to be included (Ford, et al., 1989; Payne, Braun- stein, and Carroll, 1978; Svenson, 1979). Researchers can overcome these concerns by employing the process- tracing approach, which focuses on patterns of infor- mation acquisition rather than the output (i.e., the de- cisions). In most process-tracing studies, information is presented in an information display matrix which has at least two alternatives which are characterized by at least two attributes (Ford et al., 1989). By ob- serving the information acquisition process, such as the amount of time spent making decisions, sequences of information acquisition and the amount of infor- mation accessed, researchers have gained insights into cognitive processes and developed more accurate pre- dictive models (Einhorn & Hogarth, 1981). Process tracing methods Several methods have been used to monitor informa- tion acquisition processes, such as verbal protocols (Jarvenpaa, 1989), eye movement recording (Russo & Dosher, 1983), information display boards (Payne, 1976), and the Mouselab system (Payne, Bettman, & Johnson, 1988). In earlier process-tracing studies, such as those carried out toward the end of the 1970s, participants were asked to make a decision by look- ing at information on cards located in envelopes on an information display board (Payne, 1976). In order to access information about a particular attribute for a particular alternative, the participant had to open the associated envelope and read the card, which would then be placed back in the envelope. In the meantime, Corresponding author: Zeliha Yıldırım, Middle East Technical University, Üniversiteler Mahallesi Dumlupınar Bulvarı No:1, Ankara, Turkey; e-mail: zl- hyldrm@gmail.com 10.11588/jddm.2019.1.63149 JDDM | 2019 | Volume 5 | Article 4 | 1 mailto:zlhyldrm@gmail.com mailto:zlhyldrm@gmail.com https://doi.org/10.11588/jddm.2019.1.63149 Yıldırım & Taşabat: Dynamic MouselabWEB the researcher would write down the sequence of ac- quisitions and the number of times that the envelopes were opened. Computerized process tracing, such as the Mouselab system, employs computer graphics to display information displaying an information display matrix (Payne et al., 1988). With this system, the values of attributes are concealed in boxes on a screen instead of in envelopes. Each box is opened when the mouse cursor is moved over it (mouseover), and it re- mains open until the cursor is moved away (mouseout). Only one box can be opened at a time (Payne et al., 1988). In addition to offering all the features of Mouse- lab, MouselabWEB makes it possible to carry out research on the Internet (Willemsen & Johnson 2004, 2011). MouselabWEB (www.mouselabweb.org), which uses Web technology (Javascript, HTML, PHP, and MySQL), has open-source codes, released under GNU General Public License v3.0, so researchers can add new features. Moreover, participants do not need to download plugins or other software; all they need is access to the Internet and a mouse to use a Mouselab- WEB page. Along with MouselabWEB there are other computerized process tracing tools, such as ISCube (Tabatabai, 1998), MouseTrace (Jasper & Shapiro, 2002), phased narrowing (Jasper & Levin, 2001; Levin & Jasper, 1995), active information search (Huber, Wider, & Huber, 1997; Williamson, Ranyard, & Cuth- bert, 2000), P1198 (Andersson, 2001) and Computer- Shop (Huneke, Cole, & Levin, 2004). Decision strategies According to Payne, Bettman, and Johnson (1992), a decision strategy is a “sequence of mental and ef- fector (actions on the environment) operations that transform some initial state of knowledge into a fi- nal knowledge state so that the decision maker per- ceives that the particular decision problem is solved” (p. 109). Knowing about decision makers’ cognitive processes enables us to infer the decision strategies used and it also enables us to predict future decisional behavior and decision outcomes (Payne, Braunstein, & Carroll, 1978). Furthermore, if we are better in- formed about people’s decision strategies we can de- sign better decision support systems (Browne, Pitts, & Wetherbe, 2007; Montgomery, Hosanagar, Krishnan, & Clay, 2004; Bettman et al., 1993). Although it is sel- dom possible to precisely identify a particular decision strategy, researchers can identify types of strategies by using process measures (Ford et al., 1989). In gen- eral, decision strategies can be classified as being either compensatory or non-compensatory. With compen- satory strategies, decision makers make tradeoffs be- tween different values of multiple attributes and they engage in extensive information processing (Stevenson et al., 1990). When a large amount of information is gathered in the process, this usually indicates that par- ticipants are employing compensatory decision-making strategies (Ford et al., 1989). The weighted adding strategy, for example, is a compensatory strategy that requires decision makers to multiply each attribute’s subjective value by its importance weight to obtain an overall value for each alternative (Abelson & Levi, 1985). Non-compensatory strategies are selective in the sense that decision makers restrict their attention to only part of the available information (Beach & Mitchell, 1978) and eliminate alternatives that do not meet their requirements. One example of this is the lexicographic strategy (LEX), which selects the option with the best value for the most important attribute. According to the cost/benefit framework, the choice of a decision strategy is a function of both costs, i.e. the effort required to use a rule, and benefits, i.e. the abil- ity of a strategy to select the best alternative (Beach & Mitchell, 1978; Russo & Dosher, 1983). By using this framework, the decision maker has a large repertoire of decision strategies, and he/she chooses one of these strategies contingent on the decision tasks’ character- istics, such as complexity of the decision task and time pressure, to adapt to decision settings (Bettman et al., 1993). In addition to decision task-based variables, an- other important variable that affects the use of various decision-making strategies is the level of involvement of the decision maker. Involvement is considered to be an important mediating variable of consumer behavior (Mitchell, 1979). For example, an extensive informa- tion search might indicate a higher level of involvement (Mittal, 1989). Process measures When participants process information in a decision task, computerized process tracing tools record time- dependent mouse events, e.g., mouseover (opening of boxes), mouseout (closing of boxes), the order in which boxes are opened, the amount of time boxes remain open, selected options, and total elapsed time since the display first appears on the screen automatically. By using this event-based data, several measures can be computed to characterize decision-making behav- ior, which are generally divided into depth, content, and sequence of searches (Ford et al., 1989; Jacoby, Jaccard, Kuss, Troutman, & Mazursky, 1987; Jasper & Shapiro, 2002, p. 370). Depth measures focus on de- termining how extensively the participants attempted to acquire information. This includes decision-making time (Hogarth, 1975; Pollay, 1970), proportions of in- formation sought (Payne, 1976), reacquisition rates (Jacoby, Chestnut, Weigl, & Fisher, 1976), number of acquisitions (Bettman et al., 1993), and the av- erage amount of time spent per item of information acquired (Bettman et al., 1993). Content measures are used to quantify the relative weights assigned to the various types of information and they refer to ex- actly what information was acquired and which op- tions were chosen. Content measures include the total amount of time spent on the information in the boxes (Bettman et al., 1993), time spent on the most im- portant attribute/-s, and the proportion of time spent on the most important attribute. It is common to ask subjects to rate the importance they assign to at- 10.11588/jddm.2019.1.63149 JDDM | 2019 | Volume 5 | Article 4 | 2 https://doi.org/10.11588/jddm.2019.1.63149 Yıldırım & Taşabat: Dynamic MouselabWEB tributes by using a Likert rating scale question after the decision task. When this is done, the attribute with the highest rating is considered to be the most important one. Sequence measures are used to de- scribe whether information selection behavior was pri- marily attribute- or alternative-based. An alternative- based search indicates that decision makers first con- sider several attributes of the same alternative before proceeding to the next one. In contrast, when they compare alternatives across attributes first, this is re- ferred to as an attribute-based approach. A common metric for defining search patterns is the Search Index (SI) (Payne, 1976), which measures the relative use of alternative-based versus attribute-based searches. The index ranges from -1 to +1, indicating a completely attribute-based or a completely alternative-based in- formation search, respectively. If both types of search patterns are used equally, the index is 0. A positive index value is often assumed to indicate the usage of compensatory strategies (e.g., weighted adding strate- gies), whereas a negative index value is interpreted as being an indicator for more non-compensatory strate- gies (e.g., elimination-by-aspects strategies). Böcken- holt and Hynan (1994) proposed the Strategy Measure (SM), another index for search patterns, as they found in a simulation study that the SI is unreliable when the number of alternatives and attributes is not iden- tical. When the number of attributes is larger than the number of alternatives, a positive SI index is more likely and when the number of attributes is smaller than the number of alternatives, a negative SI index is more probable. The SM index also ranges from -1 to +1. Other variables for sequences of searches are variability in the amount of information searched per option (Payne, 1976), and variability in the amount of information searched per attribute (Klayman, 1982). Lower values for these two measures indicate that de- cision processing is more evenly spread across all the attributes or alternatives, respectively. A new tool: Dynamic MouselabWEB MouselabWEB MouselabWEB focuses on the acquisition of infor- mation to make inferences on the nature of the un- derlying cognition and test existing decision models. To do that, it presents information on an attributes- alternatives matrix. In most computer-based displays, the attribute values are hidden behind boxes. Partic- ipants navigate through a task by moving the mouse and they can reveal information as much as they need to make a decision. Most computers record time and sequence of acquisitions with sufficient precision (1/60 th of a second), resulting in a record of box open- ings and closings. The interpretation of information- acquisition data is based on two testable assumptions (Costa-Gomes, Crawford & Broseta, 2001) about the relationship between search and cognition: The first, occurrence, states that if information is used by a decision-maker, it must have been seen by opening a box. The second, adjacency, assumes that informa- tion is acquired, rather than memorized because of limitations in short-term memory and the low cost of (re-)acquisition. Motivation behind Dynamic MouselabWEB Dynamic MouselabWEB was motivated by the follow- ing aims: 1. Overcome limitations of predetermined informational structures One of the limitations of IDMs is that the researcher has to determine the options, attributes and infor- mation available (Jacoby et al., 1987). This can be alleviated by using an individualized structure., Aschemann-Witzel and Hamm (2011) suggests as one of the possibilities for further development of IDMs: Another method suggested by Jacoby et al. (1987, p. 155) is an individualized IDM; this takes account of the criticism of IDM based on overly stark predefined informa- tional structures: Before the IDM-survey is carried out, each individual participant is asked which of the product attributes are im- portant to him or her, so that only these cri- teria are then offered in the matrix. (p. 4) 2. Create stimulus close to real-life decision settings In the age of the Internet, consumers tend to narrow down the options and criteria via online shopping web- sites so they can choose the best option. In light of this, researchers should prepare IDMs with this kind of dynamic structure. Allowing decision makers to cre- ate their own IDMs rather than working with prede- fined ones resembles certain real-life information envi- ronments more closely. 3. Encourage researchers to discover new decision strategies As Huber (1980) states, future research should not be limited to decision strategies which are discussed in the literature: The list of simple strategies discovered so far covers only a subset of all possible simple strategies. ... At least some of the simple strategies are incomplete. For example, the lexicographic-ordering strategy assumes that the decision maker first selects the most im- portant dimension. It is not clear what hap- pens if there are two or more most important dimensions. (pp. 187–188) 10.11588/jddm.2019.1.63149 JDDM | 2019 | Volume 5 | Article 4 | 3 https://doi.org/10.11588/jddm.2019.1.63149 Yıldırım & Taşabat: Dynamic MouselabWEB 4. Increase the involvement of participants in decision processes In decision research, involvement is directly related to motivation to get information. We believe that if par- ticipants can decide what they want to know about the options that they are considering, they will be more motivated to get more information in the decision pro- cess. 5. Make it freely available for other researchers Dynamic MouselabWEB is free software like Mouse- labWEB, so researchers can redistribute it and/or modify it under the terms of the GNU General Public License v3.0. Technical aspects Dynamic MouselabWEB is an extended version of MouselabWEB and uses all the features of Mouse- labWEB. In this section, we list the basic features of MouselabWEB and discuss the technical requirements for MouselabWEB, and then we explain the structure of Dynamic MouselabWEB. Basic features An experiment based on MouselabWEB generally has more than one HTML page, such as an intro- duction page, pages for warm-up tasks, a decision task page and a survey page. Each page is linked to the following page by defining “nextURL,” which is placed in the source code. MouselabWEB pages can be generated using the MouselabWEB Designer (http://www.mouselabweb.org/designer/index.html), an online editor program. The main purpose of the Designer is to generate the MouselabWEB boxes structure, which can be designed by using different features. According to the study aim, researchers can use default features or create new features. Mainly, MouselabWEB has these basic features: 1) there are different options for box openings and box closings, 2) boxes can be fixed or counterbalanced or random, 3) you can include a header row/column for alternatives and attributes, 4) the layouts of boxes and selection buttons can be changed by altering CSS files, 5) the background scripting automatically saves the data, and the process data is sent to and stored on a database, 6) a questionnaire can be included before/after the decision task. Dynamic features MouselabWEB uses MySQL, which is a relational database management system. We created three more tables in addition to the table that stores the process data to make MouselabWEB dynamic. The functions of these three tables are as follows: • Talternatives stores information about the alterna- tives, • Tattributes stores information about the attributes, • Tvalues stores the values of the attributes. The Electronic Supplementary Material (ESM) 1 ex- plains the columns of these three tables and the in- formation that they store. In order to use Dynamic MouselabWEB codes, you have to create these tables in a database. Our study tables can be created by running the SQL script in ESM 2. In Dynamic MouselabWEB two additional screens are present before the decision task: 1) “Choose the alternatives screen”, which uses information from the Talternatives table in the database (Figure 1 shows a capture of this screen; see ESM 3 for page source code. 2) “Choose the attributes screen”, which uses informa- tion from the Tattributes table in the database (Figure 2 shows a screen capture). This page’s source code is in Electronic Supplementary Material 4. Although we choose to display these two screens in that order, it is possible to change the order of the screens or omit one of them and use predefined at- tributes /alternatives instead. After these two selections, the names of columns and rows, and information of boxes can be obtained from the database by using the tables’ associations. To do this, we added steps in the source code of the decision task screen. The source code of the decision task is provided ESM 5. For a screen capture of a decision task, see Figure 3. To create a dynamic decision task, researchers should determine the alternatives, the attributes, and the attributes’ values in advance. Then they need to create three tables in the database to keep their in- formation and should make necessary changes in the codes of the sets of alternatives/attributes. Figure 4 illustrates the process of creating a study in Dynamic MouselabWEB. 10.11588/jddm.2019.1.63149 JDDM | 2019 | Volume 5 | Article 4 | 4 http://www.mouselabweb.org/designer/index.html https://doi.org/10.11588/jddm.2019.1.63149 Yıldırım & Taşabat: Dynamic MouselabWEB Figure 1. A screen capture of a set of alternatives screen. Figure 2. A screen capture of a set of attributes screen. Technical requirements Dynamic MouselabWEB has the same tech- nical requirements as MouselabWEB. We recommend to visit the download page (http://www.mouselabweb.org/download.php), where you can find the program files and a guide that explains how to install the software and launch an experiment. We encourage you to create a MouselabWEB page using MouselabWEB Designer to get familiar with the software. You can design an experiment in an offline environment or with an online version of MouselabWEB. Basically, you need a web server and PHP/MySQL capabilities on your system as well as a database to store the study data. You can find all the necessary files that you have to put in the study folder in ESM 6. Experiment In order to illustrate the dynamic structure of Dy- namic MouselabWEB, we conducted an experiment that compares the results of two decision tasks us- ing the same amount of information presented either in Dynamic MouselabWEB or in MouselabWEB (the control). We formulated three hypotheses based on the pro- cess measures: H1: The dynamic structure of a decision task has an effect on decision-making behavior. H2: Participants using a dynamic structure will access more information and spend a longer amount of time on a decision task, which indicates a difference in terms of involvement. H3: The Dynamic MouselabWEB group will place less attention on the most important attribute than the control group. We mainly wanted to make a general statement about the effect of the dynamic structure as it is stated in H1. Even though H1 is related to H2, we want to use different analysis techniques to test these two hy- potheses. We used more than one dependent variable to represent decision-making behavior to test H1. On the other hand, H3 suggests that these two groups have different approaches for the most important attribute. Since the dynamic decision task has more than one im- portant dimension, we assume that the group in this task give less attention to the most important attribute than the other group. Method Participants The target group was Turkish undergraduates study- ing in their final year in statistics departments. 32 un- dergraduates (19 female, 9 male, 4 not answered, Mage 10.11588/jddm.2019.1.63149 JDDM | 2019 | Volume 5 | Article 4 | 5 http://www.mouselabweb.org/download.php https://doi.org/10.11588/jddm.2019.1.63149 Yıldırım & Taşabat: Dynamic MouselabWEB Figure 3. A screen capture of a decision task. Figure 4. The flowchart of a study in Dynamic MouselabWEB. The dotted lines represent the background scripts that show screens and obtain information from the database, whereas the unbroken lines denote researchers/participants’ activities. = 22.9, SD = 0.71) from various universities selected a job on Dynamic MouselabWEB and 20 undergrad- uates (10 female, 10 male, Mage= 23.6, SD = 1.47) selected a job on MouselabWEB. Programs We designed two job selection tasks with four alter- natives and five attributes. The stimuli in Mouselab- WEB have a traditional setup in which the alterna- tives and the attributes are selected by the researcher, whereas in the dynamic decision task, participants can choose the alternatives and attributes from the sets before the task. Design In the Dynamic MouselabWEB study, to define the set of alternatives, we determined sectors in which people who graduated in Statistics might work. These sectors are IT, banking, insurance, industry, market research, and education. Then we used job search websites to determine the particular occupations. In the prepara- tion phase, we selected 21 jobs and prepared a study link that includes the decision task, and we sent this link to 10 undergraduates and asked for their feed- back. Based on their responses, we omitted 11 jobs and added 2 new jobs to the set. The set of attributes has 20 attributes which were collected from various studies. To determine the attribute values, we turned to eight specialists who have worked for many years in their sectors. We asked them to rate 20 attributes on a 7-point scale (from worst to best) for each job and each sector (overall rate). After the specialists had as- sessed all the jobs, we consolidated these assessments into a single table which has 20 attributes and 12 jobs. Procedure In the Dynamic MouselabWEB task, participants were asked to choose four jobs from the set of alternatives and five attributes from the set of attributes. The op- tion “I do not want to choose” was added to the sets, and if this option was selected, the predetermined al- ternatives/attributes were shown in the main decision task. In the MouselabWEB task, participants were asked to choose one job in the main decision task in which the alternatives and attributes were prede- fined. This decision task also had four jobs and five attributes. Both studies include introduction pages and two warm-up tasks before the stimuli. The survey links, which included the decision tasks, were sent to students between March and May 2018. They were 10.11588/jddm.2019.1.63149 JDDM | 2019 | Volume 5 | Article 4 | 6 https://doi.org/10.11588/jddm.2019.1.63149 Yıldırım & Taşabat: Dynamic MouselabWEB Table 1. Means and Standart Deviations of the Dependent Variables and t-test Results. Decision Task Group t-test for Equality of Means Dynamic MouselabWEB MouselabWEB Group (n = 32)) Group (n = 20) t df p Process measures Mean SD Mean SD Percent of unique boxes examined 0.84 0.18 0.70 0.26 2.33 50 .02 Number of acquisitions 42.16 19.25 26.60 15.66 3.04 50 <.01 Reacquisitions Number 25.41 17.14 12.70 11.66 3.18 50 <.01 Total time in decision (s) 42 24 31 16 2.10 50 .04 Time spent on the most important attribute (s) 8.32 5.90 8.86 7.20 -0.29 46 .78 Proportional time on the most important attribute 0.20 0.10 0.32 0.24 -2.19 24 .04 Strategy Measure -0.41 3.93 -0.95 3.20 0.52 50 .61 VarAlt 0.03 0.03 0.05 0.06 -1.59 50 .12 VarAtt 0.02 0.02 0.03 0.05 -1.46 22 .16 Note. VarAlt: variance in the proportion of time spentprocessing each alternative, VarAtt: variance in the proportion of time spent processing each attribute. The significance level is .05. informed about the project by an email and invited to the study. They did not receive any compensation in return. Data preparation The process data was used to create nine variables re- flecting on decision-making behavior. To measure the total amount of processing, we used four depth mea- sures: 1) total time spent on decision(s), 2) number of acquisitions, 3) number of reacquisitions, 4) percent- age of unique boxes examined. To assess selectivity in processing, we used two content measures: 1) time spent on the most important attribute(s), 2) propor- tional time spent on the most important attributes. To define the pattern of searches, we use three se- quence measures: 1) Strategy Measure, 2) variance in the proportion of time spent processing each alter- native (VARalt), 3) variance in the proportion of time spent processing each attribute (VARatt). Results Since we expected the dependent variables to be cor- related (0.2 < r < 0.8), we initially performed a Mul- tivariate Analysis of Variance (MANOVA) with de- pendent variables that meet MANOVA’s assumptions. These dependent variables are: Total time in decision- making, number of acquisitions, time spent on the most important attribute, and variance in proportion of time per alternative. The result of the MANOVA shows that the decision-making behavior of partici- pants who selected their attributes and alternatives before the decision task is statistically different from the control group, F (4,43) = 3.18, p = 0.02. As indi- cated in H1, we can conclude that the dynamic struc- ture of the decision task has an impact on decision- making behavior. An independent samples t-test was used to examine the effects in more detail. Table 1 presents the descriptive statistics (means and SDs) of the dependent variables and t-test results. Table 1 shows that the two groups are significantly different in terms of all the depth measures: percent of unique boxes examined, number of acquisitions, reac- quisitions number, total time in decision. As predicted in H2, the dynamic group processed more information (i.e., higher number of acquisitions, higher number of reacquisitions, and higher percentage of unique cells examined) and spent longer time in decision process, so we conclude that this group was involved more in decision-making. Lastly, the two groups differed in terms of “proportion of time spent on the most impor- tant attribute” (p = 0.04). As stated in H3, it can be concluded that when participants define the attributes and the alternatives of the decision task, they spend less time on the most important attribute. The data file is placed in Electronic Supplementary Material 7. Discussion The idea underlying Dynamic MouselabWEB emerged from the observation that there are no open-source tools available that can create individualized IDMs. In this paper, we have described Dynamic Mouse- labWEB, which was designed to create individualized IDMs that make it possible for participants to choose the attributes and the alternatives before the decision task. By using Dynamic MouselabWEB, researchers can avoid the problem of predetermined informational structures and generate more realistic decision set- tings. Additionally, we believe that if participants can decide what they want to know about the options that they will consider, they will be more motivated to get more information about the decision process. To define Dynamic MouselabWEB’s position in deci- sion studies, we can compare it with The method of Active Information Search (AIS). Huber, Wider, and Huber (1997) introduced the method of AIS, in which 10.11588/jddm.2019.1.63149 JDDM | 2019 | Volume 5 | Article 4 | 7 https://doi.org/10.11588/jddm.2019.1.63149 Yıldırım & Taşabat: Dynamic MouselabWEB the subject gets a basic description of the task, and has to ask questions to receive additional information. These questions are recorded and answers are provided in printed form. In an AIS experiment, the researcher has to prepare possible questions in advance to present their answers during the experiment. Similar to this, in Dynamic MouselabWEB the researcher has to de- fine all attributes and alternatives about the task to display them on the screens. Additionally, both meth- ods put the participant in an active position. The main difference between these two approaches is that in an AIS experiment, the participants have to ask questions about what they need to know to solve the problem, but in Dynamic MouselabWEB the partic- ipants see all possible attributes and alternatives at once and choose among them to create the task. In order to test the program, a study was conducted to compare the decision-making behavior of two groups which selected a job from same-size decision tasks pre- sented in Dynamic MouselabWEB versus Mouselab- WEB. The results of the study show that 1) the dy- namic structure of a decision task has an effect on decision-making behavior; 2) the dynamic group pro- cessed more information and spent more time on de- cision processes; and 3) the dynamic group spent less time on the most important attribute. The job choice task is just one example of any number of decision- making contexts, and this new program can be used for many decision-making tasks with larger sample sizes. We encourage researchers to use Dynamic Mouselab- WEB to observe decision-making behavior because in- dividualized IDMs include only those criteria that are important for decision makers and only those options that have the potential to be selected. Thus, Dynamic MouselabWEB creates a decision-making environment with more than one important dimension, an issue that Huber (1980) raised. Naturally, the outcomes of this should be explored by future research. Since it is an open-source program, other researchers can easily ac- cess the codes and adapt them to their research. Electronic Supplementary Material Supplementary material available online at https://doi.org/10.11588/jddm.2019.1.63149 – ESM1.DatabaseTables (doc). – ESM2.CreateDMWTables(sql). – ESM3.ChooseAlterScreen(php). – ESM4.ChooseAttriScreen(php). – ESM5.DecisionScreen(php). – ESM6.MouselabWEB_Files(zip). – ESM7.Data(sav). Acknowledgements: The authors would like to thank the participants and the experts who contribute this study without getting any reward. Declaration of conflicting interests: The authors de- clare that the research was conducted in the absence of any commercial or financial relationships that could be constructed as a potential conflict of interest. Author contributions: Zeliha was the lead researcher who designed the experiment, collected the data, anal- ysed it for this work. Semra was the supervisor who served as a constant guiding light for this work. Handling editor: Andreas Fischer Copyright: This work is licensed under a Creative Com- mons Attribution-NonCommercial-NoDerivatives 4.0 In- ternational License. Citation: Yıldırım, Z. & Taşabat, S. E. (2019). Dy- namic MouselabWEB: Individualized information display matrixes. Journal of Dynamic Decision Making, 5, 4. doi:10.11588/jddm.2019.1.63149. Received: 23 June 2019 Accepted: 7 October 2019 Published: 28 December 2019 References Abelson, R. & Levi, A. (1985). Decision making and decision the- ory. In G. Lindzey & E. Aronson (Eds.), The handbook of social psychology (Vol. 1, pp. 231–309). Munich: Random House. Andersson, P. (2001). P1198: Software for tracing deci- sion behavior in lending to small businesses. Behavior Re- search Methods, Instruments, & Computers, 33(2), 234–242. doi:10.3758/bf03195370 Aschemann-Witzel, J., & Hamm, U. (2011). Measuring con- sumers’ information acquisition and decision behavior with the computer-based information-display-matrix. Methodology. doi:10.1027/1614-2241/a000018 Beach, L. R., & Mitchell, T. R. (1978). A contingency model for the selection of decision strategies. Academy of management review, 3(3), 439–449. doi:10.5465/amr.1978.4305717 Bettman, J. R., Johnson, E. J., Luce, M. F., & Payne, J. W. (1993) Correlation, conflict, and choice. Journal of Experimen- tal Psychology: Learning, Memory, and Cognition, 19, 931–51. doi:10.1037//0278-7393.19.4.931 Bettman, J. R., Luce, M. F., & Payne, J. W. (1998). Constructive consumer choice processes. Journal of Consumer Research, 25, 187–217. doi:10.1086/209535 Böckenholt, U., & Hynan, L. S. (1994). Caveats on a process- tracing measure and a remedy. Journal of Behavioral Decision Making, 7(2), 103–117. doi:10.1002/bdm.3960070203 Brehmer, B. (1994). The psychology of linear judgement mod- els. Acta Psychologica, 87(2-3), 137–154. doi:10.1016/0001- 6918(94)90048-5 Browne, G. J., Pitts, M. G., & Wetherbe, J. C. (2007). Cognitive stopping rules for terminating information search in online tasks. MIS Quarterly, 89–104. doi:10.2307/25148782 Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers. doi:10.4324/9780203771587 10.11588/jddm.2019.1.63149 JDDM | 2019 | Volume 5 | Article 4 | 8 https://doi.org/10.11588/jddm.2019.1.63149 https://doi.org/10.11588/jddm.2019.1.63149 https://doi.org/10.3758/bf03195370 https://doi.org/10.1027/1614-2241/a000018 https://doi.org/10.5465/amr.1978.4305717 https://doi.org/10.1037//0278-7393.19.4.931 https://doi.org/10.1086/209535 https://doi.org/10.1002/bdm.3960070203 https://doi.org/10.1016/0001-6918(94)90048-5 https://doi.org/10.1016/0001-6918(94)90048-5 https://doi.org/10.2307/25148782 https://doi.org/10.4324/9780203771587 https://doi.org/10.11588/jddm.2019.1.63149 Yıldırım & Taşabat: Dynamic MouselabWEB Costa-Gomes, M., Crawford, V. P., & Broseta, B. (2001). Cog- nition and Behavior in Normal-Form Games: An Experimental Study. Econometrica, 69(5), 1193–1235. doi:10.1111/1468- 0262.00239 Dawes, R. M. (1979). The robust beauty of improper linear mod- els in decision making. American Psychologist, 34(7), 571–582. doi:10.1037//0003-066x.34.7.571 Einhorn, H. J., Kleinmuntz, D. N., & Kleinmuntz, B. (1979). Linear regression and process-tracing models of judgment. Psychological Review, 86(5), 465–485. doi:10.1037//0033- 295x.86.5.465 Einhorn, H. J., & Hogarth, R. M. (1981). Behavioral decision the- ory: Processes of judgment and choice. Annual Review of Psy- chology, 32, 53–88. doi:10.1146/annurev.ps.32.020181.000413 Ford, J. K., Schmitt, N., Schechtman, S. L., Hults, B. M., & Doherty, M. L. (1989). Process tracing methods: Contribu- tions, problems, and neglected research questions. Organiza- tional Behavior and Human Decision Processes, 43(1), 75–117. doi:10.1016/0749-5978(89)90059-9 Harte, J. M., & Koele, P. (2001). Modelling and describing human judgement processes: The multiattribute evaluation case. Think- ing & Reasoning, 7(1), 29–49. doi:10.1080/13546780042000028 Hogarth, R. M. (1975). Cognitive processes and the as- sessment of subjective probability distributions. Journal of the American Statistical Association, 70(350), 271–289. doi:10.1080/01621459.1975.10479858 Huber, O. (1980). The influence of some task variables on cog- nitive operations in an information-processing decision model. Acta Psychologica, 45(1–3), 187–196. doi:10.1016/0001- 6918(80)90031-1 Huber, O., Wider, R., & Huber, O. W. (1997). Active infor- mation search and complete information presentation in natu- ralistic risky decision tasks. Acta Psychologica, 95(1), 15–29. doi:10.1016/s0001-6918(96)00028-5 Huneke, M. E., Cole, C. & Levin, I. P. (2004). How Varying Lev- els of Knowledge and Motivation Affect Search and Confidence during Consideration and Choice. Marketing Letters, 15(2–3), 67–79. doi:10.1023/b:mark.0000047385.01483.19 Jacoby, J., Jaccard, J., Kuss, A., Troutman, T., & Mazursky, D. (1987). New directions in behavioral process research: Implica- tions for social psychology. Journal of Experimental Social Psy- chology, 23(2), 146–175. doi:10.1016/0022-1031(87)90029-1 Jacoby, J., Chestnut, R. W., Weigl, K. C., & Fisher, W. (1976). Pre-purchase information acquisition: Description of a process methodology, research paradigm, and pilot investigation. ACR North American Advances,3, 306-314. Jarvenpaa, S. L. (1989). The Effect of Task Demands and Graphical Format on Information Processing Strate- gies. Management Science, INFORMS, 35(3), 285–303. doi:10.1287/mnsc.35.3.285 Jasper, J. D., & Shapiro, J. (2002). MouseTrace: A bet- ter mousetrap for catching decision processes. Behavior Re- search Methods, Instruments, & Computers, 34(3), 364–374. doi:10.3758/bf03195464 Jasper, J. D., & Levin, I. P. (2001). Validating a new process tracing method for decision making. Behavior Re- search Methods, Instruments, & Computers, 33(4), 496–512. doi:10.3758/bf03195408 Klayman, J. (1982). Simulations of six decision strategies: Com- parisons of search patterns, processing characteristics, and re- sponse to task complexity. Center for Decision Research, Gradu- ate School of Business, University of Chicago. Levin, I. P., & Jasper, J. D. (1995). Phased narrowing: A new process tracing method for decision making. Organiza- tional Behavior and Human Decision Processes, 64(1), 1–8. doi:10.1006/obhd.1995.1084 Mittal, B. (1989). Measuring purchase-decision in- volvement. Psychology & Marketing, 6(2), 147–162. doi:10.1002/mar.4220060206 Mitchell, A. A. (1979). Involvement: a potentially important me- diator of consumer behavior. ACR North American Advances. Montgomery, A. L., Hosanagar, K., Krishnan, R., & Clay, K. B. (2004). Designing a better shopbot. Management Science, 50(2), 189–206. doi:10.1287/mnsc.1030.015 Payne, J. W. (1976). Task Complexity and Contingent Processing in Decision Making: An Information Search and Protocol Anal- ysis. Organizational Behavior and Human Performance, 16(2), 366–387. doi:10.1016/0030-5073(76)90022-2 Payne, J. W., Bettman, J. R., & Johnson, E. J. (1988). Adaptive strategy selection in decision making. Journal of Experimental Psychology: Learning, Memory, and Cognition, 14(3), 534–552. doi:10.1037//0278-7393.14.3.534 Payne, J. W., Braunstein, M. L., & Carroll, J. S. (1978). Ex- ploring predecisional behavior: An alternative approach to deci- sion research. Organizational Behavior and Human Performance, 22(1), 17–44. doi:doi.org/10.1016/0030-5073(78)90003-x Payne, J. W., Bettman, J. R., & Johnson, E. J. (1992). Behavioral decision research: A constructive processing per- spective. Annual review of psychology, 43(1), 87–131. doi:10.1146/annurev.ps.43.020192.000511 Pollay, R. W. (1970). The structure of executive decisions and decision times. Administrative Science Quarterly, 15(4) 459–471. doi:10.2307/2391339 Russo, J. E., & Dosher, B. A. (1983). Strategies for multiattribute binary choice. Journal of Experimental Psychology: Learn- ing, Memory, and Cognition, 9(4), 676. doi:10.1037//0278- 7393.9.4.676 Stevenson, M. K., Busemeyer, J. R., & Naylor, J. C. (1990). Judg- ment and decision-making theory. In M. D. Dunnette & L. M. Hough (Eds.), Handbook of industrial and organizational psychol- ogy (pp. 283–374). Palo Alto, CA, US: Consulting Psychologists Press. Svenson, O. (1979). Process descriptions of decision making. Organizational Behavior and Human Performance, 23, 86-112. doi:10.1016/0030-5073(79)90048-5 Tabatabai, M. (1998). Investigation of decision making process: A hypermedia approach. Interacting with Computers, 9(4), 385– 396. doi:10.1016/s0953-5438(97)00009-x Westenberg, M. R. M., & Koele, P. (1994). Multi-attribute evalua- tion processes: Methodological and conceptual issues. Acta Psy- chologica, 87(2-3), 65–84. doi:10.1016/0001-6918(94)90044-2 Willemsen, M. C., & Johnson, E. J. (2004). Mouselabweb: Per- forming sophisticated process tracing experiments in the partic- ipants home. In Society for Computers in Psychology annual meeting, Minneapolis. Willemsen, M. C., & Johnson, E. J. (2011). Visiting the decision factory: Observing cognition with MouselabWEB and other in- formation acquisition methods. In M. Schulte-Mecklenbeck, A. Kuehberger, J.G. Johnson (Eds.), A handbook of process tracing methods for decision research, pp. 21-42. Abingdon: Routledge Williamson, J., Ranyard, R., & Cuthbert, L. (2000). Risk man- agement in everyday insurance decisions: Evidence from a pro- cess tracing study. Risk, Decision and Policy, 5(1), 19–38. doi:10.1017/s1357530900000090 10.11588/jddm.2019.1.63149 JDDM | 2019 | Volume 5 | Article 4 | 9 https://doi.org/10.1111/1468-0262.00239 https://doi.org/10.1111/1468-0262.00239 https://doi.org/10.1037//0003-066x.34.7.571 https://doi.org/10.1037//0033-295x.86.5.465 https://doi.org/10.1037//0033-295x.86.5.465 https://doi.org/10.1146/annurev.ps.32.020181.000413 https://doi.org/10.1016/0749-5978(89)90059-9 https://doi.org/10.1080/13546780042000028 https://doi.org/10.1080/01621459.1975.10479858 https://doi.org/10.1016/0001-6918(80)90031-1 https://doi.org/10.1016/0001-6918(80)90031-1 https://doi.org/10.1016/s0001-6918(96)00028-5 https://doi.org/10.1023/b:mark.0000047385.01483.19 https://doi.org/10.1016/0022-1031(87)90029-1 https://doi.org/10.1287/mnsc.35.3.285 https://doi.org/10.3758/bf03195464 https://doi.org/10.3758/bf03195408 https://doi.org/10.1006/obhd.1995.1084 https://doi.org/10.1002/mar.4220060206 http://dx.doi.org/10.1287/mnsc.1030.0151 https://doi.org/10.1016/0030-5073(76)90022-2 https://doi.org/10.1037//0278-7393.14.3.534 https://doi.org/10.1016/0030-5073(78)90003-x https://doi.org/10.1146/annurev.ps.43.020192.000511 https://doi.org/10.2307/2391339 https://doi.org/10.1037//0278-7393.9.4.676 https://doi.org/10.1037//0278-7393.9.4.676 https://doi.org/10.1016/0030-5073(79)90048-5 https://doi.org/10.1016/s0953-5438(97)00009-x https://doi.org/10.1016/0001-6918(94)90044-2 https://doi.org/10.1017/s1357530900000090 https://doi.org/10.11588/jddm.2019.1.63149