USF Libraries
USF Digital Collections

An assessment of paired similarities and card sorting

MISSING IMAGE

Material Information

Title:
An assessment of paired similarities and card sorting
Physical Description:
Book
Language:
English
Creator:
Dwyer, Theodore James
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla.
Publication Date:

Subjects

Subjects / Keywords:
method comparison
disimilarities matrix
similarities collection
individual difference scaling
multidimensional scaling
Dissertations, Academic -- Psychology -- Masters -- USF   ( lcsh )
Genre:
government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: Alcohol expectancies have been shown to be predictive of risk for alcohol problems. Experimental research studies have challenged participants' expectancies with the end result demonstrating a mediational effect on participant drinking. Cognitive research using priming and word recognition tasks have led to the theory that expectancies operate in an associative network. Using dissimilarities information this network has been mapped using multidimensional scaling. The current techniques for collecting dissimilarities information directly in alcohol expectancy research has been limited to the use of the paired comparisons tasks. In order to investigate the utility of a different similarities task a comparison was made between a card sorting task and paired comparisons. The overall comparisons of matrices and Individual Difference Scaling (INDSCAL; Carroll & Chang, 1970) results followed the expected trends and generally supported the hypotheses that the two methods would provide essentially the same information. However, a possible method effect for gender was observed. The method effect was seen when comparing across methods within the females dichotomized by drinker category. Further studies are necessary to replicate these findings and to attempt to identify which method has the effect.
Thesis:
Thesis (M.A.)--University of South Florida, 2003.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Theodore James Dwyer.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 64 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001441485
oclc - 53994032
notis - AJM5925
usfldc doi - E14-SFE0000158
usfldc handle - e14.158
System ID:
SFS0024854:00001


This item is only available as the following downloads:


Full Text

PAGE 1

An Assessment of Paired Similarities and Card Sorting by Theodore James Dwyer A thesis submitted in partial fulfillment of the requirements for the degree of Master of Arts Department of Psychology C ollege of Arts and Sciences University of South Florida Major Professor: Mark Goldman Ph.D. Vicky Phares, Ph.D. Kristen Salomon, Ph.D. Date of Approval: November 12, 2003 Keywords: multidimensional scaling, individual difference scaling, similarities collection, disimilarities matrix method comparison Copyright 2003 Theodore J. Dwyer

PAGE 2

i Table of Contents List of Tables iv List of Figures v Abstract vi Introduction 1 Expectancy Research 1 Expectancies in Alcohol 1 Correlational Findings 2 Changes in Expectancies 2 Longitudinal Findings 3 Experimental Evidence 3 Cognitive Explanation of the Expectancy Process 4 Cognitiv e Evidence of Alcohol Expectancies 4 Visual Analog 5 Techniques for Collecting Similarity Data 5 Comparison of MDS Collection Techniques 7 Rationale for the Study 8 Hypotheses 8 Method 10

PAGE 3

ii Participants 10 Study Design 10 Instruments 11 Paired Comparison Task 11 Card Sorting Task 12 Distracter Task 13 Demographics and Alcoho l Use Questionnaire 14 Procedure 14 Results 15 Description of Sample 15 Analysis for Order Effect 16 Classification of Drinker Types 17 O verall Method Comparison 17 Individual Difference Scaling 18 IND SCAL Comparisons 19 Comp arison of Groups 22 Comparison of Drinker Types in Relation to Dimensions 22 Analysis for Further Dimensions 25 Discussion 26 Differences Across Methods Within Gender 27 Summary and Conclusion 31 References 33

PAGE 4

iii Appendices 39 Appendix A: Paired Comparison Task 4 0 Appendix B: Card Sorting Task 47 Appendix C: Addition Task 48 Appendix D: Demographics and Alcohol Use Questionnaire 57

PAGE 5

iv List of Tables Table 1 Participant Ethnicity and Total Group N 15 Ta ble 2 Age and Drinking by Gender for Low Drinkers (LD) and High Drinkers (HD) 16 Table 3 INDSCAL Variance Accounted for and Stress of Solution 19 Table 4 Dimension Weights 20 Table 5 Angle Between Groups Within Each M ethod 21

PAGE 6

v List of Figures Figure 1 Male Dimension Weights 23 Figure 2 Female Dimension Weights 24 Figure 3 Dimension Weights Drinker type by Method 24 Figure 4 Overall Card Sorting Solution 28 Figure 5 Overall Paired Comparison Solution 28 Figure 6 Gender Dimension Weights 30

PAGE 7

vi An Assessment of Paired Similarities and Card Sorting Theodore James Dwyer ABSTRACT Alcohol Expectancies have been shown to be pred ictive of risk for alcohol problems. Experimental research studies have challenged participants expectancies with the end result demonstrating a mediational effect on participant drinking. Cognitive research using priming and word recognition tasks hav e led to the theory that expectancies operate in an associative network. Using dissimilarities information this network has been mapped using multidimensional scaling. The current techniques for collecting dissimilarities information directly in alcohol expectancy research has been limited to the use of the paired comparisons tasks. In order to investigate the utility of a different similarities task a comparison was made between a card sorting task and paired comparisons. The overall comparisons of matr ices and Individual Difference Scaling (INDSCAL; Carroll & Chang, 1970) results followed the expected trends and generally supported the hypotheses that the two methods would provide essentially the same information. However, a possible method effect for g ender was observed. The method effect was seen when comparing across methods within the females dichotomized by drinker category. Further studies are necessary to replicate these findings and to attempt to identify which method has the effect.

PAGE 8

1 An Asses sment of Paired Comparison and Card Sorting in Expectancy Research. Introduction Alcohol expectancies are cognitive constructs or beliefs about the rewarding qualities of alcohol consumption. Alcohol expectancies have been demonstrated to be linked to al cohol consumption. Additionally, expectancies have been manipulated, using true experimental designs, to show that they mediate drinking levels. In order to better understand the expectancy process, cognitive mapping procedures have been used for modeling the structure and visualizing the expectancy network. This study will explore an alternative method for collecting information used in cognitive mapping of alcohol expectancies. Expectancy Research The course of expectancy research begins in the 1930s with Tolman (1932) who postulated the existence of a cognitive variable that predicts behavioral outcome. Several other researchers followed Tolmans line of thought for inclusion of the cognitive variable expectancies within the more traditional stimulu s response conceptualization of behavior (MacCorquodale & Meehl 1953; Rotter, 1954; Bolles, 1972). Expectancies in Alcohol In the 1960s, Merry (1966) challenged the loss of control theory by administering both alcohol and placebo to recovering alcoh olics. These findings,

PAGE 9

2 together with other researchers findings in balanced placebo studies (Engle & Williams, 1972; Marlatt, Demming, & Reid, 1973), demonstrated the need for an explanation of consumption patterns and responses to drinking alcohol (or p lacebo) that did not strictly conform with the pharmacological effects of the drink consumed. Tolmans theory of expectancies was found not only applicable to alcohol consumption but also provided an excellent explanation for reported effects that did no t correspond to the pharmacological effects of alcohol. Correlational Findings Brown and colleagues demonstrated that adults expected a variety of positive activities as a result of alcohol consumption and that these expectancies were related to their dr inking patterns (Brown, Goldman, Inn, & Anderson, 1980). These findings were confirmed by other researchers (Southwick, Steel, Marlatt, & Lindell 1981; Rosenow, 1983), and were extended by Christiansen, Goldman, and Inn (1982) who demonstrated evidence of expectancies in children before their first drinking experience. These findings in children were replicated and extended into children as young as six years old (Miller, Smith, & Goldman, 1990; Dunn and Goldman, 1996). Changes in Expectancies In addition to the previous research which established the existence of expectancies in children, research has also shown that these expectancies could change with age (or change over the course of the lifespan, childhood, etc). Expectancies were found to be primari ly negative (e.g. rude, dizzy) in the youngest cohort of children, with a consistent shift towards more positive expectancies (e.g. outgoing, less nervous) as

PAGE 10

3 children approached adolescence and presumably their first direct drinking experience (Miller, Sm ith, & Goldman 1990). These findings were replicated by Dunn and Goldman (1996). Longitudinal Findings After demonstrating that expectancies were present before the first drinking experience, further research into the temporal relationship was conducted u sing longitudinal designs. Christiansen and colleagues (1989) reported that expectancies for positive outcomes predicted prospectively the onset of drinking, and expectancies for improved cognitive and motor performance predicted problem drinking (Christi ansen et. al. 1989). Others found a reflective relationship between expectancies and drinking, such that positive expectancies increased as drinking increased, and positive expectancies decreased as drinking decreased (Christianson, Smith, Roehling, & Gold man 1989; Sher, Wood, Wood, & Raskin 1996; Smith, Goldman, Greenbaum, & Christianson, 1995). Experimental Evidence A further essential step in establishing that expectancies have causal status was to demonstrate the mediational link between the levels of expectancies and actual drinking behavior, using true experimental designs. Decreases in drinking were found in several studies, which experimentally manipulated expectancies, by challenging participants expectancies (Females Massey and Goldman, 1988; M ales Darkes and Goldman 1993). Increases in drinking were found when expectancies were experimentally manipulated using cognitive priming (Roehrich and Goldman, 1995, Stein Goldman and Del Boca 1997). These studies provided evidence that expectancies ca n be

PAGE 11

4 experimentally manipulated in order to produce a specific effect on drinking levels. Therefore these studies show experimentally that expectancies mediated drinking levels. Cognitive Explanation of Expectancy Process Tolmans conceptualization of expe ctancies as a cognitive construct provides an excellent segue to using cognitive psychology to explain how alcohol expectancies work. Within cognitive psychology there are several different explanations for cognitive processes; one of these models has bee n described as semantic networks, consisting of interconnected concepts or nodes (Collins and Loftus, 1975). The activation of one of the nodes in this semantic network provides a partial activation to connected concepts within the network (Collins and Lo ftus, 1975). Thus, expectancies can be explained using the concept of spreading activation within a semantic network, where activating one portion of the network causes activation of the related network through the links that bind them together. Cognitive Evidence of Alcohol Expectancies Concurrent with the investigation of expectancies and drinking, cognitive tasks investigating alcohol expectancies have provided evidence that is consistent with Collins and Loftuss semantic networks and spreading activat ion models (Roehrich and Goldman 1995; Stein, Goldman and Del Boca 2000; Kramer and Goldman 2003; Rather and Goldman 1994; Dunn and Goldman 1996). For example, this relationship has been shown in studies of cognitive priming, which utilized both the word s tem completion and Stroop tasks in an alcohol context or with alcohol expectancy words (Kramer and Goldman, 2003). Furthermore, Roehrich and Goldman (1995) demonstrated that alcohol

PAGE 12

5 expectancy prime words produced more drinking than non alcohol primes, an d Stein and Goldman (1996) showed that alcohol related cues also produced more drinking than non alcohol cues. Visual Analog In an attempt to model the overall relationship between expectancy concepts and to provide a visualization of the expectancy netw ork itself, Rather and Goldman (1994) used Multidimensional scaling (MDS) techniques. MDS procedures generate models that are sometimes referred to as cognitive maps, or semantic networks (Collins & Loftus 1975; Collins & Quillian 1969). Rather and Goldma n (1994) used sixteen alcohol expectancy words, to create120 paired comparisons. The resulting co occurrence matrix resulted in an MDS solution, which provides a visual analog to the cognitive space of alcohol expectancies. The MDS solution arrived at by Rather and colleagues consisted of two dimensions. These dimensions have been characterized as valence and arousal. Similar MDS solutions for expectancies have been found by other researchers (Dunn & Goldman, 1996). Techniques for Collecting Similarity Data The paired comparison task provides the co occurrence, or similarity data needed for an MDS solution, utilizing every possible permutation of sets of two stimuli from a list of stimuli. Paired comparison is analogous to a similarity judgment between each possible two stimulus combination. Paired comparisons allow for judgments based upon a participants decision concerning the relationship between each pair of stimuli. The collection effort, therefore, remains unaffected by the experimenters preconc eptions

PAGE 13

6 about the structure of the content (Rosenberg, 1982). The result of a paired comparison task is a matrix providing similarity information that indicates how the participant perceives the relationships among all the stimuli. Some of the difficulti es associated with paired comparison tasks are that they can take large amounts of time to administer, high levels of concentration, and considerable participant effort. As the number of stimuli increases, the number of comparisons increases at a rate of n *(n 1)/2, where n is the number of stimuli being used. This means that with 16 words, there are 120 comparisons; with 30 words there are 435. Also, it has been suggested that there may be dimensions that paired comparisons do not capture (Drasgow, 1976, as cited in Rosenberg, 1982). Overall the paired comparison method provides a useful technique for the collection of similarity information (Torgerson, 1958). Another common method of collecting similarity data is card sorting (Rosenberg, 1982). Card so rting and paired comparisons are similar in many ways. Like paired comparisons, card sorting allows the collection effort to remain unaffected by the experimenters preconceptions about the structure of the content (Rosenberg, 1982) Furthermore, the two methods are similar in that card sorting allows for judgments based upon a participants decision concerning the overall relationship of one concept to all others to be entered into a data matrix. However, unlike paired comparisons, one advantage of card s orting is that participants can make decisions about the entire set of stimuli at the same time. This simultaneous decision element eliminates the multiple pair wise individual comparisons that are inherent in paired comparisons, reducing the amount of ti me required to compare large numbers of stimuli. However, sorting tasks

PAGE 14

7 have not been used to study alcohol expectancies. They have, however, been used in various other domains including perceived attractiveness (Ashmore, Solomon, & Longo, 1996), educati onal planning (Maiden & Hare, 1998; Streveler, Miller & Boyd, 2001), and perceived personality traits (Rosenberg & Olshan, 1970; Davidson, 1972). Comparison of MDS Collection Techniques Previous research using personality terms and kinship terms compared the sorting method with other co occurrence data methods. For example, Rosenberg and Olshan (1970) compared co occurrence methods and demonstrated a high correlation between sorting and comparisons using 60 trait adjectives. An examination of sorting and paired comparisons of personality data by Van der Kloot and Van Herk (1991) also demonstrated high correlations between the methods. However, Drasgows attempt to predict the multidimensional structure of paired comparison data using sorting data (Rosenbe rg, 1982) was not as conclusive. Interestingly, Drasgow demonstrated that the MDS of the data from the sorting method not only captured similar relationships as found in paired comparison, but also may have captured dimensions that may not have been obtai ned using the paired comparison method (Rosenberg, 1982). Because studies have discovered a high correlation between paired comparison data and card sorting data (Rosenberg & Olshan, 1970; Van der Kloot & Van Herk, 1991), it is likely that the resulting s imilarities matrices from the two methods would be similar for expectancy data. However, Drasgows finding suggests that the card sorting task may yield additional dimensions not found using the paired comparison method. Therefore it may be

PAGE 15

8 important to c ompare the two methods to determine if they provided the same information with regard to alcohol expectancy data. Rationale for the Study Currently, the method used for directly collecting similarity matrix data of individuals expectancies for alcohol is the paired comparison method. Although card sorting is another method which has been demonstrated to be useful when collecting similarities matrix data, it has not been used within alcohol expectancy research or expectancy research in general. There ha ve been no attempts to determine if the matrix resulting from card sorting is similar to that found using the paired comparison task in expectancy research. In light of the implication of Drasgows finding that card sorting may provide access to dimension s that paired comparisons may miss, it is important to compare and contrast the methods to investigate if card sorting is useful for expectancy research. If the two methods provide similar data, card sorting could facilitate future expectancy studies by pr oviding a quicker method of collecting essentially the same data. This study will collect both paired comparison information and sorting data information, it will then compare, and contrast the methods using the resulting data matrix. Hypotheses Given th e findings of researchers comparing non expectancy data (Rosenberg and Olshan 1970; van der Kloot and van Herk 1991) using correlations between sorting results and paired comparisons, it was hypothesized that the data matrices for expectancy data will demo nstrate convergence across methods. In other words, the correlation

PAGE 16

9 between paired comparison dissimilarity matrix for expectancy data and the card sorting dissimilarity matrix for expectancy data would be high. Because previous research has shown that expectancies are causally related to drinking, patterns should be observable when mapping the individual differences for separate drinking groups. It is therefore further hypothesized that drinking groups will provide disparate results from each other in relation to the expectancy network when examined on both paired comparison and card sorting tasks. Specifically, there will be an observable difference within gender between the heavier drinkers when compared to the lighter drinkers using Individual Diffe rence Scaling (INDSCAL; Carroll & Chang, 1970), which will be consistent across collection methods.

PAGE 17

10 Method Participants Participants in this study were 85 undergraduate students from the College of Arts and Sciences. A majority of the students were re cruited from the psychology participant pool; however, seven were recruited from an Interdisciplinary Social Sciences statistics class. Three participants were removed based on failure to meet the inclusion criteria (two for age above 28 and one who repor ted not drinking). Analyses were conducted on the remaining 82 participants (36 males and 46 females). Participants mean age was 21.2 (SD=2.43) with a range of 18 to 27. The ethnic make up of the study participants reflected the published statistics fr om the University of South Florida (01 02 school year). Participants identified themselves as Caucasian (59.8%) African American (19.5%), Hispanic/Latino/Latina (9.8%), Asian/Asian American (4.9%), and Other (6.1%). In order to ensure that all participa nts were drinkers the participant pool selection program was used to only recruit participants who report drinking alcohol. Only one participant reported not drinking after being selected using the initial criteria. The non drinking participant was exclud ed from analysis based on this criterion. Study Design Each participant was randomly assigned to complete either the paired comparison measure or a card sorting measure first. A correlation was performed between the 16 words from the paired comparison and the same 16 words from the card sorting task.

PAGE 18

11 Dissimilarity matrices where used to compute the correlations providing 120 ratings for each to demonstrate that the two methods provided the same type of information. Individual Difference Scaling (INDSCAL; Carroll & Chang, 1970) was used to produce solutions which were compared to those previously reported in the literature; after visually inspecting the dimensions to ensure they were oriented the same across solutions. The direction of differences between drinker types were compared to those reported in the literature (Rather & Goldman, 1994; Dunn & Goldman, 1996, 1998; Dunn & Yniguez, 1999; Cruz & Dunn, 2003). Convergence between the methods was investigated by comparing the each INDSCAL solutions patter n of differences across dimensions. This was accomplished by comparing the direction of deviation toward the derived dimensions for each solution and comparing the pattern across solutions. Comparisons were conducted within subjects therefore steps were t aken to control for order. To control for order effects, participants were randomly assigned to one of two orders of administration (the paired comparison and card sort task were counterbalanced); and a conceptually different distracter task (i.e., math pr oblems) was included between each of the measures (Nelson & Goodmon, 2003). After completing all tasks, each participant also completed a demographic form which provided information on the quantity and frequency of drinking. Instruments Paired Comparison Task Materials, including the instructions, rating scale, and word pairs used for the paired comparison task are shown in Appendix A. The paired comparison was carried out using

PAGE 19

12 the same techniques and 16 expectancy words used by Rather and Goldman (199 4). The paired comparison task is a paper and pencil task which consists of 120 comparisons on eight pages with the following instructions: In this experiment you will be presented with adjectives that describe some typical effects that people sometimes experience when they have been drinking alcohol and under the influence of alcohol. These adjectives will be presented in pairs for each pair of alcohol effects. Consider how likely or unlikely it is that you would feel or experience the two effects at t he same time. (Very likely = 1, Likely = 2, Slightly Likely = 3, Equally Likely = 4, Slightly Unlikely = 5, Unlikely = 6, Very Unlikely = 7). Card Sorting Task The words that were used for the card sorting task are listed in Appendix B. The card sorting task included a set of 32 words, 16 from the original paired comparison (Rather & Goldman 1994) and an additional 16 extracted from the large set of terms from which the original alcohol expectancy words were selected. The 32 words selected were shown sim ultaneously in front of the participant on 3 x 5index cards. The participants received the following instructions for the card sorting task: These are adjectives that describe some typical effects that people sometimes experience when they have been drinking alcohol and are under the influence of alcohol. Each adjective is on one of these cards. Please sort these words into piles of effects that you would feel or experience together when drinking. Make as many or as few piles as you want, please try to make no more than 10 piles but you can if you

PAGE 20

13 want. Please look at all of the cards before you start sorting. Let me know when you are done. Do you understand what I have asked you to do? Participants were allowed to make any changes until they were completely satisfied with their groupings. Once they had sorted the cards, participants were asked to provide a name/label for each of the groups. After naming/labeling the groups, they were asked to rate each group on the dimensions of valence and arousa l. Participants then identified their groupings of stimuli as positive, neutral or negative in valence; and high, neutral or low in arousal. Stimuli selection for the card sorting task was conducted using occurrence data from first word associate data. Si xteen words were selected from the remaining (116) set of words from which the 16 included in the paired comparison task were selected. The frequency of occurrence information was from first word associate data collected in our lab for several larger stud ies. Words were identified by selecting the expectancy words with the highest number of occurrences in the first word associate data from each quadrant of the MDS solution found in previous research (Goldman, 1999),until an additional sixteen words have be en selected. Distracter Task The math problems that were used for the distracter task are shown in Appendix C. The distracter task consisted of sheets of three by two digit addition problems. Each participant was given a packet of addition problems, and told This is the next task, complete as many problems as quickly and as accurately as you can. The participants

PAGE 21

14 performed this task for 10 minutes in order to ensure sufficient attention switching from the expectancy stimuli (Nelson & Goodmon, 2002; Ne lson, personal communication). Demographics and Alcohol Use Questionnaire The Demographics and Alcohol Use Questionnaire that was used is shown in Appendix D. The demographics questionnaire consisted of basic demographic questions (i.e. age, gender, and e thnicity) with additional questions on the quantity and frequency of alcohol consumption. Procedure Participants were given an informed consent to read and sign. They then, based on random assignment, performed either the paired comparison or card sortin g task first, as described above. Upon completion of their first assigned task, participants then completed the addition problems. After working on the distracter task for ten minutes the participants completed the remaining similarity task. After comple tion of both the card sorting and the paired comparison tasks, they were given the demographics and drinking questionnaire. They were debriefed and thanked for their participation. Participants were awarded experimental points in accordance with the psyc hology departments participant pool policy. All informed consent forms were kept separate from responses to all other questionnaires in order to maintain confidentiality, ensuring that participants responses could not be associated with their identity.

PAGE 22

15 Results Description of Sample Analyses for overall correlations were conducted on the entire sample of 82 participants. Analyses for differences in drinker type were conducted both across the entire samples and for each gender. As can be seen in Table 1 the ethnic makeup of the subgroups by gender was consistent with the overall group. Table 1 Participant Ethnicity and Total Group N All Participants Females Males N (% of total) 82 (100%) 46 (56.1%) 36 (43.9%) High drinker n (%) 36 (45.1%) 20 (43.5%) 17 (47.2%) Ethnicity Caucasian 49(59.8%) 26 (56.5%) 23 (63.9%) African American 16(19.5%) 10 (21.7%) 6 (16.7%) Hispanic/Latino(a) 8(9.8%) 4 (8.7%) 4 (11.1%) Asian/Asian American 4(4.9%) 2 (4.3%) 2 (5.6%) Other 5(6.1%) 4 (8.7%) 1 (2.8%) The mean age of all participants was 21.16 years (2.43) with a range of 18 to 27 The mean age of females was 21.2 years (SD = 2.32) with a range of 18 to 27 years and the mean age of males was 21.1 years (SD=2.61) with a range of 18 to 27 years. Overall

PAGE 23

16 Participants reported drinking an average of 3.9 (SD=2.09) standard drinks per occasion. Females reported drinking 3.7 (SD=1.94) standard drinks per occasion and males reported drinking 4.2 (SD=2.26) standard drinks per occasion (with a range of 1 to 9 s tandard drinks for all groups). The average number of drinks for each of the drinking groups within gender can be found in Table 2. Table 2 Age and Drinking by gender for Low Drinkers (LD)and High Drinkers (HD) Drinking Age N M SD range M SD range Females All 46 3.7 1.94 1 to 9 21.2 2.32 18 to 27 HD 20 5.4 1.52 4 to 9 20.9 2.25 18 to 27 LD 26 2.3 .79 1 to 3 21.5 2.37 18 to 27 Males All 36 4.2 2.26 1 to 9 21.1 2.61 18 to 27 HD 17 6.1 1.94 2 to 9 20.7 2.39 1 8 to 24 LD 19 2.6 .77 1 to 4 21.47 2.80 18 to 27 Analyses for Order Effect Analyses were first conducted to determine if there had been an effect from the order of administration of the collection methods. These analyses were completed by taking the co rrelation between each of the data collection techniques first and second collection points. The correlation was r(82) = .936 (p < .01) for the card sorting and r(82)

PAGE 24

17 = .970 (p < .01) for the paired comparison. Based on the correlations between each of th e different collection methods it appears that, due to random assignment to order and the Distracter task, there was no effect for order. Therefore order of administration was not considered in subsequent analyses. Classification of Drinker Types Partici pants reported quantity and frequency of alcohol consumption was used to estimate the total number of standard drinks consumed per month. Using this estimate participants who drank 40 or more drinks per month were placed into the high drinker category (t his figure was chosen because it best dichotomized the gender categories). Further, participants who fell below this number but who, by current standards, would be classified as binge drinkers (Wechsler & Toben, 2001) based on their reported quantity of dr inking (4 drinks per occasion for women and 5 per occasion for men) were also placed in the high drinker category. Overall Method Comparison Analyses of the overall matrices of card sorting and paired comparisons were conducted. This was accomplished by t aking the ratings of likelihood (higher values indicating lower likelihood of co occurrence) for every possible combination of words in the paired comparison task and correlating it with the non occurrence (dissimilarity) data for the same combination of w ords from the sorting data. Across all participants, data for the card sort was significantly correlated with data from paired comparison, r (82) = .733 (p < .01). This result does not fully support the hypothesis that the two methods would

PAGE 25

18 be highly cor related. It does however provide a moderately high correlation that indicates a large degree of overlap between the two methods. Individual Difference Scaling Individual Difference Scaling (INDSCAL) is a method of analysis used for comparing groups in MD S by analyzing multiple matrices produced by different sub samples in relation to each other. The first step in the analysis is to derive a solution for the separate matrices in the same space. The resulting multidimensional space then serves as the solu tion against which each groups matrix is compared. In order for the comparison to be made a solution is generated for each of the matrices. Table 3 provides the amount of variance accounted for by each solution for the matrices, as well as the stress rat ing, which is a measure of fit used to demonstrate optimum dimensionality. The percentage of the variance (R squared) is a measure of the variance which is accounted for by the distances found in the matrix. The two dimensional solutions reported are consi dered optimal based on dimensional selection techniques for MDS solutions using large changes in stress to identify dimensionality (Spence and Graef, 1974; Davison 1983, 1992; Borg and Groenen 1997). The amount of variance accounted for by the two dimensi onal solution together with the stress rating of the solution is listed in Table 3.

PAGE 26

19 Table 3 INDSCAL Variance Accounted for and Stress of Solution. High Drinkers Low Drinkers Overall solution 3 rd Dimension % % % change in Groups variance stress variance stress variance stress stress Sorting Males 790 .198 .814 .191 .802 .191 .05 Females .804 .200 .871 .164 .838 .183 .07 All .804 .200 .871 .164 .838 .183 .07 Paired Comparison Males .758 .204 .800 .184 .779 .195 .06 Females .753 .208 .745 .210 .749 .209 .07 All .799 .188 .800 .185 .800 .186 .06 INDSCAL Comparisons The comparison of each matrix with the derived stimulus configuration provides a subject weight on each of the dimensions found in th e stimulus space. These subject weights provide a measure of the importance of each dimension for each group when compared to the overall solution and can be used for further comparisons within the configuration space. Subject weights may not be used for a direct comparison across configuration spaces as each configuration space is unique to the solution for the specific groups included in the analysis. However, subject weights may be used to discuss overall patterns based on identification of the dimens ions of the solution as they compare to other

PAGE 27

20 solutions. In the case of expectancies, the dimensional names from solutions found by previous research (Rather & Goldman, 1994; Dunn & Goldman, 1996, 1998, 2000; Dunn & Yniguez, 1999; Cruz & Dunn, 2003) were used. A comparison across measures can also be discussed using the angle between the subject weights on each of the dimensions from the origin. This provides information about the two groups in relation to each of the derived dimensions. Thus a comparis on of groups across methods should be discussed in terms of the angle of separation between groups for the solution and the group weights in relation to each dimension. Table 4 lists the dimension weights of each of the groups and Table 5 lists the angle between groups within each method. Table 4 Dimension Weights Dimension 1 Dimension 2 Method and Group weight weight Card Sorting all high drinkers .763 .505 Card Sorting all low drinkers .759 .567 Paired comparison all high drink ers .725 .522 Paired comparison all Low drinkers .635 .630 Card Sorting females high drinking .686 .578 Card Sorting females low drinking .778 .516 Paired comparison females high drinking .684 .534 Paired comparison females low drinking .568 .650 Card sorting males high drinking .789 .410 Continued on the next page

PAGE 28

21 Table 4 (continued) Dimension 1 Dimension 2 Method and Group weight weight Card sorting males low drinking .752 .49 9 Paired comparison males high drinking .716 .496 Paired comparison males low drinking .680 .581 Cards sorting male .704 .585 Cards sorting female .695 .600 Paired comparison male .719 .532 Paired comparison female .729 .520 Table 5 Angle B etwe en G roups W ithin E ach M ethod Method by Group angle within group Card sorting by type of drinker 3.30 Paired comparison by type of drinker 8.99 Card sorting female by type of drinker 6.53 Paired comparison female by type of drinker 10.83 Card sorti ng male by type of drinker 6.09 Paired comparison male by type of drinker 5.78 Card sorting by gender 2.44 Paired comparison by gender 1.03

PAGE 29

22 Comparison of Groups A direct comparison across INDSCAL solutions using dimension weights is not possible. T he solution and dimensional weights are specific to each INDSCAL solution. However, the solutions still provide a frame of reference from which general trends can be observed. Using visual inspection of previous MDS and INDSCAL solutions (Rather & Goldma n, 1994; Dunn & Goldman, 1996, 1998; Dunn & Yniguez, 1999; Cruz & Dunn, 2003), dimensions for the present solutions were labeled for ease of reference. To maintain consistency across solutions the dimensions were oriented in the same direction. The dimen sions observed in the INDSCAL solutions for both card sorting and paired comparisons where consistent with those identified by earlier research. Each dimension was labeled either Arousal Sedation or Positive Negative based on the dimension that it matched in previous research. The direction of declination from the dimensions is consistent across comparisons. High drinkers consistently deviate from low drinkers toward the same dimension (see figures 1 thru 3). This declination was consistent with what has been observed in other research (Rather & Goldman, 1994). Therefore the differences support the hypothesis that drinker type would demonstrate a consistent pattern across methods. Comparison of D rinker T ypes in R elation to D imensions The declination fr om each of the dimensions was examined to determine if the directions of the type of drinkers were consistent with those findings of previous literature (Rather and Goldman, 1992; Dunn and Goldman 1996, 1998). The dimension weights published for different drinker types (Rather and Goldman, 1994) were used as a

PAGE 30

23 comparison between levels of drinking and the derived dimensions. The declination from the arousal sedation dimension across previous research was the least for heavy drinkers while the declination from the positive/negative dimension was the least for the lighter drinkers. The same pattern can be seen within both methods with the heavier drinkers deviating less from the arousal dimension and lighter drinkers deviating less from the valence dimensio n. The declination from the dimensions is consistent across comparisons. Therefore these differences further support the hypothesis that drinker types would demonstrate consistent patterns across methods. Figure 1. Male Dimension Weights

PAGE 31

24 Figure 2 Female Dimension Weights Figure 3. Dimension Weights Drinker T ype by Method

PAGE 32

25 Analysis for a F urther D imension As noted earlier, Drasgow suggested that a different dimensiona l solution might be found using card sorting as compared to those found using paired comparisons. The most commonly used method of determining dimensionality is to search for an elbow in the stress data when the amount of change in stress by number of di mensions levels off (Spence and Graef, 1974; Davison 1983, Borg and Groenen 1997). As can be seen in Table 3, a solution for a third dimension for card sorting does not change the stress any more than the change observed for paired comparisons. Thus it ap pears that for alcohol expectancies the card sorting method does not capture a different dimensional solution.

PAGE 33

26 Discussion The results of the comparisons largely support the hypothesis that the two collection methods provide similar information. The fir st hypothesis, that the two methods would be highly correlated, was not fully supported, although a moderately high correlation was found between the card sorting and paired comparison matrices. Although the correlation was not as high as those found in r esearch using Personality items r=.960 and strategies for getting ones way r=.806 (Van der Kloot & Van Herk, 1991), the observed correlation was sizable. The differences seen in the correlations may be an indicator of the effect within females and across drinker types discussed later. The implication for this difference is that although each of the methods provide the same type of data; paired comparisons (with an angle of deviation by drinker type of 8.99) may be better at identifying a real difference b etween subtypes of drinkers within gender. The second hypothesis, that the differences between drinker types by gender would be consistent across methods, was supported by the observed separation between drinker types. This finding was consistent with the pattern observed in previous research (Rather & Goldman 1994). The second hypothesis was further reinforced by the observed declination from each solutions dimensions which were consistent with the deviations from the dimensions seen in previous research (Rather & Goldman, 1994; Dunn & Goldman, 1996, 1998; Dunn & Yniguez, 1999; Cruz & Dunn, 2003).

PAGE 34

27 Differences Across Methods Within Gender An observed difference within the results which was not expected (discussed below) raises interesting questions abou t differences by gender and drinker type for each of the methods. A visual comparison of the MDS solutions for card sorting (figure 4) and paired comparison (figure 5) demonstrated quadrants that were very similar when either of the solutions is rotated 1 80 degrees. Since there was a relatively large degree of overlap between the methods, this observed consistency was expected. Considering that the same individual participants provided information for each of the methods in the analysis, and an order eff ect was ruled out based on the observed high correlation between the different collection order positions for similar methods, the amount of difference between drinkers should have been consistent across the methods for each gender or type of drinker. Inte restingly this was not the case for all of the categories. When a comparison was made using the males across method of collection (Figure 1) the difference in the angle was only .31 degrees while a comparison of the angles for gender (Figure 6) was a diff erence of only 1.41 degrees. This angular consistency was not seen when considering the angle of separation seen for the female drinker types (a 4.3 degree difference; Figure 2) and for the overall drinker types (a 5.69 degree difference; Figure 3). The s mall differences seen in the male and gender comparisons were what should be expected if there was no effect for method. If the differences were derived from females alone the differences should have been seen in the angular differences of the gender comp arison. If the differences stemmed from the method of collection alone there should have been a similar difference across all of the

PAGE 35

28 Figure 4 Overall Card Sorting Solution Figure 5 Overall Paired Comparison Solution

PAGE 36

29 angular comparisons. The differences observed in the drinker type data were mirrored in the female data and almost nonexistent in the male and gender data. As identified earlier, the comparisons that provide these angular differences w ere being made based on data collected from the same individuals. Therefore, the angular differences across females and overall drinker types seem to imply a method effect seen in females when dichotomized by type of drinker. Several limitations of the in vestigation should be considered before utilizing the findings in future research. First and foremost, the sample should be considered. The pool of participants was restricted to those individuals within College of Arts and Science classes whose instruct ors provide extra credit for participation in experiments. Thus the results may not be generalizable beyond the Psychology participant pool. All of the participants were college students; therefore the results may not generalize beyond a college sample. Further, the sample was limited in age (18 27). Therefore the results may not generalize to different ages (these methods have not been used in expectancies for other age ranges). Second, temporal limitations for collection of alcohol consumption data shou ld be considered. The data were collected at one time point and over a constrained three week period. The three week period immediately preceded the Mid term examinations for most of the undergraduate classes which has been shown to be a decreased period of consumption for college students (Del Boca, Darkes, Greenbaum &Goldman 2003). Collection of drinking data at one time point is not sufficient for an identification of multiple drinker subgroups (which at any one time point may have similar drinking p atterns) such as those seen in larger longitudinal studies (Schulenberg,

PAGE 37

30 OMalley, Bachman, Wadsworth & Johnston, 1996; Del Boca, Darkes, Greenbaum & Goldman, in press). Thus the generalizability may be limited by the lack of longitudinal Figure 6 Gender Dimension Weights type data to classify drinker types. Finally, the study was designed to determine if there was a difference between the methods of data collection it was not designed to identify how the participants were co nceptualizing the tasks. Participants were asked if they understood the card sorting task, there was no record kept of those who asked for further clarification (further clarification consisted of reading the portion of the instructions that they did not understand to them a second time). There was also no question asked as to whether or not the participants understood the paired comparison task (although care was

PAGE 38

31 made to clarify the rating scale). Therefore if participants were conceptualizing the tasks differently there is no way to tell from this study. Categorization of drinkers was accomplished by selecting those who met the criteria for binge drinking (Wechsler & Toben, 2001) and for those who drank an estimated 40 or more drinks per month. This fig ure was arrived at because it dichotomized the sample and targeted participants who would be drinking high levels throughout the month. This number may not seem like a high level of drinking because it would result in an average of 1.33 standard drinks a day, if the 40 drinks are spread across a hypothetical month (30 days). However, by considering the drinking trends demonstrated by Del Boca and colleagues (in press) this level is much higher than it appears. Del Boca and colleagues (in press) observed t hat college student drinking is planned based on demands throughout the month and was shown to mostly occur over the weekends. This changes the drinking of 40 drinks a month from 1.3 drinks a day to close to binging levels (4.4 per day assuming 3 weekends a month). Summary and Conclusions The overall solutions and individual differences follow the expected trends and support the hypotheses. However, subtle differences within the methods might have been indicative of a different method of conceptualizing or a different approach to the separate tasks for one of the subtypes of drinkers within females. It is clear that additional research aimed at gaining a better understanding of the observed phenomena is necessary. It will be necessary to replicate these findings and to try and identify if there is a different approach to conceptualizing the task for drinker subtypes.

PAGE 39

32 Should a decision need to be made on which of the two methods should be used to collect similarities data in expectancies, consideration of the possible differences observed within drinker type for the method of data collection should be included in this decision. Because the past literature in alcohol expectancies has used paired comparisons as a direct method of comparison it would be prud ent to identify if there is a method effect related to card sorting before instituting this method and attempting to compare it with past research.

PAGE 40

33 References Ashmore, R. Solomon, M & Longo, L (1996) T hinking about fashion models looks: A multi dimensional approach to the structure of perceived attractiveness. Personality and Social Psychology bulletin 22 1083 1104 Bauman, K. E. & Chenoweth, R .L. (1984) The relationship between the consequences adolescents expect from smoking and their beha vior: A factor analysis with panel data. Journal of Applied Social Psychology 14 28 41. Bauman, K. E., Fisher, L A., Bryan, E S. & Chenoweth, R L. (1985) Relationship between subjective expected utility and behavior: A longitudinal study of adolescent d rinking behavior. Journal of Studies on Alcohol 46 32 38. Bolles, R. C. (1972). Reinforcement, expectancy, and learning. Psychological Review, 79 394 409. Borg I. & G roenen P. (1997) Modern Multidimensional Scaling. Springer series in statistics S pringer Verlag Brown, S A. (1993) Drug effect expectancies and addictive behavior change. Experimental & Clinical Psychopharmacology 1, 55 67. Brown, S.A., Goldman, M. S., Inn, A & Anderson, L R. (1980) Expectations of reinforcement from alcohol: Th eir domain and relation to drinking patterns. Journal of Consulting & Clinical Psychology 48 419 426 Carey, K B. (1995) Alcohol related expectancies predict quantity and frequency of heavy drinking among college students. Psychology of Addictive Beha viors 9, 236 241. Carroll, J.D. & Chang, J. J. (1970) Analysis of Individual Differences in Multidimensional scaling via an N way generalization of Eckart Young Decomposition, Psychometrika 35 283 319 Christiansen, B A., Smith, G T., Roehling, P V. & Goldman, M S. (1989) Using alcohol expectancies to predict adolescent drinking behavior after one year. Journal of Consulting & Clinical Psychology 57 93 99.

PAGE 41

34 Christiansen, B.A., Goldman, M.S., & Inn, A. (1982). Development of alcohol related expecta ncies in adolescents: Separating pharmacological from social learning influences. Journal of Consulting and Clinical Psychology 50 336 344. Collins, A M., & Loftus, E F. (1975) A spreading activation theory of semantic processing. Psychological Review 82 407 428. Collins, A .M., & Quillian, M. R (1969) Retrieval time from semantic memory. Journal of Verbal Learning & Verbal Behavior 8 240 247. Cruz I. & Dunn M. (2003) Lowering the risk for early alcohol use by challenging alcohol expectancies in elementary school children. Journal of Consulting and Clinical Psychology 71 493 503 Darkes, J & Goldman, M S. (1993) Expectancy challenge and drinking reduction: Experimental evidence for a mediational process. Journal of Consulting & Clinical Psyc hology 61 344 353. Davidson M L (1972) An empirical comparison of card sorting and paired comparisons judgments as methods of gathering data in a multidimensional scaling study: an exploratory study, University of Illinois 1972 Davison M L (1992) Multidimensional Scaling Malibar, FL: Kreiger. Del Boca, F K., Darkes, J Goldman, M S. & Smith, G T. (2002) Advancing the expectancy concept via the interplay between theory and research. Alcoholism: Clinical & Experimental Research 26 926 935. Del Boca F.K., Darkes J. Greenbaum P. & Goldman M.S. (in press) Up close and personal: Temporal variability in the drinking of individual college students during their first year Dun n, M. E. & Goldman, M. (1996) Empirical modeling of an alcohol expecta ncy memory network in elementary school children as a function of grade Experimental & Clinical Psychopharmacology 4 209 217.

PAGE 42

35 Dunn M. & Yniguez, I. (1999) An experimental demonstration of the influence of alcohol advertising on the organization of alc ohol expectancies in memory among 4 th and 5 th grade children. Experimental and Clinical Psychopharmacology 8 566 575 Engle, K.B., & Williams, T. K. (1972). Effect of an ounce of vodka on alcoholics desire for alcohol. Quarterly Journal of Studies on Alcohol 33 1099 1105. Goldman, M S. (1999) Expectancy operation: Cognitive neural models and architectures. In I. Kirsch ( Ed .), How expectancies shape experience Washington, DC : APA Books 41 63. Goldman, M S., Brown, S A., Christiansen, B A. & Smit h, G T. (1991) Alcoholism and memory: Broadening the scope of alcohol expectancy research. Psychological Bulletin 110 137 146. Klock H. & Buhmann J (1998) Data visualization by Multidimensional Scaling: a deterministic Annealing Approach Institute f or Informatik III Bonn, Germany : Rheinische Fredrich Willhelms Universitat Kramer D., & Goldman (2003) Using a modified stroop task to implicitly discern the cognitive organization of alcohol expectancies Journal of Abnormal Psychology 112 171 175 Leigh, B C. & Stacy, A W. (1991) On the scope of alcohol expectancy research: Remaining issues of measurement and meaning. Psychological Bulletin 110 147 154. MacCorquodale, K., & Meehl, P.E. (1953). Preliminary suggestions as to a formalization of expectancy theory. Psychological Review 60 55 63. Maiden N. A. & Hare M. (1998) Problem domain categories in requirements engineering International Journal human computer studies 49 281 304 Marlatt, G.A., Demming, B., & Reid, J.B. (1973). Los s of control drinking in alcoholics: An experimental analogue. Journal of Abnormal Psychology 81 233 241.

PAGE 43

36 Marlatt, G.A. & Rohsenow, D J. (1980). Cognitive processes in alcohol use: Expectancy and the balanced placebo design. In N. K. Mello (Ed.), A dvances in substance abuse: Behavioral and biological research 1 159 199. Greenwich, CT: JAI Press Massey, R.F., & Goldman, M S. (1988, August) Manipulating expectancies as a means of altering alcohol consumption. Paper presented at the 96th Annua l Convention of the American Psychological Association, Atlanta, GA. Merry, J. (1966) The "loss of control" myth Lancet 4 1257 1258. Miller, G A. (1991) Lexical echoes of perceptual structure. In G. Lockhead & J. Pomerantz ( Ed s.), The perception of structure: Essays in honor of Wendell R. Garner. Washington, DC : American Psychological Association 249 261 Nelson, D L. & Goodmon, L B., (2003) Disrupting attention: The need for retrieval cues in working memory theories. Memory & Cognition 31 65 76 Newcomb, M D., Chou, C Bentler, P. M. & Huba, G. J. (1988) Cognitive motivations for drug use among adolescents: Longitudinal tests of gender differences and predictors of change in drug use. Journal of Counseling Psychology 35 426 438 Oh, M & Raftery A.E. (2000) Baysian Multidimensional Scaling and choice of Dimension. Technical report no 379 University of Washington : Department of Statistics Rather, B C. & Goldman, M S. (1994) Drinking related differences in the memory organization o f alcohol expectancies. Experimental & Clinical Psychopharmacology 2 167 183 Rather, B C., Goldman, M S., Roehrich, L & Brannick, M (1992) Empirical modeling of an alcohol expectancy memory network using multidimensional scaling. Journal of Abnorma l Psychology 101 174 183 Roehrich, L & Goldman, M S. (1995) Implicit priming of alcohol expectancy memory processes and subsequent drinking behavior. Experimental & Clinical Psychopharmacology 3 402 410 Rosenberg, S (1982) The method of Sorti ng in Multivariate research with applications selected from cognitive psychology and person perception. Multivariate Applications in Social Sciences N. Hirschberg & L. Humphreys (Eds.) Hillsdale NJ : Erlbaum, 117 142

PAGE 44

37 Rosenberg, S. & Kim, M (1975) Th e method of sorting as a data gathering procedure in multivariate research. Multivariate Behavioral Research 10 489 502 Rosenberg, S & Olshan, K (1970) Evaluative and descriptive aspects in personality perception. Journal of Personality & Social Psychology 16 619 626 Rohsenow, D.J. (1983). Drinking habits and expectancies about alcohols effects for self versus others. Journal of Consulting and Clinical Psychology 51 752 756. Rotter, J.B. (1954). Social learning and clinical psychology Englewood Cliffs, N.J.: Prentice Hall, Inc. Schulenberg, OMalley, Bachman, Wadsworth & Johnston, (1996) Getting drunk and growing up: trajectories of frequent binge drinking during the transition to young adulthood. Journal of studies on Alcohol 57 289 304 She r, K J., Wood, M D., Wood, P K. & Raskin, G (1996) Alcohol outcome expectancies and alcohol use: A latent variable cross lagged panel study. Journal of Abnormal Psychology 105 561 574 Smith, G T., Goldman, M S., Greenbaum, P E. & Chri stiansen, B A. (1995) Expectancy for social facilitation from drinking: The divergent paths of high expectancy and low expectancy adolescents Journal of Abnormal Psychology 104 32 40. Spence I & Graef J (1974) The determination of the underlying d imensionality of an empirically obtained Matrix of proximities. Multivariate Behavioral research 9 331 342 Stacy, A W., Newcomb, M D. & Bentler, P M. (1991) Cognitive motivation and drug use: A 9 year longitudinal study. Journal of Abnormal Psycholog y 100 502 515 Stein, K D., Goldman, M S. & Del Boca, F K. (2000) The influence of alcohol expectancy priming and mood manipulation on subsequent alcohol consumption. Journal of Abnormal Psychology 109 106 115 Sussman, Dent, Simon, Galaif, Moss, Graig & Johnson (1995) Immediate impact of social influence oriented substance abuse prevention curricula in traditional and continuation high schools Drugs & Society 8 65 81 Tolman, E. C. (1932). Purposive behavior in animals and men. New York: Cent ury Company. Torgerson, W (1958) Theory and method of scaling New York : W iley

PAGE 45

38 Tulving, E. (1972). Episodic and semantic memory. In E. Tulving & W. Donaldson (Eds.), Organization of memory. 381 403. New York: Academic Press. Tulving, E. (1985). How many memory systems are there? American Psychologist, 40 385 398. Van der Kloot, W A. & Van Herk, H (1992) Multidimensional scaling of sorting data: A comparison of three procedures. Multivariate Behavioral Research 26 563 581 Wechsler, H, L ee, J.E., Kuo, M., Seibring, M., Nelson T.F., & Lee H. (2002) Trends in college binge drinking during periods of increased prevention efforts. Journal of American College Health 50 203 217.

PAGE 46

39 Appendices

PAGE 47

40 Appendix A: Paired Comparison Task Think of the EFFECTS OF ALCOHOL on you. Consider how likely or unlikely it is that you would feel or experience the two effects at the same time (Very likely =1 Likely =2 slightly likely =3 equally likely = 4 slightly unlikely =5 unlikely =6 very unlikely =7). 1. Funny Irresponsible Very Likely 1 2 3 4 5 6 7 Very Unlikely 2. Sick Sad Very Likely 1 2 3 4 5 6 7 Very Unlikely 3. Sleepy Smart Very Likely 1 2 3 4 5 6 7 Very Unlikely 4. Irresponsible Hap py Very Likely 1 2 3 4 5 6 7 Very Unlikely 5. Dizzy Sad Very Likely 1 2 3 4 5 6 7 Very Unlikely 6. Funny Happy Very Likely 1 2 3 4 5 6 7 Very Unlikely 7. Dangerous Obnoxious Very Likely 1 2 3 4 5 6 7 Very Unlikely 8. Intoxicated Irresponsible V ery Likely 1 2 3 4 5 6 7 Very Unlikely 9. Happy Dangerous Very Likely 1 2 3 4 5 6 7 Very Unlikely 10. Relaxed Sad Very Likely 1 2 3 4 5 6 7 Very Unlikely 11. Obnoxious Sick Very Likely 1 2 3 4 5 6 7 Very Unlikely 12. Sad Intoxicated Very Likely 1 2 3 4 5 6 7 Very Unlikely 13. Irresponsible Dangerous Very Likely 1 2 3 4 5 6 7 Very Unlikely 14. Intoxicated Smart Very Likely 1 2 3 4 5 6 7 Very Unlikely 15. Dizzy Obnoxious Very Likely 1 2 3 4 5 6 7 Very Unlikely 16. Irresponsible Talkative Very Likely 1 2 3 4 5 6 7 Very Unlikely 17. Funny Smart Very Likely 1 2 3 4 5 6 7 Very Unlikely 18. Stupid Funny Very Likely 1 2 3 4 5 6 7 Very Unlikely 19. Horny Stupid Very Likely 1 2 3 4 5 6 7 Very Unlikely

PAGE 48

41 Appendix A (Continued) Think of th e EFFECTS OF ALCOHOL on you. Consider how likely or unlikely it is that you would feel or experience the two effects at the same time (Very likely =1 Likely =2 slightly likely =3 equally likely = 4 slightly unlikely =5 unlikely =6 very unlikely =7). 20. Obnoxious Irresponsible Very Likely 1 2 3 4 5 6 7 Very Unlikely 21. Dangerous Sad Very Likely 1 2 3 4 5 6 7 Very Unlikely 22. Intoxicated Sick Very Likely 1 2 3 4 5 6 7 Very Unlikely 23. Sleepy Funny Very Likely 1 2 3 4 5 6 7 Very Unlikely 24. Relaxed Happy Very Likely 1 2 3 4 5 6 7 Very Unlikely 25. Horny Sad Very Likely 1 2 3 4 5 6 7 Very Unlikely 26. Sick Horny Very Likely 1 2 3 4 5 6 7 Very Unlikely 27. Smart Happy Very Likely 1 2 3 4 5 6 7 Very Unlikely 28. Funny Relaxed Ver y Likely 1 2 3 4 5 6 7 Very Unlikely 29. Intoxicated Talkative Very Likely 1 2 3 4 5 6 7 Very Unlikely 30. Irresponsible Horny Very Likely 1 2 3 4 5 6 7 Very Unlikely 31. Horny Sleepy Very Likely 1 2 3 4 5 6 7 Very Unlikely 32. Happy Sad Very Li kely 1 2 3 4 5 6 7 Very Unlikely 33. Relaxed Horny Very Likely 1 2 3 4 5 6 7 Very Unlikely 34. Obnoxious Funny Very Likely 1 2 3 4 5 6 7 Very Unlikely 35. Dangerous Horny Very Likely 1 2 3 4 5 6 7 Very Unlikely 36. Irresponsible Sick Very Likel y 1 2 3 4 5 6 7 Very Unlikely 37. Dizzy Stupid Very Likely 1 2 3 4 5 6 7 Very Unlikely 38. Happy Sleepy Very Likely 1 2 3 4 5 6 7 Very Unlikely

PAGE 49

42 Appendix A (Continued) Think of the EFFECTS OF ALCOHOL on you. Consider how likely or unlikely it is th at you would feel or experience the two effects at the same time (Very likely =1 Likely =2 slightly likely =3 equally likely = 4 slightly unlikely =5 unlikely =6 very unlikely =7). 39. Talkative Dizzy Very Likely 1 2 3 4 5 6 7 Very Unlikely 40. Funny Intoxicated Very Likely 1 2 3 4 5 6 7 Very Unlikely 41. Obnoxious Relaxed Very Likely 1 2 3 4 5 6 7 Very Unlikely 42. Smart Horny Very Likely 1 2 3 4 5 6 7 Very Unlikely 43. Irresponsible Relaxed Very Likely 1 2 3 4 5 6 7 Very Unlikely 44. Obnox ious Smart Very Likely 1 2 3 4 5 6 7 Very Unlikely 45. Sleepy Irresponsible Very Likely 1 2 3 4 5 6 7 Very Unlikely 46. Stupid Obnoxious Very Likely 1 2 3 4 5 6 7 Very Unlikely 47. Confident Intoxicated Very Likely 1 2 3 4 5 6 7 Very Unlikely 48. Relaxed Sleepy Very Likely 1 2 3 4 5 6 7 Very Unlikely 49. Happy Obnoxious Very Likely 1 2 3 4 5 6 7 Very Unlikely 50. Smart Relaxed Very Likely 1 2 3 4 5 6 7 Very Unlikely 51. Sleepy Obnoxious Very Likely 1 2 3 4 5 6 7 Very Unlikely 52. Talka tive Relaxed Very Likely 1 2 3 4 5 6 7 Very Unlikely 53. Stupid Sick Very Likely 1 2 3 4 5 6 7 Very Unlikely 54. Talkative Happy Very Likely 1 2 3 4 5 6 7 Very Unlikely 55. Sad Stupid Very Likely 1 2 3 4 5 6 7 Very Unlikely 56. Sleepy Sick Very Likely 1 2 3 4 5 6 7 Very Unlikely 57. Dangerous Dizzy Very Likely 1 2 3 4 5 6 7 Very Unlikely

PAGE 50

43 Appendix A (Continued) Think of the EFFECTS OF ALCOHOL on you. Consider how likely or unlikely it is that you would feel or experience the two effect s at the same time (Very likely =1 Likely =2 slightly likely =3 equally likely = 4 slightly unlikely =5 unlikely =6 very unlikely =7). 58. Stupid Sleepy Very Likely 1 2 3 4 5 6 7 Very Unlikely 59. Talkative Smart Very Likely 1 2 3 4 5 6 7 Very Unli kely 60. Confident Dangerous Very Likely 1 2 3 4 5 6 7 Very Unlikely 61. Intoxicated Dizzy Very Likely 1 2 3 4 5 6 7 Very Unlikely 62. Stupid Irresponsible Very Likely 1 2 3 4 5 6 7 Very Unlikely 63. Obnoxious Talkative Very Likely 1 2 3 4 5 6 7 V ery Unlikely 64. Smart Dizzy Very Likely 1 2 3 4 5 6 7 Very Unlikely 65. Relaxed Stupid Very Likely 1 2 3 4 5 6 7 Very Unlikely 66. Funny Confident Very Likely 1 2 3 4 5 6 7 Very Unlikely 67. Sick Dizzy Very Likely 1 2 3 4 5 6 7 Very Unlikely 68. Confident Smart Very Likely 1 2 3 4 5 6 7 Very Unlikely 69. Dizzy Sleepy Very Likely 1 2 3 4 5 6 7 Very Unlikely 70. Horny Intoxicated Very Likely 1 2 3 4 5 6 7 Very Unlikely 71. Talkative Horny Very Likely 1 2 3 4 5 6 7 Very Unlikely 72. H appy Stupid Very Likely 1 2 3 4 5 6 7 Very Unlikely 73. Confident Sick Very Likely 1 2 3 4 5 6 7 Very Unlikely 74. Dizzy Funny Very Likely 1 2 3 4 5 6 7 Very Unlikely 75. Sick Talkative Very Likely 1 2 3 4 5 6 7 Very Unlikely 76. Smart Dangerou s Very Likely 1 2 3 4 5 6 7 Very Unlikely

PAGE 51

44 Appendix A (Continued) Think of the EFFECTS OF ALCOHOL on you. Consider how likely or unlikely it is that you would feel or experience the two effects at the same time (Very likely =1 Likely =2 slightly likel y =3 equally likely = 4 slightly unlikely =5 unlikely =6 very unlikely =7). 77. Intoxicated Relaxed Very Likely 1 2 3 4 5 6 7 Very Unlikely 78. Smart Stupid Very Likely 1 2 3 4 5 6 7 Very Unlikely 79. Sad Irresponsible Very Likely 1 2 3 4 5 6 7 Ve ry Unlikely 80. Sick Dangerous Very Likely 1 2 3 4 5 6 7 Very Unlikely 81. Talkative Sad Very Likely 1 2 3 4 5 6 7 Very Unlikely 82. Happy Horny Very Likely 1 2 3 4 5 6 7 Very Unlikely 83. Obnoxious Confident Very Likely 1 2 3 4 5 6 7 Very Unlik ely 84. Irresponsible Dizzy Very Likely 1 2 3 4 5 6 7 Very Unlikely 85. Sleepy Intoxicated Very Likely 1 2 3 4 5 6 7 Very Unlikely 86. Dangerous Sleepy Very Likely 1 2 3 4 5 6 7 Very Unlikely 87. Confident Relaxed Very Likely 1 2 3 4 5 6 7 Very Unlikely 88. Irresponsible Smart Very Likely 1 2 3 4 5 6 7 Very Unlikely 89. Smart Sad Very Likely 1 2 3 4 5 6 7 Very Unlikely 90. Confident Happy Very Likely 1 2 3 4 5 6 7 Very Unlikely 91. Horny Funny Very Likely 1 2 3 4 5 6 7 Very Unlikely 92. Intoxicated Dangerous Very Likely 1 2 3 4 5 6 7 Very Unlikely 93. Confident Irresponsible Very Likely 1 2 3 4 5 6 7 Very Unlikely 94. Sad Obnoxious Very Likely 1 2 3 4 5 6 7 Very Unlikely 95. Funny Talkative Very Likely 1 2 3 4 5 6 7 Very Un likely

PAGE 52

45 Appendix A (Continued) Think of the EFFECTS OF ALCOHOL on you. Consider how likely or unlikely it is that you would feel or experience the two effects at the same time (Very likely =1 Likely =2 slightly likely =3 equally likely = 4 slightly unlik ely =5 unlikely =6 very unlikely =7). 96. Sad Confident Very Likely 1 2 3 4 5 6 7 Very Unlikely 97. Talkative Dangerous Very Likely 1 2 3 4 5 6 7 Very Unlikely 98. Sleepy Confident Very Likely 1 2 3 4 5 6 7 Very Unlikely 99. Intoxicated Happy V ery Likely 1 2 3 4 5 6 7 Very Unlikely 100. Sick Relaxed Very Likely 1 2 3 4 5 6 7 Very Unlikely 101. Funny Sick Very Likely 1 2 3 4 5 6 7 Very Unlikely 102. Horny Obnoxious Very Likely 1 2 3 4 5 6 7 Very Unlikely 103. Relaxed Dangerous Very L ikely 1 2 3 4 5 6 7 Very Unlikely 104. Dizzy Horny Very Likely 1 2 3 4 5 6 7 Very Unlikely 105. Obnoxious Intoxicated Very Likely 1 2 3 4 5 6 7 Very Unlikely 106. Sick Happy Very Likely 1 2 3 4 5 6 7 Very Unlikely 107. Stupid Intoxicated Very L ikely 1 2 3 4 5 6 7 Very Unlikely 108. Sad Sleepy Very Likely 1 2 3 4 5 6 7 Very Unlikely 109. Sick Smart Very Likely 1 2 3 4 5 6 7 Very Unlikely 110. Sleepy Talkative Very Likely 1 2 3 4 5 6 7 Very Unlikely 111. Horny Confident Very Likely 1 2 3 4 5 6 7 Very Unlikely 112. Talkative Stupid Very Likely 1 2 3 4 5 6 7 Very Unlikely 113. Dizzy Confident Very Likely 1 2 3 4 5 6 7 Very Unlikely 114. Dangerous Funny Very Likely 1 2 3 4 5 6 7 Very Unlikely

PAGE 53

46 Appendix A (Continued) Think of the EFFECTS OF ALCOHOL on you. Consider how likely or unlikely it is that you would feel or experience the two effects at the same time (Very likely =1 Likely =2 slightly likely =3 equally likely = 4 slightly unlikely =5 unlikely =6 very unlikely =7). 115. Stupid Confident Very Likely 1 2 3 4 5 6 7 Very Unlikely 116. Happy Dizzy Very Likely 1 2 3 4 5 6 7 Very Unlikely 117. Confident Talkative Very Likely 1 2 3 4 5 6 7 Very Unlikely 118. Sad Funny Very Likely 1 2 3 4 5 6 7 Very Unlikely 119. Dan gerous Stupid Very Likely 1 2 3 4 5 6 7 Very Unlikely 120. Relaxed Dizzy Very Likely 1 2 3 4 5 6 7 Very Unlikely When you have completed this task let the experimenter know.

PAGE 54

47 Appendix B: Card Sorting The Following words will be the 32 stimuli includ ed in the card sort task. The sorting task will be accomplished using 32 3 X 5 index cards. Each index card will have one stimulus on it for the sort. Confident Dangerous Dizzy Emotional Energetic Foolish Forceful Funny Happy Horny Incoherent Intox icated Irresponsible Mean Mellow Nervous Noisy Obnoxious Pass out Relaxed Sad Sick Sleepy Smart Social Stupid Talkative Unbearable Unhappy Unpredictable Verbal Woozy

PAGE 55

48 Appendix C: Addition task 754 + 53 344 + 89 697 + 43 152 + 93 285 + 96 511 + 61 629 + 87 880 + 29 856 + 63 645 + 96 549 + 59 332 + 54 406 + 65 270 + 70 973 + 52 2 43 + 79 562 + 34 499 + 23 157 + 85 807 + 79 920 + 44 808 + 25 423 + 50 456 + 89 435 + 89 865 + 56 345 + 98 456 + 5 6 877 + 56 159 + 65 735 + 45 955 + 65

PAGE 56

49 Appendix C (Continued) 468 + 85 466 + 65 672 + 45 895 + 65 645 + 57 215 + 67 123 + 58 355 + 45 132 + 54 688 + 54 755 + 23 658 + 87 651 + 15 657 + 66 325 + 99 888 + 56 785 + 56 235 + 54 222 + 66 735 + 23 658 + 73 951 + 65 568 + 86 654 + 78 879 + 45 698 + 65 758 + 55 625 + 78

PAGE 57

50 Appendix C (Conti nued) 459 + 32 951 + 54 753 + 68 874 + 65 185 + 48 658 + 98 655 + 54 895 + 59 785 + 65 486 + 56 951 + 54 657 + 95 657 + 65 159 + 78 654 + 56 654 + 86 954 + 56 654 + 85 159 + 77 846 + 91 177 + 54 658 + 56 591 + 95 543 + 95 795 + 62 594 + 65 654 + 82 198 + 56

PAGE 58

51 Appendix C (Continued) 864 + 62 627 + 33 753 + 65 183 + 65 752 + 35 985 + 55 687 + 45 358 + 15 954 + 75 954 + 68 934 + 75 116 + 35 732 + 65 492 + 91 376 + 59 194 + 54 649 + 75 924 + 49 874 + 37 738 + 81 875 + 54 159 + 48 735 + 95 553 + 78 821 + 64 645 + 72 651 + 56 197 + 95

PAGE 59

52 Appendix C (Continued) 254 + 16 354 + 95 951 + 35 235 + 16 876 + 94 357 + 65 116 + 35 754 + 19 654 + 36 564 + 15 759 + 65 435 + 12 156 + 45 245 + 65 323 + 84 321 + 98 651 + 32 346 + 58 613 + 54 987 + 63 516 + 57 684 + 63 516 + 84 352 + 16 546 + 98 435 + 16 546 + 87 316 + 54

PAGE 60

53 Appendix C (Continued) 651 + 35 735 + 16 959 + 57 321 + 89 872 + 46 576 + 46 548 + 98 162 + 37 635 + 35 498 + 43 216 + 58 762 + 16 576 + 98 546 + 51 673 + 21 698 + 76 521 + 65 765 + 21 354 + 26 654 + 69 846 + 51 321 + 65 468 + 46 546 + 54 687 + 51 321 + 95 41 6 + 24 323 + 78

PAGE 61

54 Appendix C (Continued) 984 + 65 616 + 87 654 + 13 216 + 57 324 + 98 791 + 48 745 + 16 156 + 86 519 + 84 321 + 76 513 + 65 651 + 98 762 + 19 876 + 56 765 + 46 876 + 46 878 + 35 746 + 87 984 + 32 169 + 87 354 + 63 513 + 54 654 + 32 546 + 87 325 + 41 687 + 52 432 + 37 687 + 35

PAGE 62

55 Appendix C (Continued) 468 + 79 846 + 35 216 + 57 687 + 35 135 + 44 654 + 68 321 + 68 732 + 16 986 + 79 465 + 46 987 + 35 416 + 35 464 + 35 241 + 68 765 + 43 213 + 57 686 + 21 324 + 68 732 + 13 576 + 87 432 + 16 576 + 35 213 + 57 687 + 43 213 + 57 654 + 16 598 + 43 213 + 73

PAGE 63

56 Appendix C (Continued) 216 + 35 687 + 31 686 + 54 321 + 65 498 + 76 352 + 13 546 + 87 631 + 34 687 + 32 135 + 46 987 + 32 135 + 49 873 + 21 685 + 79 687 + 98 746 + 35 323 + 54 383 + 82 683 + 68 939 + 36 837 + 28 887 + 29 357 + 98 112 + 20 383 + 82 683 + 68 939 + 36 837 + 28

PAGE 64

57 Appendix D: Demographics and Alcohol use q uestionnaire Demographics and Alcohol Use Questionnaire Date of Birth: ____/____/_____ Sex: 0) Female 1) Male Day /Month/Year Ethnicity: 0) Caucasian (non Hispanic) 1) African American 2) Latino/Latina 3) Asian 4) Other Have you ever had an alcoholic dri nk? (0) Yes (1) No About how frequently do you drink alcohol? 0) Never 1) Once a year or less 2) 3 4 times a year 3) Once a month 4) 2 3 times a month 5) 2 3 times a week 6) 4 5 times a week 7) 6 7 times a week On occasions when you drink alcohol, about how many drinks do you t ypically consume? Please estimate the actual number of drinks, where: 1 drink = approximately 1 can of beer, or = 1 glass of wine or wine cooler, = 1 serving of liquor or a mixed drink 0) None 1) One Drink 2) 2 3) 3 4) 4 5) 5 6 6) 7 8 7) 9 or more


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001441485
003 fts
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 031203s2003 flua sbm s000|0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0000158
035
(OCoLC)53994032
9
AJM5925
b SE
SFE0000158
040
FHM
c FHM
090
BF121
1 100
Dwyer, Theodore James.
3 245
An assessment of paired similarities and card sorting
h [electronic resource] /
by Theodore James Dwyer.
260
[Tampa, Fla.] :
University of South Florida,
2003.
502
Thesis (M.A.)--University of South Florida, 2003.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 64 pages.
520
ABSTRACT: Alcohol expectancies have been shown to be predictive of risk for alcohol problems. Experimental research studies have challenged participants' expectancies with the end result demonstrating a mediational effect on participant drinking. Cognitive research using priming and word recognition tasks have led to the theory that expectancies operate in an associative network. Using dissimilarities information this network has been mapped using multidimensional scaling. The current techniques for collecting dissimilarities information directly in alcohol expectancy research has been limited to the use of the paired comparisons tasks. In order to investigate the utility of a different similarities task a comparison was made between a card sorting task and paired comparisons. The overall comparisons of matrices and Individual Difference Scaling (INDSCAL; Carroll & Chang, 1970) results followed the expected trends and generally supported the hypotheses that the two methods would provide essentially the same information. However, a possible method effect for gender was observed. The method effect was seen when comparing across methods within the females dichotomized by drinker category. Further studies are necessary to replicate these findings and to attempt to identify which method has the effect.
590
Adviser: Goldman, Mark
653
method comparison.
disimilarities matrix.
similarities collection.
individual difference scaling.
multidimensional scaling.
0 690
Dissertations, Academic
z USF
x Psychology
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.158