Graduate School of Psychology
R
ESEARCHM
ASTER’
SP
SYCHOLOGYI
NTERNSHIPP
ROPOSAL&
R
EPORTStatus: Internship proposal with results and discussion Date: 02 / 07 / 2014
1. WHO AND WHERE
Student
Name : Florian Wanders
Student ID number : 10620885
Address : Billitonstraat 6-1
Postal code and residence : 1094 BC, Amsterdam
Telephone number : 0648544303
Email address : florian.wanders@student.uva.nl
Supervisor(s)
Within ResMas (obligatory) : Prof. dr Gerben van Kleef
Specialisation : Social Psychology
External supervisor(s), if any : Marc Heerdink
Second assessor :
Research center / location : UvA Number of credits (1 ec = 28 hrs) : 18 At least 18 ec
2. TITLE AND SUMMARY OF THE RESEARCH PROJECT 2.1. Title:
Think Risk Think Leader. The implicit association between risk and leader prototypes. 2.2 Abstract of the report
People perceive individuals who take high risks as more powerful than individuals who do not take such risks. One explanation for this finding is that people implicitly associate risk with a leader prototype and therefore with high levels of power. Two of the three studies reported here support the hypothesis that people implicitly associate risk with a leader prototype (Study 1a, Study 2). Study 1b appeared to suggest the opposite, namely that people associate leader prototypes with safety, rather than with risk. However, this is likely due to an operationalization of the category leader that is inappropriate in this context. Overall, the results suggest that an implicit association between risk and leader might indeed explain why people perceive individuals who take high risks to be powerful.
Word count = 126
2.3 Summary of the original proposal 1 – max 150 words
Individuals who take high risks are perceived as more powerful than individuals who do not take such risks. One explanation for this finding is that people implicitly associate risk taking with a leader prototype and therefore with high levels of power. The proposed study will test this hypothesis in two studies. In Study 1a, an implicit association test (IAT) will provide information on the relative strength of the association between risk and a leader prototype. In Study 1b, a single-attribute IAT (SA-IAT) will provide information about the centrality of risk for a leader prototype. In both cases an IAT effect that is significantly greater than zero will provide support for the implicit association between risk and a leader prototype. Word count = 122
1 The proposal only describes Studies 1a and 1b (referred to as Study 1 and 2 in the original proposal) as
3. PROJECT DESCRIPTION – max 1000 words 3.1. Prior research:
Describe prior research, a comprehensible literature review of the research field, converging upon the research questions.
a) Describe the state of affairs in the current research field, including the theoretical framework, based on the existing body of literature.
The multimillionaire Steve Fossett raced his sled dogs across Alaska, climbed hundreds of mountains, and broke several world records, being most renowned for the first solo flight in an air balloon around the world (BBC, 2008). He was a powerful businessman who took extraordinary risks. The present study investigates this link between power and risk taking, extending previous research (Anderson & Galinsky, 2006; Van Kleef et al., 2014) with an explanation of how risk taking may increase perceptions of power.
Defining Power
Power can be defined as the extent to which one has control over oneself and others, and, at the same time, remains unaffected by social interference (Galinsky, Gruenfeld, & Magee, 2003; Keltner, Gruenfeld, & Anderson, 2003). Recent studies have nuanced the prevailing idea (Resnick, 2013) of power as a corrupting force (Kipnis, 1972) into a more positive concept. According to this new perspective power primarily activates the behavioral activation system, which in turn leads to greater action tendencies, regardless of the social value of the action (Anderson & Berdahl, 2002; Keltner et al., 2003).
Power Increases Risk Taking
The increased activity of the behavioral approach system results in increased attention to rewards and decreased attention to threats (Keltner et al., 2003). Therefore it is not
surprising that power also increases risk taking (Anderson & Galinsky, 2006): When people feel powerful, they primarily see potential rewards, but often ignore associated risks. In other words, powerful people generally have more optimistic risk perceptions than others (Anderson & Galinsky, 2006).
b) Clarify how the previous research eventuates into the research questions of the current proposal
Risk Taking and Implicit Leader Prototypes
Whereas power increases risk taking, risk taking may also increase perceptions of power (Van Kleef et al., 2014). As I will argue below, risk taking should be part of a widely accepted implicit leadership prototype in Western countries. Although leadership and power are separate constructs (Anderson & Berdahl, 2002; Magee & Galinsky, 2008), they are nevertheless correlated (Anderson, John, & Keltner, 2012) and may inform each other, especially because some definitions of implicit leader prototypes explicitly refer to powerful leaders (e.g. House, Javidan, Hanges, & Dorfman, 2002).
Leader categorization theory and leader prototype. Leadership categorization theory (Lord, Foti, & De Vader, 1984; Lord & Maher, 2002) describes how leader prototypes serve as implicit decision criteria according to which targets are unconsciously classified as leader or non-leader: If a person fits the leader prototype, he or she will automatically be categorized as leader; if not, he or she will not be categorized as leader (cf. Rosch, 1978; Shondrick, Dinh, & Lord, 2010). For example, white and male targets are readily
Homan, de Dreu, & van Vugt, 2014; Johnson, Murphy, Zewdie, & Reichard, 2008; Koenig, Eagly, Mitchell, & Ristikari, 2011; Rosette, Leonardelli, & Phillips, 2008). The proposed study builds on this research and examines, in how far the paradigms “think leader think white” (Gündemir et al., 2014; Rosette et al., 2008) and “think leader think male” (Johnson et al., 2008; Koenig et al., 2011) can be expanded to “think leader think risk”.
Risk taking as part of the leader prototype. Three lines of evidence suggest that risk taking might be part of a widely shared implicit leader prototype.
First, a series of studies consistently found that people who take high risks were perceived as more powerful than people who take low risks (Van Kleef et al., 2014).
Second, findings from cross-cultural research into leadership suggest that risk taking might even be a valued aspect of leaders. Based on ratings from 287 middle managers in the Netherlands; an analysis of leadership attributes listed in managerial job postings; an analysis of a series of interviews with CEOs published in 1994 in a major daily Dutch newspaper; and focus interviews with 20 middle managers Thierry, Hartog, Koopman, & Wilderom (2007) concluded that the attribute risk taking, was central to a positive leader prototype.2 Ratings from 272 top managers support these findings (House, Dorfman, Javidan, Hanges, & de Luque, 2014): When asked to rate attributes along a scale from “greatly inhibiting a person from being an outstanding leader” to “contributing greatly to a person being an outstanding leader”, risk taking was among those attributes that were seen as contributing to outstanding leadership. While these findings are specific to the
Netherlands, risk taking is valued even more in the USA. (Chhokar, Brodbeck, & House, 2007; House et al., 1999).
A third line of support for the notion that risk taking is part of a widely held implicit leadership prototype comes from the overrepresentation of personalities with high readiness for risk taking in leadership positions. The proportion of (subclinical)
psychopaths is significantly greater in leadership positions than in the general population (Babiak, Neumann, & Hare, 2010) and numerous studies suggest that higher levels of psychopathy result in greater readiness to take risks (e.g. van Honk, Hermans, Putman, Montagne, & Schutter, 2002). Similarly, Campbell, Hoffman, Campbell, and Marchisio (2010) suggested that narcissists actively seek out leadership positions and are also more inclined to take risks than healthy individuals. Because attributes of leader prototypes are derived from direct or indirect exposure to leaders (cf. Smith & Zárate, 1992), an
overrepresentation of personalities with high readiness for risk taking provides further support for the idea that risk taking is part of an implicit leader prototype. Thus, it appears as if risk taking is indeed an important part of an implicit leader prototype in Western countries.3
2 The implicit leadership prototype portrayed by the GLOBE studies has been criticized as only pertaining to
ideal examples of leadership (e.g. Schyns & Schilling, 2010). Accordingly, attributes that are seen as hindering from outstanding leadership might still be perceived as prototypical of a leader. In line with this argument, Schyns and Schilling (2010) found that participants considered, for example, “unpleasant” as a typical, albeit ineffective, leader attribute. However, “charismatic” was rated most consistently as typical – and effective – leader attribute, supporting the importance of risk taking – as part of charismatic leadership (Javidan & Waldman, 2003) – for a leader prototype (see Hofstede, 2006, and Javidan & House, 2006, for a debate on the validity of the GLOBE measures).
3 Risk taking may play an even greater role for the prototype of an entrepreneurial leader compared to a
3.2. Key questions
Now state the key questions, the essence of the proposal. Here, the intended research should be connected to prior research. Testable hypotheses should be derived from the key question, and the relation between theory and research hypotheses should be clearly specified.
a) Formulate a general relevant research question based on previous research.
The aim of the current studies is to extend previous studies which suggest that risk taking results in inferences of power (Van Kleef et al., 2014) and to investigate the underlying psychological mechanism.
b) Translate the general research question in a clear manner into a specific research question.
Specifically, the current studies investigate whether risk taking is implicitly associated with the prototype of a powerful leader. Such an implicit association could explain participants’ inferences of power from risk taking.
c) Translate the specific research questions into testable research hypotheses.
Two studies will be conducted to investigate the stereotype content of risk taking. Study 1a will test the relative importance of risk taking for a leader prototype. Stronger associations between two mental constructs result in greater cognitive accessibility than weaker
associations (González & Brown, 2006; Greenwald, McGhee, & Schwartz, 1998; Lord et al., 1984). Therefore, if the construct risk taking is indeed part of a leadership prototype, associations between risk taking and leader should be more readily accessible than associations between risk taking and follower (Hypothesis 1a).
Study 1b will test the centrality of risk taking for the leader prototype. The semantic activation of a mental construct should lead to the activation of related constructs (Bargh, Raymond, Pryor, & Strack, 1995; Chen, Lee-Chai, & Bargh, 2001). Therefore, if the construct risk taking is indeed a central component of a leadership prototype it should activate the construct leader. In contrast, the construct leader should not be activated in the absence of risk taking (Hypothesis 1b). The hypotheses can be summarized as follows: Leader is associated with risk, both in relative terms (Hypothesis 1a) and in absolute terms (Hypothesis 1b).
4. PROCEDURE – max 1000 words 4.1. Operationalisation
Describe how the research questions are operationalised.
a) Operationalise the research questions in a clear manner into a research design/strategy.
Study 1a will assess the relative importance of risk taking for a leader prototype with an implicit association test (IAT, Greenwald et al., 1998; Greenwald, Nosek, & Banaji, 2003). Study 1b will assess the centrality of risk taking for a leader prototype with a
single-attribute IAT (SA-IAT, Gündemir, Homan, de Dreu, & van Vugt, 2014, Study 2; Penke, Eichstaedt, & Asendorpf, 2006).
b) Describe the procedures for conducting the research and collecting the data. A pilot study will be administered to peers to obtain stimuli for the IATs in addition to those taken from Gündemir et al. (2014). During the actual study, participants will first complete an unrelated pilot study to assure that participants’ moods are comparable (cf. LaTour & LaTour, 2009). Each participant will then complete the SA-IAT (Study 1b) and the IAT (Study 1a). An unrelated experiment will be administered between the two IATs to preclude reduced reliability from administering multiple IATs (Gawronski, Deutsch, & Banse, 2012).
4.2. Sample characteristics
a) Indicate, given a power analysis, how many participants will be recruited. Also motivate whether the resulting number is feasible.
130 participants will be recruited (see Appendix B for the power analysis) over a span of three weeks. This means that nine participants should be recruited each day.
b) If a subset of participants will be excluded from the analysis given their scores on dependent variables, indicate the objective criterion to do so. For example include a phrase like: “Scores on dependent variables exceeding ± 3 sd of the mean will be excluded from the analysis “
Based on the improved scoring algorithm (Greenwald et al., 2003), data will be excluded from, or modified prior to, the analysis in three cases: First, trials with response latencies above 10,000 ms will be excluded from the analysis. Second, if more than 10% of a participant’s response latencies are below 300 ms that participant’s data will be excluded from the analysis. Third, latencies for incorrect responses will be replaced by the respective block mean for correct responses and a 600 ms error penalty will be added.
c) If a subset of participants will be excluded from the analysis given their scores on a manipulation check item, indicate the objective criterion to do so. For example include a phrase like: “Participants scoring 15 or lower on a manipulation checks item, will be excluded from the analysis”
4.3. Materials
Indicate which tests, stimuli, equipment, etc. will be used; provide sufficiently elaborate descriptions and motivate your choice. (Always report the psychometric characteristics, such as reliability and validity, if existing tests are used. If new or adapted instruments or test materials (e.g., questionnaires) will be developed, then the new
instrument must be independently validated first; only then it can be used as a testing instrument. Exception to this rule is allowed in case of questionnaires that do not contain more than one question (e.g., indicate on a 5-point scale how you feel today).
Pilot study
A paper-and-pencil pilot study will be administered to peers to find five risk-related and five safety-related words that are, respectively, most strongly associated with the categories risk and safety (cf. Nosek, Greenwald, & Banaji, 2005). Participants will be asked about their agreement that each item of a list of words retrieved from an online thesaurus (Appendix C) is central to risk taking or safety (1 = strongly disagree, 7 = strongly agree; cf. Nosek, Greenwald, & Banaji, 2005). Furthermore participants will be asked to rank-order the five items that are most central to the two categories to obtain converging evidence. Study 1a and 1b
Studies 1a and 1b employ, respectively, an IAT and a SA-IAT. Both tests assess the strength of associations between different categories (e.g. risk and leader; Greenwald et al., 1998). In both tests, category labels are displayed on either side of the computer screen, for example as in Figure 1a. Participants are then asked to categorize target words in the middle of the screen as fast and accurate as possible as belonging either to the left or the right side. A strong association between two of the categories (e.g. risk and leader) that are presented on the same side (Figure 1a) results in reduced reaction times and fewer errors compared to when the two strongly associated categories are presented on different sides (Figure 1b). For the screen in Figure 1b, a strong association between leader and risk would require participants to inhibit the tendency to (falsely) classify boss as belonging to the same side as risk (cf. Gawronski et al., 2012). In contrast, a strong association between leader and risk would facilitate performance for the screen in Figure 1a. Tables 1 and 2 provide an overview over the two versions of the IAT, which are described in detail in the remainder of section 4.3 and in section 4.4.
a b
Figure 1a and 1b. Example screens during test trials of an IAT. The
set-up is slightly different during an SAIAT, but the same principle applies also here: performance is facilitated if labels of two strongly associated categories are presented on the same side.
Study 1a. Study 1a will employ the four categories leader, follower, risk and safety, with five items per category.4 Five leader roles (boss, supervisor, leader, executive, and authority) and five follower roles (helper, assistant, subordinate, aid, and follower) will be used as
4 A comparable IAT with four items in each category resulted in an effect size and internal consistency that
was indistinguishable from that of an IAT with 8 items in each category (Nosek et al., 2005). This suggest that an increased number of items would be unlikely to improve the experiment.
items for the categories leader and follower (Gündemir et al., 2014). In addition, five risk- and five safety-related words, as identified in the pilot study, will be used for the categories risk and safety.
The IAT in Study 1a was adapted from Gündemir et al. (2014), using guidelines from Greenwald et al. (1998) and Nosek et al. (2005). First, participants are familiarized with the stimuli (i.e. leader and follower roles and risk- and safety related words) and their respective category. Participants then discriminate between categories with left- (“Q”) and a right-key responses (“P”). First, participants discriminate risk- and safety-related words (Block 1, 20 practice trials) and leader and follower roles (Block 2, 20 practice trials; see Table 1 for an overview). In Blocks 3 (20 test trials) and 4 (40 test trials), participants discriminate between all four categories. Key responses are hereby congruent, such that the categories risk and leader share the same key and the categories safe and follower share the other key. Next, participants again discriminate between risk- and safety-related words, however the key assignments are reversed compared to block one (Block 5, 20 practice trials). Lastly, participants discriminate between all four categories, however, now the categories are presented in an incongruent manner (Blocks 6, 20 test trials, and 7, 40 test trials). Administration of congruent and incongruent blocks, as well as left- and right-key assignment will be counterbalanced.
Block No. of
Trials Function Discrimination Items assigned to left-key response (“Q)
Items assigned to right-key response (“P”)
1 20 Practice Target Risk Safe
2 20 Practice Attribute Leader Follower 3 20 Test Combined Risk + Leader Safe + Follower 4 40 Test
5 40 Practice Target (reversed) Safe Risk
6 20 Test Combined (reversed) Safe + Leader Risk + Follower 7 40 Test
Table 1. Procedure of the implicit association test (IAT).
In a recent comparison of seven implicit measures (Bar-Anan & Nosek, 2013), the IAT had the highest internal reliability (average α =.88), a high test-retest reliability (average r = .45), and correlated moderately with other implicit (average r = .39) and explicit measures (average r = .35). The proposed study will use only one type of stimuli (i.e. semantic stimuli) which is thought to further increase reliability (Gawronski et al., 2012).
Study 1b. The IAT of Study 1a assesses the relative strength of the association between leader and risk taking. That is, the extent to which leader is associated with risk, relative to the extent to which follower is associated with risk. To assess the centrality of risk taking to the leader prototype, or, in other words, the absolute strength of the association between leader and risk taking, a SA-IAT will be used (Penke et al., 2006). Five leader prototype attributes (decisive, intelligent, self-confident, ambitious, and reliable) will be used as items comprising the category leader (Gündemir et al., 2014). As in the IAT, five risk-related words and five safety-related words will serve as items for the categories risk and safety (cf. Gündemir et al., 2014).
The SA-IAT in Study 1b was adapted from Gündemir et al. (2014). First, participants are familiarized with the stimuli (i.e. risk-related words and leader and follower roles) and their respective category. As in the IAT, participants then discriminate between categories with a left-key response (“Q”) and a right-key response (“P”). First, participants discriminate risk- and safety-related words (Block 1, 20 practice trials; see Table 2 for an overview). Next,
participants discriminate between all three categories (Block 2, 35 test trials). Key responses are hereby congruent, such that the categories risk and leader share the same key and the category safe is assigned to the other key. Lastly, participants again discriminate between all three categories, however, now the categories are presented in an incongruent manner (Block 3, 35 test trials). Key distributions for correct assignment will be 2:2:3 for Block 2 and 3:2:2 for Block 3 (Karpinski & Steinman, 2006) to avoid potential bias from equal proportions of correct assignments to left and right key. The order in which participants complete Blocks 2 and 3 and the left- and right key assignment will be counterbalanced (Gündemir et al., 2014).
Block No. of
Trials Function Items assigned to left-key response Items assigned to right-key response
1 20 Practice Risk Safe
2 35 Test Risk + Leader Safe
3 35 Test Risk Safe + Leader
Table 2. Procedure of the single-attribute implicit association test (SA-IAT).
Penke et al. (2006) reported good reliability (α = .82) for the SA-IAT, but small
correlations with explicit measures (r = .21). The related single-target implicit association test (ST-IAT) demonstrated good reliability (average α =.77), a high test-retest reliability (average r = .48), and a moderate correlation with other implicit (average r = .36) and explicit measures (average r = .31). The SA-IAT differs from the ST-IAT only in that the SA-IAT uses a single attribute category (e.g. prototypical leader attributes), while the ST-IAT uses a single target category (cf. Penke et al., 2006).
4.4. Data analysis
Indicate for each research question separately, how it is translated into a statistical prediction. For example: “In a repeated measures ANOVA we expect an interaction effect of the between factor x and the within factor y on the dependent variable z. Also indicate how you will correct for multiple comparisons. Only the analyses proposed here can be described as confirmatory analyses in your research report. All other have to be mentioned as exploratory.
IAT
According to the improved scoring algorithm (Greenwald et al., 2003), an IAT score (D) will be computed for each subject from the test Blocks 3, 4, 6, and 7 (cf. Gündemir et al., 2014). After treatment of extreme values (see 4.2 e), the difference of mean response times for congruent and incongruent blocks will be computed with the two subtractions Block 6 – Block 3 and Block 7 – Block 4. Each difference will be divided by its respective pooled standard deviation that is calculated based on the response times after deletion of extreme scores, but before imputation of incorrect trials. The average of the two subtractions is the individual IAT score, D. Positive values of D hereby suggest stronger associations between leader and risk and follower and safe relative to leader and safe and follower and risk. ST-IAT
D is computed similarly as for the IAT from the subtraction of Block 3 – Block 2 (cf. Gündemir et al., 2014). Positive values of D hereby suggest a stronger association between leader and risk than between leader and safe.
For both IATs a one-sided one-sample t-tests with a significance level of α = 0.05 should be used to test the hypotheses.
4.5. Modifiability of procedure
Is there room for modification of the intended procedure? Evaluation of the proposal by the RMP Thesis Committee is meaningful only if the recommendations that the Committee might have can be implemented. It is therefore required that the intended procedure can be modified before you start gathering data. In situations where procedures or operationalisations or sample characteristics cannot be modified, the Thesis Committee has to be consulted before handing in the research proposal. The committee will consider the eligibility of this project for a research internship.
5. INTENDED RESULTS - max 250 words
Clarify what the implication of possible outcomes would be (per hypothesis) for the specific and general research questions as well as for the theory. Address the following in approximately 250 words:
a) To what extent does the proposed research provide an answer to the research questions?
An IAT effect that is significantly greater than zero will provide support for the hypotheses in both studies. The IAT hereby suggests that risk taking is more strongly related to leader roles than to follower roles and that risk taking is therefore congruent with a leader prototype. The SA-IAT further suggests that such risk taking is a central part of a leader prototype.
Using different operationalizations of leader, the current studies investigated whether it is plausible that inferences of power from risk taking arise because of an implicit association of risk taking with the prototype of a powerful leader. If the hypotheses are supported future studies should establish a causal relationship. That is, future studies will need to demonstrate that people infer power from risk-taking because they rely on the implicit association between risk and leader.
b) What are the (alternative) interpretations if the results do (not) match the expectations?
If the IAT reveals an IAT effect in line with the hypothesis, but the SA-IAT does not, this suggests that risk taking is congruent with a leader prototype, but it is not a central
component. If neither IAT supports the hypotheses, this suggests that risk taking is not part of a leader prototype. In both cases, alternative explanations, such as inferences about a leader’s (high) self-esteem (e.g. Javidan & Waldman, 2003), appear more suited.
c) If applicable: what is the practical and societal relevance
If the hypotheses are supported, Steve Fossett may have been perceived as powerful leader, not only because of his successful career as powerful businessman, but also because of the extraordinary risks he took on his adventures. However, risk taking is always associated with threats. The proposed study therefore promises to offer important insights for companies, regarding the selection and evaluation of their leaders.
Word count = 268
Please refer to Appendix A for sections 6 (work plan), 9 (further steps), and 10 (signatures) of the original proposal. Sections 7 (references) and 8 (appendices) of the original proposal have been updated in this report. The main text continues next with the results and discussion section of the internship report.
6. INTERNSHIP REPORT – Results and Discussion Pilot Study
The pilot study was conducted to obtain target words for the categories risk and safety. Nineteen undergraduate students from a large university in the Netherlands completed the pilot study. Participants’ average ratings and rankings were discussed and the target words presented in Table 3 were selected as most suitable to describe the respective categories of risk and safety. The corresponding English translations are (from top to bottom): risky, hazardous, daring, dangerous, and reckless; sure, safe, careful, thoughtful, and cautious.
Risk Related Words Safety Related Words Target
word rating (1-7) Average % top ranks Target word rating (1-7) Average % top ranks riskant 6.16 84.21 zeker 6.26 68.42 gewaagd 5.79 78.95 veilig 5.95 57.89 gedurfd 5.74 68.42 voorzichtig 5.00 31.58 gevaarlijk 5.22 26.32 bedachtzaam 4.79 21.05 roekeloos 4.90 15.79 behoedezaam 4.74 21.05 Average 5.56 54.73 Average 5.35 40.00 Table 3. Results pilot study. Average ratings to the statement that the target is of great importance to its respective category were anchored from 1 (completely inaccurate) to 7 (completely accurate). ‘% top ranks’ reflects the percentage of times that the target word was listed as one of the five most important words for the respective category.
Studies 1a and 1b
Study 1a assessed the relative strength of the association between leader and risk, whereas Study 1b assessed the absolute strength of this association. A significant positive IAT effect during study 1a hereby suggests that risk taking is more strongly related to leader roles than to follower roles and that risk taking is therefore congruent with a leader prototype. This would support Hypothesis 1a. A significant positive SA-IAT effect during study 1b further suggests that such risk taking is a central part of a leader prototype. This would support Hypothesis 1b.
Participants. Most of the 140 participants were students (N = 133) at a Dutch University (93 women, 47 men, Mage = 22.58, SDage = 4.25).5 All participants completed both studies 1a and 1b and received course credit or a small monetary reward. Four participants were excluded from the analysis for Study 1a (final N = 136): three of them were intoxicated and one was not a native speaker of Dutch. An additional 7 participants were excluded from the analysis for Study 1b (final N = 129): six of them had to be excluded due to a technical error and one left the experiment to accept a phone call.
Analysis and Results – Confirmatory. An IAT score was calculated based on the improved scoring algorithm (Greenwald et al., 2003), which takes into account both reaction times and error rates. The resulting data were normal for Study 1a, yet nonnormal for Study 1b. Greenwald and colleagues (2003) suggest alternatives for calculating the IAT score that are approximately equivalent to the improved scoring algorithm (Table 4). When these approximate equivalent alternatives were incorporated, data were normal in both studies. Therefore, the analyses were conducted based on the improved scoring algorithm
5 A sample size of N = 130 was specified in the proposal. However, because it took some time before the
study was taken off the central site on which participants can register for studies, 10 additional participants had to be recorded.
with approximate equivalent alternatives.6 More positive values of the IAT effect score hereby indicate stronger associations between risk and leader.
Approximate
equivalent alternatives Corresponding steps in the improved scoring algorithm deletion of trials with response
times below 400 ms in addition to the extreme value treatment of the improved scoring algorithm (*)
extreme value treatment of the improved scoring algorithm (*)
computation of an inclusive standard deviation that is based on correct trials only
computation of an inclusive standard deviation that is based on all trials
replacement of incorrect trials with the block mean plus two times the standard deviation of correct trials of the respective block
replacement of incorrect trials with the block mean plus 600 ms
Table 4. Approximate equivalent alternatives and the corresponding steps in the
improved scoring algorithm. (*) please refer to section 4.2 b) for more
information on the extreme value treatment of the improved scoring algorithm
Study 1a investigated the relative association between risk and leader. The IAT score for the IAT from Study 1a was significantly greater than zero (M = 0.37, SD = 0.63, d = 0.60) with t(135) = 6.94, p = 001, two-tailed.78 This suggests that participants associated the combinations 'risk and leader’ and ‘safety and follower’ more strongly than they associated the combinations ‘risk and follower’ and ‘safety and leader.’ Hypothesis 1a, which specified that the relative association between risk and leader would be stronger than that between safety and leader, is therefore supported. In other words: risk is associated with leader in relative terms.
Study 1b investigated the absolute association between risk and leader. The SA-IAT score for the SA-IAT from Study 1b was significantly smaller than zero (M = -0.86, SD = 0.64, d =1.36) with t(128) = -15.40, p < .001, two-tailed.9 This suggests that participants associated leader more strongly with safety than with risk in absolute terms. Hypothesis 1b, which specified that the absolute association between risk and leader would be stronger than that between safety and leader, is therefore not supported. On the contrary, the absolute association between safety and leader was stronger than that between risk and leader. Analyses and Results – Exploratory. Study 1a used leader and follower roles (e.g. boss, secretary) to operationalize the constructs leader and follower. In contrast, Study 1b used prototypical adjectives (e.g. ambitious) to operationalize the construct leader. To test whether the two studies assessed the same construct, a correlation analysis was conducted. The correlation between IAT and SA-IAT scores was positive (r(127) = .28) and
significantly different from zero (p = .001). As stronger associations between risk and leader resulted in a more positive IAT scores in both studies, this suggests that, on average, participants who displayed a stronger relative association between leader and risk during the IAT also displayed a stronger absolute association between leader and risk during the SA-IAT. However, since participants associated leader more strongly with risk during the
6 Results did not differ whether or not approximate equivalent alternatives were incorporated 7 An alpha level of α=.05 was used to assess significance throughout the report.
8 Results did not differ when all participants were included in the analysis. 9 Results did not differ when all participants were included in the analysis.
IAT, but more strongly with safety during the SA-IAT, the correlation is a negative one in conceptual terms: on average, participants who displayed a stronger association between leader and risk during the IAT also displayed a weaker association between leader and safety during the SA-IAT.
Discussion. Even though Hypothesis 1a was supported and participants associated leader more strongly with risk than with safety in relative terms, Hypothesis 1b was not
supported. This suggests that while participants associated risk more readily with leader than with follower, leader was not associated with risk in absolute terms. However, the two studies used different operationalizations of leader. The positive correlation between the IAT effects of both studies suggests that both studies assessed the same construct, yet the small magnitude of the correlation suggests that each study assessed slightly different aspects of this construct. One possible explanation for why Hypothesis 1b was not
supported is that the prototypical leader adjectives from the SAIAT of study 1b were rather positive (e.g. ambitious, intelligent), and that participants associated these positive
characteristics with safety. Therefore a second SA-IAT was conducted that operationalized the construct leader differently.
Study 2
Study 2 was identical to Study 1b except that leader was operationalized with the five leader roles from Study 1a. The hypothesis of Study 2 was identical to that of Study 1b: People associate leader with risk in absolute terms.
Participants and Procedure. All 60 participants were students at a Dutch University (34 women, 26 men, Mage = 21.73, SDage = 2.84) and received course credit or a small monetary reward for participation (please see Appendix D for a power analysis). All participants were Caucasian, as Study 2 was administered as a filler task in an unrelated experiment that required Caucasian participants. Three participants were excluded from the analysis: One of them was excluded because of a technical error, another left the experiment during the SA-IAT, and one was intoxicated.
Analysis and Results. The analysis was identical to that for Study 1b. However, unlike in previous analyses, three additional participant were excluded as more than 10% of their response times were below 300ms (final N = 54; Greenwald et al., 2003). The SA-IAT score for the SA-IAT from Study 2 was positive (M = 0.13, SD = 0.57, d = 0.24) and marginally significant t(53) = 1.74, p = .098, two-tailed. Based on the directional hypothesis of a positive SA-IAT effect, a one-tailed t-test may be used, which suggests that
participants associated leader significantly more strongly with risk than with safety in absolute terms (p = .044, one-tailed). Hypothesis 2, which specified an association between risk and leader in absolute terms, was therefore supported. Figure 2 summarizes the findings of studies 1a through 2.
Figure 2. Mean (SA-)IAT effects and standard errors
throughout the three studies. More positive values represent stronger associations between leader and risk while more negative values represent stronger
associations between leader and safety.
General Discussion
Readiness to take risks is a valued leader characteristic in Western countries (House et al., 2014) and risk-takers are overrepresented among leaders (e.g. Campbell, Hoffman, Campbell, & Marchisio, 2010). Indirect or direct experience of this apparent connection between leader and risk may result in an implicit association between leader and risk (cf. Smith & Zárate, 1992). Such an implicit association between a prototypical (i.e. powerful) leader and risk may explain why people perceive risk takers as powerful (Van Kleef et al., 2014). Three studies investigated the existence of the hypothesized association between leader and risk. Study 1a suggested that leader is indeed associated with risk, relative to the extent to which follower is associated with risk. Yet, Study 1b appeared to suggest that leader is not associated with risk in absolute terms. On the contrary, there was a strong association between leader and safety. However, when the leader roles of Study 1a replaced the positive leader adjectives of Study 1b in Study 2, the hypothesized absolute association between leader and risk was found. This suggests that the results of Study 1b might have been due to an – in this context – inappropriate conceptualization of leader.
Overall, these results suggest that risk-taking is indeed a central aspect of a leader
prototype. The paradigms “think leader think white” (Gündemir et al., 2014; Rosette et al., 2008) and “think leader think male” (Johnson et al., 2008; Koenig et al., 2011) may
therefore be expanded to include “think leader think risk.” Organizations should be aware of this implicit association between risk and leader to facilitate unbiased selection and evaluation of their leaders (cf. Gündemir et al., 2014).
While the current studies provide support for the implicit association between leader and risk and suggest that this association may contribute to an explanation of why people perceive risk takers to be powerful (Van Kleef et al., 2014), future studies will have to establish a causal relationship: That is, future studies will need to demonstrate that people infer power from risk-taking because they rely on the implicit association between risk and leader. Considering the small size of the SA-IAT effect of Study 2, however, it is
questionable whether an implicit association between risk and leader may fully account for the robust finding that people perceive risk takers to be powerful (Van Kleef et al., 2014). One concern is that participants might not have been primed with a powerful leader stereotype. Indeed, studies suggest that risk taking plays a greater role for entrepreneurial leaders, than for leaders in general (Miner & Raju, 2004; Stewart & Roth, 2001, 2004).
Future studies should therefore refine the stimulus material to include more suitable primes. Especially for students the role of a supervisor may seem rather abstract and powerless, while the schoolyard bully or clique leader represents a much more realistic manifestation of power. Despite these considerations, the current set of studies suggests that even students entertain an implicit association between risk and the abstract concept of leader.
7. REFERENCES
List all cited literature, formatted according to the directions of the APA Manual.
Anderson, C., & Berdahl, J. L. (2002). The experience of power: Examining the effects of power on approach and inhibition tendencies. Journal of Personality and Social Psychology, 83(6), 1362– 1377. doi:10.1037//0022-3514.83.6.1362
Anderson, C., & Galinsky, A. D. (2006). Power, optimism, and risk-taking. European Journal of Social Psychology, 36, 511–536. doi:10.1002/ejsp.324
Anderson, C., John, O. P., & Keltner, D. (2012). The personal sense of power. Journal of Personality, 80(2), 313–44. doi:10.1111/j.1467-6494.2011.00734.x
Babiak, P., Neumann, C. S., & Hare, R. D. (2010). Corporate psychopathy: talking the walk. Behavioral Sciences and The Law, 28, 174–193. doi:10.1002/bsl
Bar-Anan, Y., & Nosek, B. a. (2013). A comparative investigation of seven indirect attitude measures. Behavior Research Methods, 1–89. doi:10.3758/s13428-013-0410-6
Bargh, J. a, Raymond, P., Pryor, J. B., & Strack, F. (1995). Attractiveness of the underling: an automatic power --> sex association and its consequences for sexual harassment and aggression. Journal of Personality and Social Psychology, 68(5), 768–81. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/7776181
BBC. (2008). Profile: Steve Fossett. Retrieved March 01, 2014, from http://news.bbc.co.uk/2/hi/americas/2078591.stm
Campbell, W. K., Hoffman, B. J., Campbell, S. M., & Marchisio, G. (2010). Narcissism in organizational contexts. Human Resource Management Review, 21(4), 268–284.
doi:10.1016/j.hrmr.2010.10.007
Chen, S., Lee-Chai, A. Y., & Bargh, J. A. (2001). Relationship orientation as a moderator of the effects of social power. Journal of Personality and Social Psychology, 80(2), 173–187. Retrieved from http://psycnet.apa.org/journals/psp/80/2/173/
Chhokar, J. S., Brodbeck, F. C., & House, R. J. (Eds.). (2007). Culture and Leadership across the world. The GLOBE book of in-depth-studies of 25 societies. New York, NY: Lawrence Erlbaum
Associates.
Galinsky, A. D., Gruenfeld, D. H., & Magee, J. C. (2003). From power to action. Journal of Personality and Social Psychology, 85(3), 453–66. doi:10.1037/0022-3514.85.3.453
Gawronski, B., Deutsch, R., & Banse, R. (2012). Response Interference Tasks as Indirect Measures of Automatic Associations. In K. C. Klauer, A. Voss, & C. Stahl (Eds.), Cognitive Methods in Social Psychology (Abridged E., pp. 78–123). New York, NY: Guilford Press. González, R., & Brown, R. (2006). Dual identities in intergroup contact: Group status and size
moderate the generalization of positive attitude change. Journal of Experimental Social Psychology, 42(6), 753–767. doi:10.1016/j.jesp.2005.11.008
Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. (1998). Measuring individual differences in implicit cognition: the implicit association test. Journal of Personality and Social Psychology, 74(6), 1464–80. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/9654756
Greenwald, A. G., & Nosek, B. (2001). Health of the implicit association test at age 3. Zeitschrift Für Experimentelle Psychologie, 48(2), 85–93. Retrieved from
http://www.psycontent.com/index/J8R39864G1213Q05.html
Greenwald, A. G., Nosek, B. a., & Banaji, M. R. (2003). Understanding and using the Implicit Association Test: I. An improved scoring algorithm. Journal of Personality and Social Psychology, 85(2), 197–216. doi:10.1037/0022-3514.85.2.197
Gündemir, S., Homan, A. C., de Dreu, C. K. W., & van Vugt, M. (2014). Think leader, think white? Capturing and weakening an implicit pro-white leadership bias. PloS One, 9(1), e83915. doi:10.1371/journal.pone.0083915
Hofstede, G. (2006). What did GLOBE really measure? Researchers’ minds versus respondents' minds. Journal of International Business Studies, 37(6), 882–896. Retrieved from
http://www.palgrave-journals.com/jibs/journal/v37/n6/abs/8400233a.html
House, R. J., Dorfman, P. W., Javidan, M., Hanges, P. J., & de Luque, M. F. S. (Eds.). (2014). Strategic leadership across cultures: The GLOBE study of CEO leadership behavior and effectiveness in 24 countries (1st ed.). London: Sage Publications Ltd.
House, R. J., Hanges, P. J., Ruiz-Quintanilla, S. A., Dorfman, P. W., Javidan, M., Dickson, M., & Gupta, V. (1999). Cultural influences on leadership and organizations: Project GLOBE. Advances in Global Leadership, 1(2), 177–233. Retrieved from
http://www.vnseameo.org/zakir/process.pdf
House, R. J., Javidan, M., Hanges, P. J., & Dorfman, P. W. (2002). Understanding cultures and implicit leadership theories across the globe: an introduction to project GLOBE. Journal of World Business, 37(1), 3–10. doi:10.1016/S1090-9516(01)00069-4
Javidan, M., & House, R. J. (2006). Conceptualizing and measuring cultures and their
consequences: a comparative review of GLOBE’s and Hofstede's approaches. Journal of International Business Studies, 37(6), 897–914. Retrieved from http://www.palgrave-journals.com/jibs/journal/v37/n6/abs/8400234a.html
Javidan, M., & Waldman, D. A. (2003). Exploring Charismatic Leadership in the Public Sector Measurement and Consequences. Public Administration Review, 63(2), 229–242.
Johnson, S. K., Murphy, S. E., Zewdie, S., & Reichard, R. J. (2008). The strong, sensitive type: Effects of gender stereotypes and leadership prototypes on the evaluation of male and female leaders. Organizational Behavior and Human Decision Processes, 106(1), 39–60. doi:10.1016/j.obhdp.2007.12.002
Karpinski, A., & Steinman, R. B. (2006). The single category implicit association test as a measure of implicit social cognition. Journal of Personality and Social Psychology, 91(1), 16–32.
doi:10.1037/0022-3514.91.1.16
Keltner, D., Gruenfeld, D. H., & Anderson, C. (2003). Power, approach, and inhibition. Psychological Review, 110(2), 265–284. doi:10.1037/0033-295X.110.2.265
Kipnis, D. (1972). Does power corrupt? Journal of Personality and Social Psychology, 24(1), 33–41. Retrieved from
http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2945826&tool=pmcentrez&re ndertype=abstract
Koenig, A. M., Eagly, A. H., Mitchell, A. A., & Ristikari, T. (2011). Are Leader Stereotypes Masculine? A Meta-Analysis of Three Research Paradigms. Psychological Bulletin, 137(4), 616– 642. doi:10.1037/a0023557
LaTour, K. a., & LaTour, M. S. (2009). Positive Mood and Susceptibility to False Advertising. Journal of Advertising, 38(3), 127–142. doi:10.2753/JOA0091-3367380309
Lord, R. G., Foti, R. J., & De Vader, C. L. (1984). A test of leadership categorization theory: Internal structure, information processing, and leadership perceptions. Organizational Behavior and Human Performance, 34(3), 343–378. doi:10.1016/0030-5073(84)90043-6
Lord, R. G., & Maher, K. J. (2002). Leadership and information processing: Linking perceptions and performance (New Ed (18.). London: Routledge. Retrieved from
http://www.amazon.com/Leadership-Information-Processing-Perceptions-Organizations/dp/0415099013
Magee, J. C., & Galinsky, A. D. (2008). 8 Social Hierarchy: The Self-Reinforcing Nature of Power and Status. The Academy of Management Annals, 8, 351–398. doi:10.1080/19416520802211628 Miner, J. B., & Raju, N. S. (2004). Risk propensity differences between managers and
entrepreneurs and between low- and high-growth entrepreneurs: a reply in a more
conservative vein. The Journal of Applied Psychology, 89(1), 3–13. doi:10.1037/0021-9010.89.1.3 Nosek, B. a, Greenwald, A. G., & Banaji, M. R. (2005). Understanding and using the Implicit
Association Test: II. Method variables and construct validity. Personality & Social Psychology Bulletin, 31(2), 166–80. doi:10.1177/0146167204271418
Penke, L., Eichstaedt, J., & Asendorpf, J. B. (2006). Single-Attribute Implicit Association Tests (SA-IAT) for the Assessment of Unipolar Constructs. Experimental Psychology (formerly “Zeitschrift Für Experimentelle Psychologie”), 53(4), 283–291. doi:10.1027/1618-3169.53.4.283 Resnick, B. B. (2013). How Power Corrupts the Mind. Retrieved March 01, 2014, from
http://www.theatlantic.com/health/archive/2013/07/how-power-corrupts-the-mind/277638/
Rosch, E. H. (1978). Principles of categorization. In Cognition and categorization (pp. 27–48). Hillsdale, NJ: Erlbaum.
Rosette, A. S., Leonardelli, G. J., & Phillips, K. W. (2008). The White Standard: Racial Bias in Leader Categorization. Journal of Applied Psychology, 93(4), 758–777. doi:10.1037/0021-9010.93.4.758
Schyns, B., & Schilling, J. (2010). Implicit Leadership Theories: Think Leader, Think Effective? Journal of Management Inquiry, 20(2), 141–150. doi:10.1177/1056492610375989
Shondrick, S. J., Dinh, J. E., & Lord, R. G. (2010). Developments in implicit leadership theory and cognitive science: Applications to improving measurement and understanding alternatives to hierarchical leadership. The Leadership Quarterly, 21(6), 959–978. doi:10.1016/j.leaqua.2010.10.004
Smith, E. R., & Zárate, M. A. (1992). Exemplar-based model of social judgement. Psychological Review, 1, 3–21.
Stewart, W. H., & Roth, P. L. (2001). Risk propensity differences between entepreneurs and managers: A meta-analytic review. Journal of Applied Psychology, 86(1), 145–153.
Stewart, W. H., & Roth, P. L. (2004). Data quality affects meta-analytic conclusions: a response to Miner and Raju (2004) concerning entrepreneurial risk propensity. The Journal of Applied Psychology, 89(1), 14–21. doi:10.1037/0021-9010.89.1.14
Thierry, H., Hartog, D. Den, Koopman, P., & Wilderom, C. (2007). Culture and leadership in a flat country-The case of the Netherlands. In Culture and leadership across the world: The GLOBE book of in-depth studies of 25 societies (1st ed., pp. 215–251). New York, NY: Taylor & Francis. Retrieved from
http://books.google.com/books?hl=en&lr=&id=v0txL5LnLcYC&oi=fnd&pg=PA215&d q=Culture+and+leadership+in+a+flat+country:+The+case+of+the+Netherlands&ots=Q 4twXBRtBP&sig=OeunFxP_EcJi52Xj67q4Mao-g-o
Van Honk, J., Hermans, E. J., Putman, P., Montagne, B., & Schutter, D. J. L. G. (2002).
Defective somatic markers in sub-clinical psychopathy. Neuroreport, 13(8), 1025–7. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12060801
Van Kleef, G. A., Cheshin, A., Fang, X., Heerdink, M. W., Koning, L. F., & Stamkou, E. (2014). Risk-takers are seen as powerful. Manuscript in Preparation.
8. APPENDICES
Appendix A
Sections 6 (Work Plan), 9 (Further Steps), and 10 (Signatures) of the Original Proposal.
6. WORK PLAN – max 500 words
Describe how the research project will be executed. Who is doing what and when? Is the planning of the current project realistic, efficient and feasible?
6.1 Time schedule
State the total amount of ec as noted in the internship contract (18-24 ec), 1 ec stands for 28 hours work. Present and justify a time schedule in weeks, including your time investment in hours per week. Plan some spare time, and indicate what elements can be cut / reduced if necessary.
My internship totals 18ec (1ec = 28hrs):
Official distribution of ec and hrs based on my courses
February + March: 6ec 168hrs / 8 weeks 21hrs / week April: 6ec 168hrs / 4 weeks 42hrs / week May + June: 6ec 168hrs / 8 weeks 21hrs / week
Preferred distribution of ec and hrs:
December + January: 1ec 28hrs (≈ 3hrs per paper)
February + March: 6ec 168hrs / 8 weeks 21hrs / week April: 7ec 196hrs / 4 weeks 49hrs / week May + June: 4ec 112hrs / 5 weeks 23hrs / week
The actual distribution of ec and working hours will probably fall halfway between the official and preferred distribution. Please see the document ‘Schedule_Proposal.xlsx’ for a detailed schedule.
6.2 Tasks and duties
For internship proposals only: Indicate to which aspects of the research project you will contribute.
I contributed and will contribute to all stages of the proposed study (with the exception of the general research question of finding an underlying mechanism for the association between risk taking and power, which was already provided).
6.3 Infrastructure
Where will the research take place? How is access to the facilities and materials ensured?
Participants will be recorded in six semi-cubicles in the UvA Psychology Lab in D2.20. 6.4 Budget
The compensation from the department is max € 80 for each research project. If the total expenditure exceeds the maximum compensation, then specify how the surplus will be financed. The € 80 budget may be used for printing costs (e.g. for the conference poster), travel expenses, participant payment. Specify the financial ramifications for the intended research. Please go to the secretariat of the specialization of your supervisor with your receipts. The secretariat will reimburse the costs you made up to € 80.
The proposed study will be embedded in a session of three experiments, which will last approximately one hour. For participation in the session, students will be reimbursed with ten euros (10€) or one (1) proefpersoonuur. The 80€ compensation from the department will be used for this reimbursement, but the majority of the payment will be provided through Prof. Dr. Gerben van Kleef.
Word count = 97
9. FURTHER STEPS
Make sure your supervisor submits an Ethics Checklist for your intended research to the Ethics Committee of the Department of Psychology at http://ce.psy-uva.nl/.
Submit the research proposal in PDF by email to researchmaster-psychology@uva.nl.
If you have the proposal signed by the supervisor(s) and you have scanned their signatures in the PDF, you only have to hand in a digital version of the proposal. However if the signatures are not on the PDF, please also submit a printed copy of the signed research proposal to the secretariat of the Research Master Psychology:
Universiteit van Amsterdam Research Master’s Psychology Weesperstraat 4, room 1.02 1018 XA Amsterdam
researchmaster-psychology@uva.nl
After handing in the proposal you can continue your research. 10. SIGNATURES
I hereby declare that both this proposal, and its resulting report, will only contain original material and is free of plagiarism (cf. Chapter 11 or the Research Master’s course catalogue). I hereby declare that the result section of the report will consist of two subsections, one entitled “confirmatory analyses” and one entitled “exploratory analyses” (one of the two subsections may be empty):
a) The confirmatory analysis section reports exactly the analyses proposed in section 4 of this proposal
b) The exploratory analysis section contains additional, and thus exploratory, analyses. Signature student
_______________________________
Signature ResMas supervisor: Signature Second assessor:
Appendix B
Power Analyses Study 1a and 1b
A power analysis was conducted for two separate one-sided one-sample t-tests with a significance level of α = 0.05, and a power of 0.80. Assuming that effect sizes in this study will be about half of those in the study by Gündemir et al. (2014), the current study will require a minimum of 101 participants for the IAT (expected d = 0.25) and a minimum of 71 participants for the SA-IAT (expected d = 0.30). Each participant will complete both IATs, meaning that the final sample size should be N = 130 to accommodate exclusion of responses based on the criteria detailed in section 4.2.
################################# # R-code for the Power Analysis # ################################# library(pwr) #IAT pwr.t.test(d=0.25,power=0.8,sig.level=0.05,type="one.sample",a lternative="greater") #SA-IAT pwr.t.test(d=0.3,power=0.8,sig.level=0.05,type="one.sample",al ternative="greater")
Appendix C Pilot Study
The wordlist was obtained using several online thesauri (thesaurus.com and synonym.com (in combination with google.translate.com, uitmuntend.de and vandale.nl/opzoeken) and
synoniemen.net). Words with potential resemblance to a category (e.g. ondermenend might resemble ondernemer (entrepreneur) and thus the category leader) were not included in the wordlist. Dutch native speakers were consulted and their suggestions were integrated. The English translations in brackets will not be printed.
Bedankt dat je wilt meedoen aan dit kort onderzoek! Geef alsjeblieft aan in hoeverre je eens bent met het volgend statement (maak een tik in de kolommen 1-7). Geef alstublieft ook een volgorde van de 5 meest belangrijke woorden voor het volgend statement (schrijf bij 5 woorden een 1, 2, 3, 4, of 5 in de column “Rang”. 1 bedoelt ‘meest belangrijk’).
Het statement is:
De volgende woorden zijn VAN GROOT BELANG voor het thema “risico”
1 2 3 4 5 6 7
Helemaal niet van
toepassing toepassing Niet van
Enigszins niet van
toepassing Neutral
Enigszins wel van
toepassing toepassing Wel van
Helemaal wel van toepassing Rang (1-5) riskant (risky) gedurfd (daring) kans (chance) inzet (stake) weddenschap (bet) geluk (luck) gewaagd (hazardous) avontuurlijk (adventurous) gokken (gamble) speculeren (speculate) roekeloos (reckless) impulsief (impulsive) moedig (brave) gevaarlijk
(dangerous) spontaan (spontaneous) hacheljik (dubious)
Doe alstublieft het zelfde voor het volgende statement:
De volgende woorden zijn VAN GROOT BELANG voor het thema “zekerheid”
1 2 3 4 5 6 7 Helemaal niet van toepassing Niet van toepassing Enigszins niet van toepassing Neutral Enigszins wel van toepassing Wel van toepassing Helemaal wel van toepassing Rang (1-5) voorzichtig (careful) behoedzaam (cautious) veilig (safe) zeker (sure) verzekerd (assured) discreet (discreet) berekenend (calculative) verantwoordelijk (responsible) rationeel (rational) zorgzaam (considerate) redelijk (reasonable) bedachtzaam (thoughful) opletten (watching out) comfortable (comfortable) zorgvoudig (careful) betrouwbar (reliable)
Appendix D Power Analysis Study 2
A power analysis was conducted for a one-sided one-sample t-tests with a significance level of α = 0.05, and a power of 0.80. Assuming that effect sizes in this study will be about half of that in Study 1b, the current study will require a minimum of 17 participants (expected d = 0.65). The final sample size should be N = 60 to accommodate exclusion of responses based on the criteria detailed in section 4.2.
################################# # R-code for the Power Analysis # ################################# library(pwr)
pwr.t.test(d=0.65,power=0.8,sig.level=0.05,type="one.sample",a lternative="greater")
Appendix E
Data Exploration Studies 1a through 2 Study 1a
Counterbalancing effects. Administration of congruent and incongruent blocks, as well as left- and right-key assignment was counterbalanced. A 2 (sequence: congruent first vs. incongruent first) x 2 (side: leader assigned to left vs. right) ANOVA revealed significant differences among the counterbalancing conditions. The main effects of sequence F(1, 132) = 13.20, d = 0.09) and side F(1, 132) = 13.17, d = 0.09) were both significant (p < .001).10 IAT effects were larger when congruent trials were presented first (M = 0.55, SD = 0.64, N = 70) compared to when incongruent trials were presented first (M = 0.185, SD = 0.56, N = 66). This was expected as response times are typically lower during the first combined block than during the second combined block (Greenwald & Nosek, 2001). Both
congruent and incongruent trials benefit from this increased performance during the first block, rendering the IAT effect stronger when congruent trials are presented first, and weaker when incongruent trials are presented first. IAT effects were also larger when leader was assigned to the right (M = 0.56, SD = 0.60, N = 67) compared to when leader was assigned to the left (M = 0.20, SD = 0.61, N = 69). This is in contradiction to previous findings where the IAT score was unaffected by side (Greenwald, McGhee, & Schwartz, 1998).
Raw means and percentage correct. The IAT effect incorporates both reaction times and accuracy simultaneously. To investigate reaction times and accuracy separately, a repeated measures analysis was conducted with time point (congruent vs. incongruent trial) as within-subjects variable and average response time for correct trials (RT) and percentage of correct trials (accuracy) as dependent variables. The main effects on accuracy (F(1, 135) = 22.94, p < .001, d =0.15) and on RT (F(1, 135) = 30.44, p < .001, d =0.18) were both significant. Accuracy was higher during congruent trials (M = 0.93, SD = 0.05) than during incongruent trials (M = 0.92, SD = 0.05). Similarly, RTs were lower during congruent trials (M = 766.12, SD = 206.32) than during incongruent trials (M = 843.52, SD = 209.09). These findings suggest that performance was higher during congruent trials than during incongruent trials. The findings therefore converge with those for the IAT effect and support Hypothesis 1a.
Target word. Separate IAT scores were calculated by deleting all responses to the category risk except for the focal adjective. No significant differences emerged in a repeated
measures analysis with these separate IAT scores for each risk-related word. This suggests that all risk-related target words represent the category risk equally well.
Study 1b
Counterbalancing effects. As in study 1a, administration of congruent and incongruent blocks, as well as left- and right-key assignment was counterbalanced. A 2 (sequence: congruent first vs. incongruent first) x 2 (side: leader assigned to left vs. right) ANOVA revealed no significant main effects for side. Repeating the analysis without side yielded a significant main effect of sequence (F(1, 125) = 11.32, p = .001, d =0.08).11 IAT effects were larger when incongruent trials were presented first (M = -1.06, SD = .0.66, N = 61) than when congruent trials were presented first (M = -0.69, SD = 0.57, N = 68). However, since the SA-IAT effect size was negative – indicating that participants associated leader more strongly with safety than with risk – participants likely perceived the trials labelled
10 Results did not differ when all participants were included in the analysis. 11 Results did not differ when all participants were included in the analysis.
‘incongruent’ as congruent, while they perceived trials labelled ‘congruent’ as incongruent. As is typical for any IAT (Greenwald & Nosek, 2001), the absolute SA-IAT effect size was thus larger for conceptually congruent trials (labelled ‘incongruent’) than for conceptually incongruent trials (labelled ‘congruent’).
Raw means and percentage correct. The SA-IAT effect incorporates both reaction times and accuracy simultaneously. To investigate reaction times and accuracy separately, a repeated measures analysis was conducted with time point (congruent vs. incongruent trial) as within-subjects variable and average response time for correct trials (RT) and percentage of correct trials (accuracy) as dependent variables. The main effects on accuracy (F(1, 128) = 164.34, p < .001, d =0.56) and on RT (F(1, 128) = 129.02, p < .001, d =0.50) were both significant. Accuracy was higher during conceptually congruent trials (labelled
‘incongruent’, M = 0.94, SD = 0.05) than during conceptually incongruent trials (labelled ‘congruent’, M = 0.85, SD = 0.08). Similarly, RTs were lower during conceptually
congruent trials (labelled incongruent, M = 683.93, SD = 152.80) than during conceptually incongruent trials (labelled ‘congruent’, M = 841.82, SD = 233.75). These findings suggest that performance was higher during congruent trials (labelled ‘incongruent’) than during incongruent trials (labelled ‘congruent’). The findings therefore converge with those for the SA-IAT effect of Study 1b and suggest an absolute association between leader and safety rather than between leader and risk.
Target word. Separate IAT scores were calculated by deleting all responses to the category risk except for the focal adjective. Significant differences for the separate risk-related words emerged in a repeated measures analysis with separate SA-IAT scores for each risk-related word (F(2.83, 362.76) = 18.78, p < .001, d =0.13). As Figure E1 illustrates, SA-IAT effects were least negative for daring (gedurfd, M = -0.94, SD = 0.65) and most negative for hazardous (gewaagd, M = -1.03, SD = 0.68). Scores for these words differed significantly from those of the other three risk-related words (risky, dangerous, reckless) at α = .05. This suggests that, among all risk-related words, the target word daring resulted in the weakest association between leader and safety, while the target word hazardous resulted in the strongest association between leader and safety. Leader qualities are thus not associated with daring, even less so with risky, dangerous, or reckless, and least with hazardous. The differential effects on the risk-related target words further suggest that they did not represent the category risk equally well in this context.
Figure E1. Mean SA-IAT effects and standard errors for
each risk-related word. * significant at α = .05. ** significant at α = .01. More negative values represent stronger associations with leader and safety.
Study 2
Counterbalancing effects. As in study 1b, administration of congruent and incongruent blocks, as well as left- and right-key assignment was counterbalanced. A 2 (sequence: congruent first vs. incongruent first) x 2 (side: risk assigned to left vs. right) ANOVA revealed no significant effects. Even though non-significant, the trend for sequence was in the expected direction such that IAT effects were larger when congruent trials were
presented first (M = 0.20, SD = 0.56, N = 29) than when incongruent trials were presented first (M = 0.06, SD = 0.57, N = 25).
Raw means and percentage correct. The SA-IAT effect incorporates both reaction times and accuracy simultaneously. To investigate reaction times and accuracy separately, a repeated measures analysis was conducted with time point (congruent vs. incongruent trial) as within-subjects variable and average response time for correct trials (RT) and percentage of correct trials (accuracy) as dependent variables. Only the main effect on accuracy was marginally significant (F(1, 53)= 3.76, p = 0.058, two-tailed, d =0.07). Based on the directional hypothesis (cf. Greenwald & Nosek, 2001), a one-tailed t-test may be used, suggesting that accuracy was significantly higher during congruent trials (M = 0.94, SD = 0.05) than during incongruent trials (M = 0.92, SD = 0.06, p = .029). This finding suggest that performance was higher during congruent trials than during incongruent trials. This finding therefore converges with that for the SA-IAT effect of Study 2 and supports Hypothesis 2.
Target word. Separate SA-IAT scores were calculated by deleting all responses to the category risk except for the focal adjective. No significant differences emerged in a repeated measures analysis with these separate SA-IAT scores for each risk-related word. This suggests that all risk-related target words represent the category risk equally well. Discussion Data Exploration
Three exploratory analyses were conducted on each of the three studies. The results suggest that, first, expected effects of sequence (congruent trials first vs. incongruent trials first) were significant in two studies and in the right direction in all three studies. The unexpected significant effect of right- and left-key assignment of Study 1a did not replicate across studies. Second, results for raw means and percentages correct converged with those for the IAT effects for all three studies. Third, and finally, risk-related target words did not impact IAT effects differently, except for Study 1b.
One possible explanation for why only Study 1b yielded significant differences for risk-related words is that leader was operationalized with leader qualities, rather than with leader roles as in the other two studies. The findings of Study 1b suggest that prototypical,
positive, leader qualities are not associated with daring, less so with risky, dangerous, or reckless, and least with hazardous. In fact, participants might have even interpreted risk-related words as undesirable leader qualities in this context, with daring as least negative and hazardous as most negative quality.
Whereas the positive leader qualities of Study 1b were not associated with risk, this was not the case with leader roles, which were associated with risk in both relative (Study 1a) and absolute terms (Study 2). The fact that the results of Study 1b differ from those of studies 1a and 2 in several ways (i.e. IAT effect in the opposite direction, differential effects of risk-related words), seems to suggest that the operationalization of leader in Study 1b was indeed not well suited for testing the research question. This increases confidence in
