Measuring Implicit SelfEsteem in Adolescence: Evaluating the Psychometric Qualities of an Adapted Version of the Implicit Association Test Lisan N. Kroese University of Amsterdam Student: Lisan Kroese Student number: 10323759 Professor: Marija Maric University: University of Amsterdam Word count: 4988 Word count abstract: 120
Abstract The psychometric qualities of an adapted Implicit Association Test specific for adolescents (IATA) were evaluated. The participants were 152 adolescents (82 female) aged 1218 years (M = 15.04, SD = 1.81) from primary and secondary schools in the Netherlands. Participants completed the IATA, the revised Rosenberg’s SelfEsteem Scale (RSESA) and the revised Self Perception Profile for Adolescents (SPPAR) on a computer. Results showed that the IATA had good internal consistency and divergent validity. However, interitem and itemtotal correlations of the IATA were not yet satisfactory. Therefore it is argued that this study was a good initial step towards developing a psychometrically sound IAT that can accurately measure implicit selfesteem in adolescence, but that more research is needed for further improvement. Implicit and Explicit Selfesteem Selfesteem is considered an important construct in social sciences because of its role in the development and maintenance of internalizing problems1,2. Selfesteem is described as the degree to which one values oneself (3, p. 115). Greenwald and Banaji4 recognized that there is an explicit and an implicit mode of selfesteem4. Explicit selfesteem is the conscious mode of selfesteem4,5. For example, when people are asked how they feel about themselves and they would have the time to consciously reflect on this, their answer would be an indicator of their explicit selfesteem5. Implicit selfesteem, on the other hand, is the unconscious mode of selfesteem4,5. For example, when people are asked to rate the letters of their own name on their positivity or negativity as fast as they can, without reflecting on it consciously, their positive or negative rating could be used as an indicator of their implicit selfesteem5,6.
The dominant theory about the relation between implicit and explicit selfesteem is the dual process theory7. The dual process theory describes that there is a reflective and an associative way of informationprocessing. In the reflective way of informationprocessing information is processed in a rational and conscious manner. In the associative way of informationprocessing information is processed in an automatic and unconscious manner1. As for selfesteem, explicit selfesteem would be a result of the reflective way of processing whereas implicit selfesteem would be a result of the associative way of processing. Because implicit and explicit selfesteem derive from qualitatively different cognitive processes, they can change distinctively and therefore they can differ1. In line with this, research has shown that implicit and explicit selfesteem are only moderately correlated8.9 The Importance of Implicit Selfesteem While most research has focused on explicit selfesteem, implicit selfesteem seems distinct from explicit selfesteem and should therefore be taken into account as well2. Moreover, implicit selfesteem might be an important construct to consider in the relation between selfesteem and internalizing problem1,4,9. While a direct link between implicit selfesteem and internalizing problems has not yet been found, there are numerous studies that show that the discrepancy between implicit and explicit selfesteem is an important predictor of internalizing problems1,2,10. As Leeuwis et al.2 suggested: ”Selfesteem
discrepancies are suggested to be maladaptive because they imply that individuals hold two inconsistent views of the self.” (p. 910). To detect a selfesteem discrepancy, information about the implicit selfesteem in addition to the explicit selfesteem is needed and therefore properly measuring implicit selfesteem is an important endeavor2,11,12.
The Importance of Implicit Selfesteem in Adolescence Measuring implicit selfesteem might especially be important in adolescence. Not only has research shown that implicit selfesteem tends to be at its lowest point at this period in life, but also that implicit selfesteem continuously declines during adolescence while explicit selfesteem increases at the beginning but declines at the end11. These kind of asymmetrical changes in selfesteem increase the risk for developing selfesteem discrepancies1. In adolescence, selfesteem discrepancies have been linked to social anxiety and depression with suicide ideation2,12. Taking this into consideration, it is of great importance to detect such a discrepancy in adolescence so that suitable therapy can be offered. As information about the implicit selfesteem is required for this, it might be of crucial importance to assess implicit selfesteem in adolescence13,14,15. Since implicit selfesteem seems distinct from explicit selfesteem it should also be measured in a different way4,8,11,16. While explicit selfesteem can be assessed by asking people to directly report how they feel about themselves using selfreport measures, this strategy is not suitable for measuring implicit selfesteem5,17. Implicit selfesteem, due to its unconscious nature, can only be assessed using indirect measures5. The Implicit Association Test Because of the increased interest in the role of implicit selfesteem, researchers have developed several implicit selfesteem measures8,11. One of the most widely used instruments for measuring implicit selfesteem is the Implicit Association Test (IAT)18. The IAT is a computertask that aims to measure implicit selfesteem by assessing automatic and unconscious associations related to the self7. For example, people with high implicit selfesteem would unconsciously associate positive words like “good” and “smart” with
themselves whereas for people with low implicit selfesteem this would be the case for negative words like “bad” and “stupid”7. The IAT intents to measure these associations by assessing ”the association between a targetconcept discrimination and an attribute
dimension” (7, p. 1465). The targetconcept discrimination consists of “self” versus “nonself” words and the attribute dimension consists of “pleasant” versus “unpleasant” words7. Words are shown in the centre of the screen and participants are asked to categorize these words by pressing either a left or a right key on the keyboard. An example of one trial of an IAT selfesteem is shown in Figure 1. Figure 1. Screenshot of a trial in the compatible block of a Dutch IAT selfesteem described in “Damaged selfesteem is associated with internalizing problems,” by D. H. Creemers, R. H. Scholte, R. C. Engels, M. J. Prinstein, and R. W. Wiers, 2013, Frontiers in psychology, 4, p. 2. The left key corresponds with both categories shown on the upper left of the screen while the right key corresponds with both categories shown on the upper right of the screen. For every trial a new word appears in the middle of the screen that has to be categorized by pressing either the left or the right key.
In the first two blocks of the test, participants categorize words from one dimension by pressing the corresponding key. For example, “self” words correspond with the left key while “nonself” words correspond with the right key. After this, there are two compatible and two incompatible blocks in which participants categorize words from both dimensions by pressing the corresponding key. For example, in the compatible blocks both “self” and “pleasant” words correspond with the left key, while both “nonself” and “unpleasant” words correspond with the right key (see Figure 1). Implicit selfesteem is assessed by taking the difference in reaction times between trials in the compatible block and the incompatible block7,18. If participants show faster reaction times in the compatible block than in the incompatible block, this is an indicator of high implicit selfesteem7. The block order of the IAT is shown in Table 1. Table 1 Block Order of the IAT SelfEsteem
Block No. of trials Block name Left key response
Right key response
1 20 Target Self Nonself
2 20 Attribute Pleasant Unpleasant
3 20 Compatible
practice
Self + Pleasant Nonself + Unpleasant
4 40 Compatible test Self + Pleasant Nonself +
Unpleasant
5 20 Reversed target Nonself Self
6 20 Incompatible practice Nonself + Pleasant Self + Unpleasant 7 40 Incompatible test Nonself + Pleasant Self + Unpleasant Note. Names shown for both left key and right key response columns are the names of the categories of words that correspond to that key. Each of these categories contains several words. Block 5 is the same as block 1, but with reversed corresponding keys.
Various studies have shown that the IAT possesses the best overall psychometric qualities, compared to other implicit selfesteem measures19. Reliability is the extent to which a measure produces consistent results that stay the same under the same conditions. Validity is the extent to which a measure actually measures what it is intended to measure (20, p. 12). Studies on the psychometric qualities of the IAT selfesteem provided evidence for satisfactory to good reliability and validity7,8,16,17,21. Considering reliability, studies reported good internal consistency, which is the extent to which the items of a measure produce results that are consistent with the overall measure (20, p.708). For the IAT selfesteem, a Cronbach’s α of .82 and SpearmanBrown splithalf reliabilities of .83 and .85 were found17,21. Considering validity, studies reported good divergent validity, which is the extent to which two measures of distinct concepts are unrelated22. For the IAT, correlations between implicit and explicit measures of selfesteem of .13, .17, .18 and .22 were found7,8,16,21. However, it must be noted that the usefulness of the correlation between implicit and explicit selfesteem as an indicator of divergent validity is under debate8,9,23. Central in this debate is the concept of construct validity, which is the extent to which a measure actually measures the construct that it is intended to measure22. Construct validity consists of divergent validity and convergent validity. Divergent validity, as described above, is the extent to which two measures of distinct concepts are unrelated. Convergent validity, on the other hand, is the extent to which two measures of the same concept are related22. According to the dual process theory, the moderate to low correlation mostly found between implicit and explicit selfesteem is desirable, because it demonstrates that they are distinct
Greenwald et al.23 argue that implicit and explicit selfesteem share a latent variable indicated as association strength, and therefore a higher implicitexplicit correlation is desirable since it would be an indicator of construct validity23. In their view, the implicitexplicit correlation could be higher for a more sensitive measure of implicit selfesteem, that is, in case the explicit selfesteem measure is also a proper measure. Therefore they argue that a higher implicitexplicit correlation is possible if implicit selfesteem measures would be improved, for example by using improved scoring algorithms23. Taking all this into account, two things can be concluded about measuring implicit selfesteem in adolescence. First, since the psychometric qualities of the IAT outperformed those of other implicit selfesteem measures, the IAT should be the first choice when intending to measure implicit selfesteem in adolescence7,8,16,17,21. Second, it is important to keep in mind that there is no consensus about the way the validity of the IAT should be interpreted and therefore one should be cautious when evaluating the validity of the IAT in adolescence8,9,23. The Implicit Association Test for Adolescents The IAT has been used to measure implicit selfesteem in adolescents in a small number of studies24,25,26,27. However, two limitations of previous studies are worth mentioning. First, the way in which the IAT was adapted for adolescents was lacking a systematic and argumentative procedure. The earliest study that reported using the IAT to measure selfesteem in adolescence was the study of Sandstrom and Jordan26. They used the IAT for measuring selfesteem in adults by Greenwald and Farnham16 without adapting it for the use with adolescents26. The next study was that of Bos, Huijding, Muris, Vogel and
Biesheuvel24. The only thing they reported about how they adapted their IAT was that they adapted it in accordance with the procedure described in Field and Lawson28. In this study, however, the only procedure described was that of the standard IAT design18,28. Moreover, in the study of Field and Lawson, an IAT that evaluates animal types in children is used. Since Bos et al. intended to evaluate selfesteem in adolescents, it is not exactly clear why they refer to that study for their procedure to adapt the IAT. About the stimulus words used, Bos et al. mention some examples such as “nice”, “smart”, “stupid” and “dumb”. However, why they used these stimuli words and where they got them from, is not mentioned24. It therefore remains unclear what Bos et al. adapted exactly to design their IAT. The remaining two studies that used the IAT to measure selfesteem in adolescence only noted that they used the same stimuli words for their IAT as Bos et al. did. However, they did not mention their reason for this25,27. In sum, IATs have been used to measure implicit selfesteem in adolescence, but the way they have been adapted for the use with adolescents was lacking systematic and argumentative procedure. Second, despite that studies mentioned above used some new, adapted version of the IAT, none of these were conducted with the goal of assessing the psychometric qualities of the IAT used. As a consequence, only two of the studies reported the validity and reliability of the IAT. For the validity, correlations between the IAT and explicit selfesteem measures of .02, .03 and .05 were reported24,27. For the reliability, Bos et al. reported a considerably low SpearmanBrown reliability of .49, whereas van Tuijl et al. reported satisfactory SpearmanBrown reliabilities of .74 and .77 (20, p. 709). This seems ambiguous, considering that the same stimulus words and algorithm for analysing the data were used 24,27. An explanation for that could be that with the SpearmanBrown splithalf method there are a lot of different ways in which the data set can be split into two, since splitting occurs
randomly. Thereby, the way the data were split could have influenced the results (20, p. 708). The spearmanbrown reliability being the only measure of reliability in these studies could therefore be an explanation of the inconsistent results, and other ways of investigating reliability should be used in addition. In sum, the IAT has been used to measure implicit selfesteem in adolescents, but studies on the psychometric qualities of these IAT measures in adolescence are lacking. Taking all this into account, two things can be concluded about the IAT measures that have been used to measure implicit selfesteem in adolescence. First, a study that uses and describes a systematic and argumentative procedure for adapting their IAT and choosing their stimuli words for the use with adolescents, is lacking. Second, there has not yet been a study that is conducted to evaluate the psychometric qualities of an adapted IAT. This Study In this study, the IAT selfesteem used in young adults by Creemers et al.1 was adapted for the use with adolescents. The goal of the current study was to evaluate the psychometric qualities of the adapted IAT for adolescents (IATA). The first research question was: what is the reliability of the IATA? Since the IATA was adapted systematically for the use with adolescent according to the results of a pilot study conducted in this age group, it was expected that the reliability of the IATA was at least as high as the reliability of the IAT for adults by Creemers et al.1. The second research question was: what is the validity of the IATA? Because of the existing disagreement on the utility of the implicitexplicit correlation as either divergent validity or convergent validity, this question was exploratory in nature, so that it was not guided by specific hypotheses but by the
following question: Does the IATA display construct validity by demonstrating either divergent validity or convergent validity? Method Participants The participants were 152 adolescents (82 female) aged 1218 years (M = 15.04, SD = 1.81) from seven primary and secondary schools in the Netherlands. The primary school participants were in sixth grade, the secondary school participants were in senior general secondary education or in preuniversity education. The nationality was Dutch for 94% of the participants, the remaining 6% had other nationalities. The adolescents were recruited by approaching 15 primary and secondary schools in the Netherlands. Parental and adolescent active consent were obtained before testing took place. Materials Implicit selfesteem was assessed with the IATA, which was an adapted version of the IAT selfesteem used in young adults by Creemers et al. (2013). The IAT of Creemers et al. had a significant internal consistency, assessed by computing a correlation between test and practice blocks (r = .62, p < 0.001). The stimulus words used were selected based on their relevance, measured by their positive or negative valence, in a survey study among adults (29, as cited in 30). The IATA used in this study was derived according to the following procedure. First, permission from the first author was acquired to use and adapt the IAT selfesteem of Creemers et al. (2013). Second, a pilot study was conducted among five adolescents (three male) aged 1216 years to assess if the stimulus words were understandable and applicable
for adolescents. Third, based on the pilot study, the category labels and some of the stimulus words for the attribute dimension of the IAT were changed. The stimulus words used in the
targetconcept discrimination were adopted without changing them. The original and adapted category labels and stimulus words are shown in Table 2. Table 2. Original and Adapted Category Labels and Stimulus Words of the Attribute Dimension Original category label Original stimulus words Adapted category label Adapted stimulus words
Valuable Valuable Important Important
Competent Good
Loved Popular
Powerful Strong
Successful Successful
Capable Handy
Worthless Worthless Unimportant Unimportant
Weak Weak Failed Failed Inferior Bad Loser Loser Dumb Dumb Note. Words that were changed in the adapted version of the IATA are shown in boldface. The original category labels and stimulus words were these used in the IAT described in “Damaged selfesteem is associated with internalizing problems,” by D. H. Creemers, R. H. Scholte, R. C. Engels, M. J. Prinstein, and R. W. Wiers, 2013, Frontiers in psychology, 4, p. 2. The adapted category labels and stimulus words were these used in the IATA. All stimulus words were translated from Dutch. Note that the meaning and affective valence of the stimulus words in the English translation might be slightly different than the meaning and affective valence of the Dutch words used in this study.
To measure implicit selfesteem the difference in mean reaction times between trials in the compatible block and trials in the incompatible block was computed. A high score on the IATA indicated a high implicit selfesteem1. The same block order was used as shown in Table 1. However, blocks within the IATA were counterbalanced in order to overcome the unwanted effect of block order in the IAT reported by Greenwald and Nosek (2001). They reported that a stronger association strength is usually found in the first of the combined blocks compared to the second one, irrespectively of the content of the blocks (23; as cited in 31). The IATA was programmed in Inquisit. Explicit selfesteem was assessed using a revised version of the Dutch Rosenberg’s SelfEsteem Scale (RSES)32 for adolescents and a revised version of the Dutch Self Perception Profile for Adolescents (SPPA)33. The revised version of the Dutch RSES (RSESA) was adapted for the use with adolescents and is described in the thesis study of Maan34. Based on the pilot study, in some items only one word was changed, for example “others” was changed into “adolescents” or “peers”. In other items the whole statement was changed, for example “I stand positive towards myself” was changed into “I am happy with myself”. The RSES is a selfreport questionnaire that measures global feelings of selfesteem and consists of 10 items with statements such as “Sometimes I feel completely useless”. Answers were scored on a 4point scale, ranging from 1 (totally agree) to 4 (totally
disagree). Internal consistency and convergent validity of the RSESA were good34. The revised version of the Dutch SPPA (SPPAR) was adapted to make it more applicable for the use with adolescents and is described in the thesis study of Sinceretti35. In the revised version, the left statement from the original questionnaire was taken on every trial and changed into a firstperson statement, with answers being scored on a 4point scale. The content and number of the items stayed the same. The SPPA is a selfreport
questionnaire that measures general feelings of selfworth and specific competencies in adolescents between 12 and 18 years old33. It consists of 35 items containing statements such as “I am very satisfied with the way I look”. Answer were scored on a 4point scale, ranging from 1 (not at all applicable) to 4 (very applicable). Internal consistency and construct validity of the SPPAR were good35. Procedure This study was approved by the Ethics Committee Psychology before data collection took place. Participants completed all measures on a computer within the timeframe of one lesson (ranging from 45 minutes to 60 minutes) in a computer room in their school. The participants signed an informed consent before testing. Since the order of presenting IAT and selfreport measures of selfesteem has been shown to affect the correlation between these measures, the order of the measures were counterbalanced so that half the participants completed the questionnaires first whereas the other half completed the IATA first31. Participants were asked to carry out the test seriously. The researchers were present in case anyone needed assistance. Data analysis strategy Data processing The raw data in the data analysis were used to evaluate the reliability of the IATA. To evaluate the validity of the IATA, IAT effect scores were needed (D scores). The D scores ranged from 2 to +236. To get the D scores, the data were transformed using the D600 algorithm recommended by Greenwald et al.23.
Reliability The reliability of the IATA was evaluated by assessing the internal consistency. This was done by computing a testpractice correlation, a SpearmanBrown splithalf reliability and a Cronbach’s α. Because of the possible limitations of both the SpearmanBrown method and the testpractice correlation considering data were just split in one way, a Cronbach’s α was considered to be needed in addition (20, p. 708). First, a the testpractice correlation was computed for both combined blocks. Second, difference scores between trials in the combined blocks were computed. This was done by subtracting the reaction time on a trial in the compatible block from the reaction time on the corresponding trial in the incompatible block (e.g. subtracting the reaction time of trial 1 of block 3 from the reaction time of trial 1 of block 6). This method made it possible to control for individual differences in baseline reaction time and for order effects that may have occurred8. After this, a SpearmanBrown and Cronbach’s α were computed on these difference scores. The outcome reflected the consistency of the participants to associate the self with either positive or negative words8. Validity The validity of the IATA was evaluated by assessing the construct validity. This was done by computing a Pearson correlation between the IATA and RSESA, as well as between the IAT and SPPAR. Results Of the 150 participants in this study 7 participants completed the wrong version of the IATA, that accidentally contained the original words “valuable” and “worthless”
instead of the adapted words “important” and “unimportant”. Because in this study the psychometric qualities of the adapted version of the IATA containing the adapted words were evaluated, the data of these participants were excluded from further data analysis. While transforming the raw IATA data using the D600 algorithm, trials with latencies longer than 10000 ms were eliminated, as well as participants who responded to more than 10% of their trials with a latency of less than 300 ms23. Two participants had a response latency of less than 300 ms on more than 10% of their trials. The data of these participants were excluded from further data analysis. The data of the remaining 141 participants were used for data analysis. The mean response latency on trials was 1161.56 ms (SD = 285.30) in the compatible block and 1167.23 ms (SD = 352.57) in the incompatible block. The mean D score was 0.08 (SD = 0.63). Pearson correlations between test and practice blocks were computed for the IATA. The testpractice correlation was significant for the compatible block, r = .57, p < .001, and for the incompatible block, r = .76, p < .001. The mean testpractice correlation was r = .66. SpearmanBrown and Cronbach’s α were also computed for the IATA. Results showed that the IATA had high reliability, SpearmanBrown = .88, Cronbach’s α = .92. Itemtotal correlations of the IATA ranged from .14 to .61, with 14 itemtotal correlations below .3. The mean interitem correlation was r = .17. Pearson correlations between the IATA and RSESA as well as between the IATA and SPPAR were computed to assess the association between implicit and explicit selfesteem. There was no significant association between implicit and explicit selfesteem for the correlation between the IATA and the RSESA, r = .02, p = .778, and for the correlation between the IATA and the SPPAR, r = .03, p = .767.
Discussion In this study, the psychometric qualities of the adapted IAT selfesteem for adolescents were evaluated. Results indicated that the IATA showed good reliability. This was in line with the first hypothesis because the reported reliability using the mean testpractice correlation was higher than that in the study of Creemers et al.1, this was also true for the SpearmanBrown and Cronbach’s α. The high reliability values of the IATA were similar to these reported in other studies that used the IAT selfesteem in adults and higher than the ones reported in earlier studies that used the IAT selfesteem in adolescents17,21,24,27. Based on these results, the IATA seemed to be a more reliable measure than earlier IAT selfesteem measures that were adapted for the use with adolescents. In terms of validity, the IATA showed good divergent validity. The second research question was exploratory in nature and was guided by the question if the IATA displayed construct validity by demonstrating either divergent validity or convergent validity. Results indicated that the IATA demonstrated divergent validity. The implicitexplicit correlations found between the IATA and RSESA, as well as between the IATA and SPPAR, were lower than the implicitexplicit correlations found in studies that used the IAT selfesteem in adults and similar to these in earlier studies that used the IAT selfesteem in adolescents7,8,16,21,24,27. This suggests that implicit and explicit selfesteem might even be more distinct in adolescents than in adults. The evidence for divergent validity of the IATA found in this study suggests that the constructs of implicit and explicit selfesteem in adolescence are truly distinct. The results therefore support the dual process theory8,9.
However, some limitations of this study are worth mentioning. First, despite the high SpearmanBrown and Cronbach’s α values for internal consistency, 25% of the items had an itemtotal correlation below r = .3. This indicates that the high internal consistency of the IATA might be questionable, considering that the difference items with values below r = .3 do not correlate well enough with the measure as a whole (20, p. 713). The average interitem correlation was also low, indicating that the different items did not correlate well enough with each other (20, p. 712). This suggests that the high SpearmanBrown and Cronbach’s α values found might not have been caused by the IATA truly being highly internally consistent. Since the number of items is known to influence the SpearmanBrown and Cronbach’s α values, an alternative explanation could be that the high number of items that the reliability analysis was conducted on caused the SpearmanBrown and Cronbach’s α values to go up (20, p. 709). In accordance with the D600 algorithm the practice blocks were included in the analysis, which resulted in 60 difference scores that the reliability analysis was conducted on. Thus, despite the high internal consistency values reported, the reliability of the IATA might still be questionable. Since other studies using the IAT selfesteem did not report the itemtotal or average interitem correlations, it might be that these values were unsatisfying in other studies as well. The IATA could then still be a reliable implicit selfesteem measure for adolescents, compared to other measures. However, it can be concluded that the IATA is not as highly reliable as the results suggested. Second, since the reliability of the IATA might still be questionable, deriving conclusions from the correlation found between implicit and explicit selfesteem should be done cautiously. Because of its reliability being questionable, the IATA might not have been a perfectly accurate measure of implicit selfesteem. Because of this it might not have been able to detect the true correlation between implicit and explicit selfesteem yet. The
very low implicitexplicit correlations reported in this study could have been influenced by measurement error and should therefore be interpreted cautiously. It might be that a higher correlation would have been obtained if the IATA was perfectly internally consistent. Even though the results of this study support the dual process theory, it could still be that if itemtotal and average interitem correlations were satisfying, a higher correlation could have been obtained, suggesting convergent validity. This would have been supporting the view of Greenwald et al.23. The low implicitexplicit correlations found in adolescence might be caused by the implicit selfesteem measures being less developed for adolescents, considering only a few studies used the IAT in adolescents24,25,26,27. Therefore, more research is needed in order to make a decisive argument in the debate about the relation between implicit and explicit selfesteem in adolescence. Third, because the secondary school participants in this study were all in senior general secondary education or in preuniversity education, they were not representative for the whole population of adolescents in the Netherlands. Therefore, the results found in this study only apply to a part of the population of adolescents in the Netherlands. A number of suggestions are made for future research. First, it is suggested that the reliability analysis could be conducted on all the items or the uneven items of the combined test blocks only, thereby reducing the number of difference scores to 40 or 20. The SpearmanBrown reliability values in this study were lower if conducted on either 40 or 20 items (.81 or .79). This was also true for Cronbach’s α (.88 or .79). By conducting the reliability analysis on a smaller number of items, it would be easier to conclude that a high internal consistency value found is caused by the IAT really being internally consistent. Second, it is suggested that in future research participants should be equally divided among all levels of secondary school. In this way, a representative sample of Dutch
adolescents will be used, which will be important in developing a representative IAT selfesteem for adolescents. Third, in future research it might be useful to conduct a pilot study on a larger number of adolescents, so that the choice for the adaptation of stimulus words would be based on data from a larger and therefore more representative sample. In this way, the adapted words used in the IAT selfesteem for adolescents could be more relevant and applicable for adolescents. An additional option could be to develop an IAT selfesteem from scratch. To select relevant words for adolescents, a large pilot study could be conducted in which adolescents rate a list of words on their positive or negative valence, like in the study of Hermans and de Houwer29. Based on this pilot study, words that were rated most strongly positive or negative could then be used in the new IAT selfesteem for adolescents. In this way, words that are most relevant for adolescents will be used, which could have a positive influence on the psychometric qualities of the IAT selfesteem for adolescents. The research findings in this study suggest that adapting an IAT selfesteem for adolescents in a systematic and argumentative way can produce promising results. However, there is still much room for improvement considering the psychometric qualities of the IAT. Further improving and optimizing the IAT selfesteem for adolescents is highly important, because only with a psychometrically sound measure can implicit selfesteem be measured accurately. Then adolescents can be offered the best suitable treatment based on both their implicit and explicit level of selfesteem. This is relevant because solely focusing on enhancing explicit selfesteem might in some cases, such as when both implicit and explicit selfesteem are low, have a negative effect since it could increase the implicitexplicit discrepancy14. Here, an intervention that enhances implicit selfesteem, for example with
methods of classical conditioning, might be needed in addition to an intervention than enhances explicit selfesteem, like cognitive behavioural treatment37. When a selfesteem discrepancy exists, a mindfulness component in therapy might be suitable since it has proved to increase congruence between implicit and explicit selfesteem by allowing people to bring their intuitive and explicit selfviews more in line, hereby increasing inner harmony and decreasing the implicitexplicit discrepancy13. Therefore, if this study would encourage other researchers to study and improve the IAT selfesteem for adolescents, offering suitable treatment for the variety of selfesteem problems in adolescents could be achievable. References 1. Creemers, D. H., Scholte, R. H., Engels, R. C., Prinstein, M. J., & Wiers, R. W. (2013). Damaged selfesteem is associated with internalizing problems. Frontiers in Psychology, 4, 17. 2. Leeuwis, F. H., Koot, H. M., Creemers, D. H., & van Lier, P. A. (2014). Implicit and Explicit SelfEsteem Discrepancies, Victimization and the Development of Late Childhood Internalizing Problems. Journal of Abnormal Child Psychology, 43, 909919. 3. Blascovich, J., & Tomaka, J. (1991). Measures of selfesteem. In J. P. Robinson, P. R. Shaver & L. S. Wrightsman (Eds.), Measures of Personality and Social Psychological Attitudes (pp. 115160). San Diego, CA: Academic Press. 4. Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: Attitudes, selfesteem, and stereotypes. Psychological Review, 102(1), 427. 5. Kernis, M. H. (2003). Toward a conceptualization of optimal selfesteem. Psychological
Inquiry, 14(1), 126.
6. Koole, S. L., Dijksterhuis, A., & van Knippenberg, A. (2001). What's in a name: Implicit selfesteem and the automatic self. Journal of Personality and Social Psychology, 80(4), 669685. 7. Buhrmester, M. D., Blanton, H., & Swann Jr, W. B. (2011). Implicit selfesteem: Nature, measurement, and a new way forward. Journal of Personality and Social Psychology, 100(2), 365385. 8. Bosson, J. K., Swann Jr, W. B., & Pennebaker, J. W. (2000). Stalking the perfect measure of implicit selfesteem: The blind men and the elephant revisited?. Journal of Personality and Social Psychology, 79(4), 631643. 9. Roefs, A., Huijding, J., Smulders, F. T., MacLeod, C. M., de Jong, P. J., Wiers, R. W., & Jansen, A. (2011). Implicit measures of association in psychopathology research. Psychological Bulletin, 137(1), 149193. 10. Schröder‐Abé, M., Rudolph, A., & Schütz, A. (2007). High implicit self‐esteem is not necessarily advantageous: Discrepancies between explicit and implicit self‐esteem and their relationship with anger expression and psychological health. European Journal of Personality, 21(3), 319339. 11. Cai, H., Wu, M., Luo, Y. L., & Yang, J. (2014). Implicit selfesteem decreases in adolescence: A crosssectional study. PloS one, 9(2), 15. 12. Franck, E., De Raedt, R., Barbez, C., & Rosseel, Y. (2008). Psychometric properties of the Dutch Rosenberg selfesteem scale. Psychologica Belgica, 48(1), 2535. 13. Koole, S. L., Govorun, O., Cheng, C. M., & Gallucci, M. (2009). Pulling yourself together: Meditation promotes congruence between implicit and explicit selfesteem. Journal of Experimental Social Psychology, 45(6), 12201226.
14. Korrelboom, K., de Jong, M., Huijbrechts, I., & Daansen, P. (2009). Competitive memory training (COMET) for treating low selfesteem in patients with eating disorders: A randomized clinical trial. Journal of Consulting and Clinical Psychology, 77(5), 974980. 15. Robins, R. W., & Trzesniewski, K. H. (2005). Selfesteem development across the lifespan. Current Directions in Psychological Science, 14(3), 158162. 16. Greenwald, A. G., & Farnham, S. D. (2000). Using the implicit association test to measure selfesteem and selfconcept. Journal of Personality and Social Psychology, 79(6), 10221038. 17. Rudolph, A., SchröderAbé, M., Schütz, A., Gregg, A. P., & Sedikides, C. (2008). Through a glass, less darkly? Reassessing convergent and discriminant validity in measures of implicit selfesteem. European Journal of Psychological Assessment, 24(4), 273281. 18. 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), 14641480. 19. Krause, S., Back, M. D., Egloff, B., & Schmukle, S. C. (2011). Reliability of implicit self‐esteem measures revisited. European Journal of Personality, 25(3), 239251. 20. Field, A. (2013). Discovering statistics using IBM SPSS statistics. Sage.
21. BarAnan, Y., & Nosek, B. A. (2014). A comparative investigation of seven indirect attitude measures. Behavior Research Methods, 46(3), 668688.
22. Bagozzi, R. P., Yi, Y., & Phillips, L. W. (1991). Assessing construct validity in organizational research. Administrative Science Quarterly, 36(3), 421458.
implicit association test: I. An improved scoring algorithm. Journal of Personality and Social Psychology, 85(2), 197216. 24. Bos, A. E., Huijding, J., Muris, P., Vogel, L. R., & Biesheuvel, J. (2010). Global, contingent and implicit selfesteem and psychopathological symptoms in adolescents. Personality and Individual Differences, 48(3), 311316. 25. Jong, P. J. de, Sportel, B. E., De Hullu, E., & Nauta, M. H. (2012). Cooccurrence of social anxiety and depression symptoms in adolescence: Differential links with implicit and explicit selfesteem?. Psychological Medicine, 42(03), 475484. 26. Sandstrom, M. J., & Jordan, R. (2008). Defensive selfesteem and aggression in childhood. Journal of Research in Personality, 42(2), 506514. 27. Tuijl, L. A. van, de Jong, P. J., Sportel, B. E., de Hullu, E., & Nauta, M. H. (2014). Implicit and explicit selfesteem and their reciprocal relationship with symptoms of depression and social anxiety: A longitudinal study in adolescents. Journal of Behavior Therapy and Experimental Psychiatry, 45(1), 113121. 28. Field, A. P., & Lawson, J. (2003). Fear information and the development of fears during childhood: Effects on implicit fear responses and behavioural avoidance. Behaviour Research and Therapy, 41(11), 12771293. 29. Hermans, D., & De Houwer, J. (1994). Affective and subjective familiarity ratings of 740 Dutch words, Psychologica Belgica, 34(23), 115139. 30. Franck, E., De Raedt, R., Dereu, M., & Van den Abbeele, D. (2007). Implicit and explicit selfesteem in currently depressed individuals with and without suicidal ideation. Journal of Behavior Therapy and Experimental Psychiatry, 38(1), 7585. 31. Nosek, B. A., Greenwald, A. G., & Banaji, M. R. (2005). Understanding and using the Implicit Association Test: II. Method variables and construct validity. Personality
and Social Psychology Bulletin, 31(2), 166180.
32. Rosenberg, M. (1979). Conceiving the self. New York: Basic Books.
33. Treffers, P. D. A., Goedhart, A. W., Veerman, J. W., Van den Bergh, B. R. H., Ackaert, L., & De Rycke, L. (2004). Competentie belevingsschaal voor Adolescenten. Tijdschrift voor Psychiatrie, 7, 468469. 34. Maan, Y. (2016). Onderzoek naar Aanpassing van de Rosenberg’s SelfEsteem Scale voor Adolescenten [Study on the revise of the Rosenberg’s SelfEsteem Scale for Adolescents] (Undergraduate). University of Amsterdam, Amsterdam. 35. Sinceretti, M. (2016). Psychometrische eigenschappen van de competentie belevingsschaal voor adolescenten [Psychometric features of the Dutch Self Perception Profile for Adolescents] (Undergraduate). University of Amsterdam, Amsterdam. 36. Nosek, B. A., Greenwald, A. G., & Banaji, M. R. (2007). The Implicit Association Test at age 7: A methodological and conceptual review. Automatic Processes in Social Thinking and Behavior, 265292. 37. Baccus, J. R., Baldwin, M. W., & Packer, D. J. (2004). Increasing implicit selfesteem through classical conditioning. Psychological Science, 15(7), 498502.