Doping in sport, defined as the presence of a prohibited substance or its metabolites or markers in an athlete’s sample, or evidence of the attempted use or use of a prohibited method , appears to be widespread . For example, recent analyses of biochemical data from 2,737 elite track and field athletes revealed incidents of blood doping in an average of 14% of athletes, with up to 48% positive samples in athletes from particular countries . It is thus astonishing that for example in the year 2011, only 1.2% of the 243,193 test samples, which have been analyzed in accredited WADA (World Anti-Doping Agency) laboratories, produced adverse analytical findings . Humble detection rates like this might have contributed to what has been described as a shift from detection-based deterrence to prevention-based deterrence of doping in sport [5, 6].
Evaluations of doping prevention programs raise the problem (among others) that open interviews with participating athletes, for example whether or how much their attitude towards doping has changed as a result of the prevention program, do not always lead to conclusive information. Especially, the tendency to respond in a socially desirable manner (the athlete knows what the interviewer expects to be the “right” answer) has a negative effect on the validity of respective self-report data . For example, it has been shown that answers indicated in a standardized questionnaire, from athletes who exhibited positive doping results in biochemical hair analyses, allowed no conclusions about their factual doping behavior . When it comes to detecting changes in the behavior of athletes who participated in a doping prevention program, it may be of limited value to solely rely on data from such direct inquiries therefore. However, by far the largest part of all existing studies on the effectiveness of doping prevention programs in sport is based on such self-reports .
Direct and indirect testing of athletes’ doping attitudes
There is an increasing number of empirical studies, in which (sets of) cognitive determinants of the intention to dope, or more seldom the behavior itself, have been investigated [10–12]. Most of them include and several have been devoted to the investigation of doping attitudes. This is because most psychologists will concur that attitude is one of the most important cognitive predictors of intention and/or behavior .
Three fundamental characteristics can be used to essentially define the psychological construct of attitude . First, attitudes fulfill an evaluative task in assessing objects and persons. Second, important attitudes are represented in a person’s memory. Third, attitudes can have affective, cognitive and behavioral causes, as well as bring about changes in affect, cognition and behavior themselves. Current social-cognitive theories of behavior (dual system models, as for example the Reflective Impulsive Model, RIM) assume that reflective as well as automatic cognitive processes influence behavior . It is thus important to differentiate between explicit and implicit attitudinal components . Explicit attitudes are evaluative judgments about an attitude object that result from processes of conscious, deliberate thought. They can be measured by using direct tests, for example standardized attitude questionnaires. In contrast, implicit attitudes are evaluative reactions resulting from spontaneous cognitive associations, which are automatically activated by a relevant stimulus . Reaction-time based indirect tests have been shown to be the method of choice for measuring this attitudinal component.
The Performance Enhancement Attitude Scale, PEAS , is one of the most widely used direct tests to measure (thus explicit) doping attitudes [19, 20]. This questionnaire contains items such as “doping is necessary to be competitive”, which has to be evaluated on a 6-point Likert-type scale ranging from “strongly disagree” to “strongly agree”.
Only very few studies have attempted to measure implicit attitudes related to doping in sport [8, 21, 22]. All of these drew on variations of one reaction time-based indirect test, the Implicit Association Test, IAT .
The IAT represents one class of tests, which are based on the theoretical assumption that knowledge is stored in our memory in a “networked” fashion. Nodes represent knowledge in an associative network of semantic information. If a node of the network is activated, this activation spreads within the network and associated nodes are co-activated automatically . IATs take advantage of the fact that the activation of an attitude object (of a semantic network node; e.g. through presentation of the drug name “Erythropoietin/EPO”, which can be used as a doping substance) automatically activates the evaluation associated with this attitude object, e.g. “positive” or “negative”.
The IAT is typically presented as a lexical sorting task (speed test) on the computer where two concepts (one target and one evaluative) are mapped on the same response key of the computer’s keyboard. The sorting task is easier (and thus requires less time; a few hundred milliseconds on average) when the two concepts sharing the same response key (e.g. doping and dislike) are closely associated than when the two concepts on the same key are not associated (doping and like). Several different ways of calculating test scores have been proposed . All of these approaches use the difference in response time between related and unrelated pairs, which is then interpreted as a measure of the associative strength between the target concept and attribute characteristics.
It is important to mention that there is controversy about the implicitness of the processes measured by the IAT and its test variants . It is beyond doubt at the other hand that the test has evolved as one of the standard measures of implicit attitudes in social cognition research [27, 28]. Recognizing the notwithstanding well-founded critique on the IAT, we will refer to it as an established indirect test throughout this paper, and avoid a language suggesting that IAT test scores would reflect unbiased (pure) implicit processes.
Researchers have argued in favor of reaction-time based indirect tests that, compared with questionnaires (direct tests), are easier to conceal the ultimate goal of measurement from the participant . Therefore, the IAT may be less susceptible to the social desirability bias. On the other hand, critics have argued that one weakness of the IAT is that it is sometimes difficult to prove its validity. Non-associative factors, like for example cognitive skills, familiarity and perceptual similarity of stimuli, are able to bias the measurement and could also be used to explain observed differences in reaction-times therefore . Arguments like these counter one fundamental assumption of the IAT, namely that differences in response-times correspond only to the associative strength between the categories. Another critique is that deliberate faking of the IAT seems to be possible, especially if participants are informed about the test’s general setup and content . Aside from that, until today, results from numerous studies indicate that the IAT is a sensible methodological choice for attitude measurement when socially sensitive information is under question, e.g. gender stereotypes or prejudices against other social groups [28, 32–35].
The method of choice for validating a doping IAT certainly consists of indirectly measuring the doping attitudes of athletes who have verifiably taken an illegal substance, and compare them with those of athletes who have verifiably not. The indirect test then should reveal more positive doping attitudes in the first group than in the second group. The results of one pilot study points in this direction. Petróczi et al. showed that athletes who were found guilty of taking doping substances on the basis of biochemical hair analyses indicated more positive attitudes in an IAT than athletes who tested negative . However, these results are based on an extremely small sample (6 dopers vs. 4 non-dopers) and the method of biochemical testing may have been suboptimal. For example it is unclear from the article whether a quantitative or qualitative differentiation of endogenous hormones and externally introduced substances was carried out.
Insofar as the biochemical test for the use of doping substances, which should ideally be used as an external criterion for validating experimental results, is very costly and, especially, since it is not easy to find doping positive athletes who would participate in such studies, other researchers have employed variations of this known-group validation strategy.
Petróczi, Aidman, and Nepusz expected, for example, that athletes who are regularly involved in competitions would exhibit a stronger dislike for doping . Rather unexpectedly, they could not find this difference. A study by Lotz and Hagemann, by contrast, suggested a group difference between more (bodybuilding and track-and-field) or less (handball and table tennis) doping prone sports . However, in their study, the indirectly measured attitude scores correlated with a random factor.
There is evidence that IAT results are strongly dependent on the test stimuli used . So one reason for the inconclusive evidence in these two initial studies on doping IATs [22, 35] may be that they used suboptimal test stimuli. Results from one later methodological study, which found substantial error rates and adaptational learning effects associated with both IATs, point into this direction . Another or an additional reason may be that the between-group variance of IAT scores was marginal in the first  as well as in the second study . Both have found fairly negative evaluations of doping (doping attitudes) in all groups.
The present research
This article introduces a picture-based brief-IAT (BIAT) for the indirect measurement of athletes’ doping attitudes. BIATs have already been used successfully in various other behavioral contexts [37, 38]. This methodological (brief) variant of the standard IAT contains considerably fewer (less than half) sorting trials than the standard IAT procedure, resulting in much shorter testing time (less than five minutes). We expect the development of such a time efficient variant to bring about improved test compliance on the side of the participants.
The here presented doping-BIAT uses pictures instead of word stimuli. Considerations for the use of pictures instead of word stimuli go back to the early years of IAT research [23, 39]. Picture (B)IATs appear to produce slightly smaller IAT effects than word (B)IATs . One possible explanation lies in the different representation levels of the stimuli . Words are more abstract and elicit more associations (personal images) than concrete pictures. Despite this potential drawback, we focus our development efforts on the use of pictures as stimuli. The main reason for this is that we expect this to facilitate the applicability and further examination of this doping-BIAT beyond language barriers.
The picture-based doping-BIAT will be validated using a known-group validation strategy [22, 35]. A group of athletes is approached, for which the probability of finding a greater number of subjects with positive doping attitudes is particularly high: We address bodybuilders who regularly participate in bodybuilding competitions. Studies concerning the prevalence of doping in this sport suggest that up to 40% of these athletes regularly consume doping substances [42, 43]. Using anabolic steroids for example, is seen as an integral part of this sport’s culture among most athletes . More importantly, several researchers have illustrated that bodybuilders are comparably open to confess their abuse. The doping attitudes of bodybuilders are compared with those of handball players. Unfortunately, one cannot be sure that handball has a de facto low prevalence of doping, although this is suggested by the very low figures in recent WADA reports . Due to the handball association’s strict official anti-doping policy, we expect to have a much better chance of sampling handball players with comparably less positive attitudes towards using doping substances anyhow. This choice, handball vs. bodybuilding, was supposed to maximize the between-group variance in the participating athletes’ (B)IAT scores.
We expect that both doping attitude tests, the direct as well as the indirect one, unsheathe significantly more positive personal evaluations of doping in the group of bodybuilders than in the group of handball players. This is considered an empirical indicator for the picture-based doping-BIAT being capable of providing valid information about athletes’ doping attitudes.