The Effect of Experimenter Bias and Demand Characteristics on Participant’s Responses Within the Bart Task

...

When run, the PsychoPy file asked participants to evaluate their own risk taking behaviour on a scale of 1-10 with 10 being the most risky. Participants were then asked their age, digit ratio, gender and which group they were in. The screen then went black and at the top of the screen in white Arial text the words “You have nothing banked yet” were displayed. The controls, press SPACE to pump the balloon and press RETURN to bank the sum, were displayed in the bottom of the screen. A red balloon was on the left hand of the screen and as you pressed space the balloon expanded. The balloon, unknown to the participants, was under a set of controls which meant it could be pumped to a set amount of pumps before it was burst which was randomised for each trial.

After the experiment, it was revealed to the participants that in fact the study was not about the effects of testosterone on risk-taking behaviours but instead about the effects of experimenter bias and demand characteristics on subject’s responses. This is when the subject knows the aim of the study therefore tries to conform to what they believe to be the experimenters wanted outcomes. Participants were then told in reality the average digit ratio was a lot higher, .947 for males and .965 for females and that other factors such may impact this.

Results

Appendix II shows the data when Groups A and B were split and a bivariate Pearson Correlation was run. The data collected did not provide any inferences about any linear pattern for Group A, (as shown in Appendix II, Figure 1.) however for Group B (Appendix II, Figure II) one of the pieces of data was statistically significant. The Correlation of risk rating and BART score (r=.581) based on n=21 observations with pairwise non-missing values. The two asterisks marked in the risk rating and BART score for group B show that there is a .01 significance level, the Pearson correlation coefficient is .581, which is significant (p . They have a moderate positive linear relationship meaning those with a greater risk rating are associated with a greater BART score thus insinuating that higher BART scores reflect more risk-taking behaviour . An independent two-tailed samples test was carried out for statistical significance t(43)=.812;p=.421. This is greater than .05 so there is no statistical significant difference between the two groups and shows the limited effect of demand characteristics. Although, Group B had been told that if that had a larger digit ratio they would be more masculine, the results show that these participants had only just a higher BART score (Group A M=31.49, Group B M=32.82). Group B did have a higher mean, 5.71, for risk rating versus Group A,4.75, but due to this being a scale, it bears little aid in trying to determine whether participants thought of themselves as more risk-taking due to the information the experimenter had told them.

Discussion

The results showed no significant correlation between digit ratio and BART score nor digit ratio and risk rating. Although, correlation was moderate between risk rating and BART score but only in Group B. Due to the lack of evidence, the conclusion that experimenter bias had little effect upon the participants can be drawn. Within the experiment there were, however, several limitations. The sample size in each of the groups were not equal with Group A having 24 participants and Group B having 21, which were small sample sizes limiting the amount of data collected and the conclusions that were able to be drawn from it. Within the study there were only 4 males, which were unevenly distributed with 3 males being in Group A and 1 in Group B.Another factor to consider is the use of the risk rating in this task, which calls on the participant to rate their own behaviour and quantify it in terms of a 1-10 scale. The issues with this being not only does it have a hugely limited reliability but also is widely broad in terms of variation among informants as not every participant may view the same behaviour with the same level of risk. The isolating technique of giving Group A and B different powerpoints did mean participants were not aware of others differing condition, which meant participants data were not disrupted. Another key factor is the effect of the reward value on performance. In the experiment, the reward value increased as the balloon increased so it is difficult to perceive whether participants were increasing the balloons size due to being more risk-taking or whether they were eager for a larger reward. Previous studies do, however, suggest high BART scores reflect non-strategic impulsive risk-taking tendencies, rather than calculated methods to maximise their payoff (Helfinstein et al., 2014; Lejuez et al., 2003). The statistical evidence obtained from the experiment did not show participants conforming to experimenter bias nor letting this disrupt their results. In future work, a post experimental questionnaire could be added at the end of the experiment. Rubin et al. (2010) developed a Perceived Awareness of the Research Hypothesis (PARH) scale in which participants indicate how aware they were of the researchers’ set hypotheses within the study. A mean is then calculated of the PARH scores and correlated with the key effects to indicate whether demand characteristics may be related to the research results. However, using digit ratio as a diagnostic tool for measuring ones risk taking behaviour is rather limited. Instead participants’ testosterone levels could have been measured by either a blood or saliva test, which would have given more accurate results. Another method in calculating participants’ risk-taking behaviour is the DOSPERT scale, which can be accessed online as various versions (Blais and Weber, 2006). This is a psychometric scale that assesses risk taking in several domains and would be a lot more informative than just getting participants to select their own risk-rating.

References

Blais, A.-R., & Weber, E. U. (2006) A Domain-Specific Risk-Taking (DOSPERT) scale for adult populations. Judgment and Decision Making, 1, 33-47.

Helfinstein, S. M., Schonberg, T., Congdon, E., Karlsgodt, K. H., Mumford, J. A., Sabb, F.

W., ... Poldrack, R. A. (2014). Predicting risky choices from brain activity

patterns. Proceedings of the National Academy of Sciences, 111, 2470–2475.

http://dx.doi.org/10.1073/pnas.1321728111.

Lejuez, C. W., Aklin, W. M., Zvolensky, M. J., & Pedulla, C. M. (2003). Evaluation of the