Administrator Effects on Respondent Choice

Research concerning administrator effects and the wider field of experimenter expectancy
effects (Rosenthal, 1976) has established the theory that the hypotheses and knowledge
held by an experimenter can unconsciously influence their results. Therefore, in a novel use
of a photographic line-up from an actual police investigation, this research aims to explore
the impact of administrator effects without the memory component of a recalled event.
Previous research in this area has used a traditional memory paradigm to test administrator
effects, however this has clouded the issue of whether the witness is being influenced by
the administrator or is actually remembering the event. This research removes the memory
component and therefore concentrates on the expectancy effect of the administrator.
In order to further the understanding of administrator effects, this research examines
whether there is an aspect of interpersonal behaviour which predisposes some individuals
to be more susceptible to inferences from others, or predisposes some to be more likely to
influence individuals than others. In order to do this the Fundamental Interpersonal
Relations Orientation: Behaviour (FIRO-B) instrument is utilised to examine the
interpersonal relationship behaviour of the administrator and the participant. This research
also identifies the cues emitted by the administrator by audio-recording the interaction
between the administrator and the participant.
Using an experimental design, which manipulated the knowledge of the location of the
target, five hundred and twenty six participants were asked to identify the person
responsible for the Lockerbie bombing. Line-up administrators, who were either informed of
the location of the suspect, informed of the location of an alternative suspect, or
uninformed of the location of the suspect, presented the participants with the photographic
line-up of twelve men, one of which is believed to be the person responsible for the
Lockerbie bombing. Participants were asked to pick the person they thought was the
suspect from the line-up, they then completed the FIRO-B questionnaire.
Analysis of the frequency of identifications suggests the presence of an experimenter
expectancy effect. A chi-square goodness of fit test indicates that significantly more
participants identified the target suspect when the administrator was informed of the
location of the target, than when the administrator was uninformed. Analysis of the FIRO-B
data found that target identifiers in the informed condition reported significantly higher
received control scores than non-identifiers from an informed condition matched
comparison group. Target identifiers also reported significantly higher social interactivity
and received inclusion scores than non-identifiers.
Analysis of the FIRO-B data from the line-up administrators found subtle differences in the
FIRO-B scores of the administrators achieving a high number of target identifications
compared to administrators achieving a low number of target identifications. In particular,
administrators achieving a high number of target identifications reported higher levels of
expressed control and lower levels of received control than administrators achieving a low
number of target identifications. Analysis of the transcripts of the line-ups indicate that
administrators in the informed condition interacted with their participants for longer, and
exhibited more verbal cues. Administrators who obtained a target identification also spoke
to their participants for longer. Those administrators who spoke to their participants for
longer reported higher levels of expressed control and lower levels of received control.
The results of this study point to an experimenter expectancy effect. Beyond that though
there appears to be an aspect of interpersonal behaviour that may be responsible for a
predisposition to influence or to be influenced. This thesis, in line with previous research
advocates the use of ‘double-blind’ line-up procedures in order to eradicate the possibility
of an administrator effect. However, it also highlights the importance of considering the
social interaction between the experimenter and the participant that is at the heart of social
psychology research with human participants. In particular, the damning effect on the
results of research conducted by an experimenter who assumes the dominant role in a
social interaction, with a participant who assumes the submissive role.