• No results found

2.3 Overview of Measures

2.3.8 Implicit Measure. The journal paper focused on explicit and

behavioural measures, here the implicit measure, the IRAP is explored in more detail.

The IRAP (Barnes-Holmes et al., 2006) procedure has been described in detail elsewhere (see Barnes-Holmes et al., 2010; and Hussey et al., 2015). In brief it presents visual stimuli with relational terms such as true, false, same, opposite, and participants are instructed to respond in a particular direction.

Different latencies are said to be indicative of some previous learned verbal

178 behaviour and it demonstrates adequate psychometric properties (see Golijani-Moghaddam et al., 2013; further information provided below).

2.3.8.1 Psychometric properties of the IRAP. A brief summary of the psychometric properties is outlined below.

2.3.8.1.1 Construct validity of the IRAP. Construct validity relates to whether an assessment measures what it claims to measure. Given that implicit cognition cannot be readily observed, it is difficult to determine if it is actually assessed by the IRAP. Therefore, construct validity is dependent on our belief regarding whether what we are measuring actually exists (Sechrest, 2005).On the condition that the construct exists, then both convergent and discriminant validity is necessary in order to estimate construct validity.

2.3.8.1.2 Convergent validity. Measures which are supposed to be related on the same construct and are shown to correlate are said to have convergent validity. The IRAP has demonstrated convergent validity by correlating with explicit measures assessing the same construct (Cullen, Barnes-Holmes et al., 2010; Power et al., 2009), and furthermore with the IAT when assessing the same construct implicitly (Roddy, Stewart & Barnes‐

Holmes, 2011). However, it is important to note that correlation does not equal causation and just because a measure correlates with another does not mean that the construct is necessarily true. Another means of assessing convergent validity is by examining the differences between groups of individuals who would be hypothesised to score differently. For example, the IRAP has proved effective in examining group differences based on food preferences (Barnes-Holmes, Murtagh, Barnes-Holmes & Stewart, 2011) and sexual preferences amongst sexual offenders (Dawson, Barnes-Holmes, Gresswell, Hart & Gore, 2009).

2.3.8.1.3 Discriminant validity. Discriminant validity is dependent on a measure not making associations. A way of assessing discriminant validity is exploring non-correlational data between constructs considered different. The IRAP appears to bypass social desirability by picking up on different information in comparison to self-report measures (Nosek, Greenwald & Banaji, 2007). This has been found in a number of studies, however, one such example relates to responses on a measure assessing racial stereotyping (Barnes-Holmes et al., 2010). Patterns of divergence have also been reported in the study by Dawson

179 and colleagues (2009) whereby similar response patterns on the IRAP were found in explicit responses.

2.3.8.1.4 Criterion validity. Criterion validity also referred to as concrete validity is classified when a measure demonstrates its usefulness in strongly relating to other behaviours and constructs. There are two types of criterion validity; concurrent and predictive. Regarding the former, concurrent validity, this is dependent on how well the IRAP correlates with known responses on areas of difference. One example of this relates to examining attitudes towards meat and vegetables amongst vegetarians and non-vegetarians whereby the IRAP found support for the expected beliefs of the two groups (Barnes-Holmes et al., 2010). Predictive validity on the IRAP, in terms of the extent to which is may predict a particular outcome, has been evidenced in a range of studies. For example, in one study assessing participants’ fear of spiders, the IRAP has been shown to predict participants’ subsequent avoidance of such insects (Nicholson & Barnes-Holmes, 2012).

2.3.8.1.5 Face validity. This explores whether a measure assesses what it aims to. As an implicit measure, the IRAP has been considered by experts in the area of implicit measurement as having face validity (LeBel &

Paunonen, 2011). However, from a participant perspective, it may be viewed as having poor face validity as although they may be aware of the target being measured, the individual may be uncertain regarding how their responses will be assessed (Golijani-Moghaddam et al., 2013).

2.3.8.1.6 Test re-test reliability. Test retest reliability on the IRAP has been reported as r =.49, indicating adequate reliability and stability as the responses have been found to be consistent on two occasions (Cullen et al., 2010).

2.3.8.1.7 Internal consistency. Internal consistency has been reported to range from .23 to .85 (Golijani-Moghaddam et al., 2013). However, generally the reliability of the IRAP is considered limited, this may be due to it being a state rather than a trait measure.

2.3.9 Developing the IRAP. In this study, an IRAP targeting implicit emotional eating was developed on an individual basis; all items were idiographic and based on each participant’s key beliefs identified through a stem completion task in the screening process (informed by a literature search and items on an emotional eating questionnaire). Participants were asked to

180 indicate how much they agreed with each statement and were also asked to generate their own items or belief statements regarding their emotional eating habits, and then rank these in order, from most to least applicable. If the

participant reported eating in response to both positive and negative emotions, then the six most highly rated items for both types of emotions were included in the IRAP. If the participant reported eating more in response to one type of emotion (e.g., negative) then the six most highly rated negative items were input into the IRAP along with the six lowest ranked items for eating in response to the opposite type of emotion. Refer to figure 4 for an example of a participant who presented with a stated belief that they ate more in response to negative emotions than in response to positive emotions.

.