Chapter 4 Stage One: designing and developing an online Implicit Association Test to measure stereotypes of empathy in scientists
4.3. Step II: SE-IAT Procedural design
After having selected the materials for the SE-IAT, the next step was to design the procedure of how to administer the test. IAT has been used with procedural variations, usually without any intention to collect data to discriminate alternate versions. It is important to consider that measurement procedures are tools and different types of research require different kinds of tools (Gawronski & De Houwer, 2013). There is no procedure as the "best" one, only the most suitable paradigm for particular research aims. The present study followed the most widely used standard IAT procedure recommended by the founder of the test, Nosek et al.(2007), which has shown good psychometric properties in many existing studies. Potential construct-unrelated variables that need extra vigilance are also tackled with various means during the procedure. Different aspects with the procedure design, including the IAT structure, the
number of trials, the inter-trial intervals and the order of compatible and incompatible tasks are explained in detail in the following sections.
4.3.1. Seven-block design
First of all, the SE-IAT adopts the standard 7-block structure design. There are three main categorization tasks in the SE-IAT: single-category classification (Block 1, 2, and 5), compatible configuration of double categorization (Block 3 and 4), and incompatible configuration of double categorization (Block 6 and 7).
The SE-IAT starts by training participants to press the left response key ("E" on keyboard) when a Science item appears on the screen and the right response key ("I" on keyboard) when a Liberal arts item appears on the screen. In the following Block 2, participants are trained to press left for Rationality items and right for Empathy items. The next Blocks 3 and 4 combine both discrimination tasks, making the so-called compatible combined blocks in which items representing Science and Rationality share the same right response key, whereas those representing Liberal Arts and Empathy share the left response key. The following Block 5 is again a single discrimination task and switches the positions of target categories, such that Liberal Arts items are assigned to the left and Science items are assigned to the right this time. The final Blocks 6 and 7 again combines the attribute and the previously reversed target discrimination, making the so-called incompatible combined blocks in which the Liberal arts and
Rationality now share the same right response key and the Science and Empathy items
share the left key. It is worth noting that in both combined blocks, the first set (Block 3 and 6) is for practice and the second set is the actual testing set (Block 4 and 7). Table 4.4 presents a schematic overview of the 7 blocks of the SE-IAT.
Table 4.4 Seven blocks in the SE-IAT
Block N trials Task
Response Key Assignment
Left key (E) Right key (I)
1 20 Target discrimination Science Liberal Arts
2 20 Attribute discrimination Rationality Empathy
3 20 Initial combined task Science,
Rationality
Liberal Arts, Empathy
4 40 Initial combined task Science,
Rationality
Liberal Arts, Empathy
5 40 Reversed target
discrimination Liberal Arts Science
6 20 Reversed combined task Liberal Arts,
Rationality
Science, Empathy
7 40 Reversed combined task Liberal Arts,
Rationality
Science, Empathy
Note. Nosek, Greenwald, and Banaji (2005) recommend increasing the number of trials in the fifth block to 40, in order to combat compatibility-order effect. As such, the SE- IAT has adopted this change and involved 40 instead of 20 trials in this particular block.
For participants who possess a stereotype-congruent belief that empathy is more related with liberal arts than with science, the incompatible combined blocks should be more difficult than the compatible combined blocks. On the contrary, the incompatible combined blocks should be easier for participants who possess a counter-stereotype belief that empathy is more associated with science than with liberal arts. Magnitude of associations is indexed both by the speed of responding (faster responding indicating stronger associations) and the frequency of errors (fewer errors indicating stronger associations).
4.3.2. Number of trials
A trial is defined as the time from the onset of a single stimulus item to the correct categorization of that item (Nosek, Greenwald, et al., 2007). For each trial, participants rapidly classify one item into the corresponding category. Regarding the number of trials in each block, evidence indicates that including 20 trials in Block 1, 2, 3 and 6 (that are blocks for practice) and 40 trials in Block 5 (that is the Block of reversed target discrimination) and critical Block 4 and 7 (that are Blocks of combined tasks for test) yields good psychometric properties (Lane et al., 2007; Nosek, Greenwald, et al., 2007). There is no evidence for any benefit of using more trials. A total number of 200 trials are included in the SE-IAT.
As can be seen from Table 4.4, the number of trials for practice in Block 5 (40 trials) of the subsequent combined blocks doubles the amount of trials for practice in Block 3 (20 trials) of the precedent combined blocks. This is designed with the aim of reducing the undesirable order effect of combined blocks. Considering that the precedent combined task usually interferes with the performance of the subsequent combined task, Nosek et al., (2005) suggested to double the number of trials for practice in the subsequent combined task to provide more time for participants to get prepared for the upcoming new categorization task in the subsequent combined blocks, as such to reduce the order effect.
4.3.3. Intertrial intervals
The interval between occurrence of one trial and the following trial – the intertrial interval – is set as 150 milliseconds in the SE-IAT. Though Greenwald et al., (1998) found no appreciable effect of using longer intertrial interval (750ms), a relatively short interval allows shorter test time, which is important to reduce the fatigue effect (Cohen et al., 2003). In this way, the procedure of the seven blocks can be administered within five minutes (Nosek, Greenwald, et al., 2007).
4.3.4. Order of combined blocks
As mentioned before, one of the common construct-unrelated effects observed on the IAT is the tendency for the precedent combined task to interfere with performance in the subsequent combined task (Klauer, 2005; Lane et al., 2007; Nosek et al., 2005). Specifically, participants who complete the compatible combined blocks first and the incompatible combined blocks later usually showed larger IAT effects than those who complete the combined blocks in a reversed order. Nosek et al. (2005) reported that the well-documented order effects on the IAT could not always be eliminated by extra practice trials in the subsequent combined blocks. Therefore, the order of the two critical combined tasks is also counterbalanced across participants with the purpose of controlling the order effect.
To conclude, the SE-IAT adopted the standard 7-block structure design with two single- categorisation blocks, two compatible combined blocks, one reversed target categorization block, and two incompatible combined blocks. There are 200 trials in total and the intertrial interval is set as 150 milliseconds. The order of the compatible and incompatible combined blocks is counterbalanced for the purpose of controlling the undesired order effect.
4.4. Step III: Creating and testing the SE-IAT website