09.12.2012 Views

2003 IMTA Proceedings - International Military Testing Association

2003 IMTA Proceedings - International Military Testing Association

2003 IMTA Proceedings - International Military Testing Association

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

of personality. Second, the same responses would be used to generate both an effectiveness score<br />

and personality scores. Examinees who try to maximize their personality scores would risk<br />

lowering their effectiveness scores. That is, examinees cannot ignore the effectiveness of any<br />

actions when answering test items. Thus, an SJT that produces both an effectiveness score and<br />

personality scale scores might be able to eliminate—or considerably reduce—faking.<br />

Several personality traits were identified that are relevant to performing well as a Soldier<br />

in the U.S. Army (Sager & Russell, <strong>2003</strong>). We had developed a situational judgment test in<br />

recent previous research for the Army (Knapp et al., 2002). The SJT score correlated<br />

significantly not only with performance measures but also with several personality scales.<br />

Unfortunately, attempts to develop scales based on item scores were unsuccessful. This was not<br />

surprising considering that SJTs tend to be heterogeneous even at the item level. Thus, in the<br />

current approach, we are developing personality scales based on the scores for individual actions.<br />

This effort has two notable aspects. First, we generated parallel civilian situations from<br />

military situations. Second, the test simultaneously measures both the respondent’s judgment and<br />

several of the respondent’s personality traits. We developed descriptions of military situations<br />

that a Soldier might experience during the first few months in the Army. For each situation, we<br />

developed response options comprising actions that Soldiers might take in those situations. We<br />

feared, however, that a test consisting of military situations might not be appropriate for civilian<br />

applicants. Many applicants might not understand the situations, and those that had some<br />

military knowledge might have an unfair advantage on the test. Therefore, we developed a<br />

parallel civilian situation for most military situations. The remainder of this paper provides more<br />

detailed descriptions of the development of civilian items and trait scales. The results of a recent<br />

pilot test will also be discussed.<br />

DEVELOPMENT OF THE MILITARY AND CIVILIAN SJTS<br />

To develop the military SJT, we asked new Soldiers and the NCOs (Non-Commissioned<br />

Officers) who train them to write descriptions of problem situations relevant to new Soldiers<br />

(during the first few months in the Army). These could be actual situations they had observed or<br />

hypothetical situations. Other Soldiers and NCOs wrote descriptions of actions that Soldiers<br />

might take in these situations. We edited and dropped actions until there were no more than<br />

about nine actions per situation. We asked Soldiers and NCOs to write each situation targeted to<br />

one of the specific predictor constructs we wanted the SJT to measure.<br />

At this point, we asked NCOs and Soldiers to write parallel civilian situations based on<br />

the military situations. After editing these situations, we picked the best parallel civilian situation<br />

for each military situation. Then we asked NCOs and Soldiers to write actions for the civilian<br />

situations. We edited these actions and reduced the number of actions per situation to about nine.<br />

We developed an initial scoring key using 10 NCOs (drill instructors and other trainers of<br />

new Soldiers). Then we gave the draft test items (military and civilian items) to Soldiers in<br />

training. Each item was answered by about 12 Soldiers. Based on these two data collections, some<br />

options were dropped and a few others were edited or added. An option was dropped if the NCOs<br />

disagreed substantially on an option’s effectiveness. Typically, options with standard deviations<br />

above 2.00 (on a 7-point scale) were dropped. In contrast, options were also dropped if there was<br />

541<br />

45 th Annual Conference of the <strong>International</strong> <strong>Military</strong> <strong>Testing</strong> <strong>Association</strong><br />

Pensacola, Florida, 3-6 November <strong>2003</strong>

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!