Tuesday 16th July
Wednesday 17th July
Thursday 18th July
Friday 19th July
Download the conference book
Psychological short scales for survey research - advantages and potential limitations |
|
| Convenor | Professor Beatrice Rammstedt (GESIS Leibniz Institute for the Social Sciences) |
| Coordinator 1 | Dr Constanze Beierlein (GESIS Leibniz Institute for the Social Sciences) |
Psychological variables are becoming more and more attractive for survey researchers who strive at explaining diverse social, economical, and political phenomena. In order to obtain good psychometric quality, the majority of psychological measures contain multiple items for assessing a single construct. However, using multiple items is often not feasible in large-scale social surveys due to limitations in time and money. In addition, the length of scales has been found to be associated with higher rates of attrition and of refusals (Stanton et al., 2002). Thus, researchers have begun to reduce the number of already established scales or to develop new and shorter scales for survey research. We invite researchers from different disciplines to share and discuss their experiences using short scales in surveys. The symposium is open to a wide range of topics in this field of research. For example, presentations may address the development and validation of short scales for specific constructs, different methodological approaches in meeting the challenges of short scales construction, comparisons of short and long scales.
The locus of control scale is intended to measure the extent to which people report to have control over their own achievements rather than be dependent on others or even fate. We examine a shortened version of this scale in a convenience sample of international students (N=170). Besides possible differences regarding the reported degree of locus of control due to true differences between students, we examine possible differences of scores explained by cultural background (students in the test come from all over the world), and the degree of foreign language proficiency (in this case, English).
While implementing a survey, there are two possibilities: 1) to give respondents a sequence of questions and construct a psychometrical scale; 2) to ask one question, i.e. to have only one item for each assessment construct. Results of social surveys and testing depend on the way of questioning but this aspect is poorly investigated. The intention is to present a correlation study. It shows to what extent the results obtained from the same respondents after applying one item approach and psychometrical scales coincide or do not coincide. Measured construct - interests of non-formal learning. At the beginning, using traditional Likert scale, based on one item approach, respondents were asked about their cognitive interests, e.g. in economical, political, legal areas. In the other case, respondents were given a detailed psychometrical scale composed of multiple items e.g. human rights; labor, civil law; etc. The sample of the study (N=729) - mixed population. Number of primary items - 97; number of scales N=10; the length of the scales ranges from 6 to 18 items. Multivariate and one-dimensional regression was applied. Primary items of the scales were defined as independent variables, and one item estimate in Likert scale's form was defined as a dependent variable. Further on, one item estimate was defined as an independent variable and psychometric scale's score was defined as dependent variable. Conclusion - results' coincidence only about 25%. Questioning method has a decisive impact on the results of the survey.
The personality construct Need for Cognition describes the "individual's tendency to engage in and enjoy effortful cognitive endeavors"(Cacioppo & Petty, 1982). It is an important predictor or control variable in various studies in the social sciences, and as such needs to be measured economically.
We present the construction and validation of a German Need for Cognition short scale (NFC-K) based on the German Need for Cognition scale presented by Bless et al. (1994). The NFC-K provides the opportunity to measure the construct with only four items while maintaining objectivity, reliability, and validity.
The original 32-item scale was reduced to four items, on the one hand based on statistical item characteristics as reported in Bless et al. (1994), Cacioppo and Petty (1982) and our own pilot study (N=85) and on the other hand with regards to content to cover the two facets "to engage in and enjoy thinking".
In a set of studies the psychometric properties of the short scale were assessed: The NFC-K correlates highly with the original scale (r= .79; p <.01, N=161), has a satisfying test-retest reliability (r=.78; p<.01, N=73) and shows the expected relations with other psychological constructs such as openness (r=.28, p <.01), achievement motivation (r=.31, p <.01) and social desirability (r=.05, p=.72) as assessed in a third study (N=276). In study four convergent, discriminant and criterion validity will be compared between the NFC-K and the original German scale.