LPI DATA ANALYSIS August 2008
Barry Z. Posner Santa Clara University
The attached pages provide a psychometric update on the LPI and the conclusion is “that the LPI continues to be a reliable and valid instrument.” In addition, breakdowns of the LPI by demographic variables are provided. The current sample includes data collected through the LPI Online from 2005 – 2007. Demographic information is voluntarily provided (~ 25% response rate).
STRONG PSYCHOMETRIC PROPERTIES RELIABILITY Reliability of the LPI was tested through analysis of internal reliability (Table 1). All five leadership practices had consistently strong internal relaiability coefficients, for both the Self and Observers formats. Cronbach alpha coefficients greater than .70 are generally regarded as very good.
Table 1 Internal Reliability Coefficients (Cronbach alpha) for the Five Practices of Exemplary Leaders
MTW All Respondents (N =651,602) Self Only (N = 48,620) Observers Only (N = 602,982) .84 ISV .91 CTP .86 EOA .86 ETH .91
VALIDITY Analysis of validity was tested in several ways. An “impact” scale was created, using responses from Observers only. Respondents were asked 10 questions (using five-point Likert scales) regarding their feelings of and assessments about team spirit, organizational pride, behavioral commitment, motivation, productivity, clarity of expectations, levels of appreciation, effectiveness, management trust, and efficacy in the workplace. Internal reliability for the
Impact scale was 0.93. The correlations shown in Table 2 between the Impact scale and the five leadership practices were all statistically significant (p < .001). Table 2 Correlations of Impact with Leadership Practices
[LPI-Observer Responses Only]
MTW Impact .28
All correlations are statistically significant (p < .001).
Three approximately equal-sized groups were...