Supplemental Materials
Leader Affective Presence and Innovation in Teams
by H. P. Madrid et al., 2016, Journal of Applied Psychology
http://dx.doi.org/10.1037/apl0000078
In addition to the analyses presented in the paper, we conducted a series of
multilevel confirmatory analyses, using data from Study 1, to determine the divergent and
convergent validity of team leader affective presence. Affective presence is a novel
construct in personality research in general and in organizational psychology in particular;
thus, testing the validity of affective presence relative to other similar constructs is relevant
for the conceptual and empirical robustness of this construct. Leader affective presence is
defined as the positive or negative feelings that leaders consistently elicit in their team
members. Regarding divergent validity, leader affective presence should be related to, but
different than, team member work-related affect, because team members’ affect might be
explained in part by leader affective presence and in part by affective traits of team
members and work events unfolding within teams. Similarly, leader affective presence
should be related to, but different than, leader work-related affect. In addition, in terms of
convergent validity, team member ratings of leader affective presence should be
substantially correlated to leaders’ self-reported affective presence. If the above criteria for
divergent and convergent validity are satisfied, strong validity would be inferred for the
leader affective presence construct.
In the team member survey, team member work-related affect was measured with
three items for positive affect (enthusiastic, joyful, inspired; α = .89) and three items for
negative affect (nervous, anxious, tense; α = .85) (Madrid & Patterson, 2014; Warr, Bindl,
Parker, and Inceoglu, 2013). The question frame for the affect measures was “during the
last month working in your team, how often have you felt…? [1: never/almost never – 5:
always/almost always]. In the team leader survey, team leaders rated their work-related
affect, using the same measures adopted with team members (positive affect α = .71;
negative affect α = .61). Furthermore, leaders rated their own affective presence using the
same measures utilized with team members (happy, enthusiastic and inspired for positive
affective presence (α = .73); stressed, tense and worried for negative affective presence (α =
.70)). In this case, the measure was framed as “when interacting with the members of your
team, to what extent do you usually make them feel…? [1: not at all – 5: a great extent].”
We conducted the multilevel confirmatory factor analyses by fixing the first loading
of each factor equal to 1.0 (marker variable) and evaluating the goodness-of-fit of the
models tested with model chi-squared (χ2) and approximate fit indices (i.e., RMSEA ≤ .05,
SRMR ≤ .08, CFI ≥ .95 and TLI ≥ .95) (Kline, 2011). Results of those analyses were as
follows. First, a four-factor model in which team members’ ratings of their leader’s positive
and negative affective presence were included alongside team member positive and
negative affect showed a very good goodness-of-fit (χ2 = 236.88, df = 105, p = .67;
RMSEA = .06; SRMR = .05; CFI = .95; TLI = .94; latent correlation between positive
affective presence and positive affect ρ = .75, and between negative affective presence and
negative affect ρ = .73). In this model, residual variance of an item of positive affect was
constrained to zero, because it showed a negative value at the team level – a Heywood case
– which is typically observed when estimating multilevel confirmatory factor analyses with
a small number of observations at the upper level (Byrne, 2012). These results supported
team leader affective presence and team member affect as related but distinct constructs.
Second, a four-factor model that included self-ratings of positive and negative
leader affective presence and self-ratings of positive and negative leader affect showed
excellent goodness-of-fit (χ2 = 48.73, df = 48, p = .95; RMSEA = .01; SRMR = .00; CFI =
.99; TLI = .99; latent correlation between positive affective presence and positive affect ρ =
.82, and between negative affective presence and negative affect ρ = .44). Therefore, leader
affective presence and leader affect were supported as related but distinct constructs. Taken
together, the above results are supportive of divergent validity for leader affective presence.
Third, the analyses showed positive latent correlations between ratings of positive
affective presence provided by team members and team leaders (ρ = .59), and between
ratings of negative affective presence provided by team members and team leaders (ρ =
.70). This provides evidence for convergent validity, suggesting that team members and
team leaders rated leader affective presence in a congruent way.
References
Madrid, H. P., & Patterson, M. G. (2014). Measuring affect at work based on the valence
and arousal circumplex model. Spanish Journal of Psychology, 17, 1–12.
doi:10.1017/sjp.2014.54
Warr, P., Bindl, U., Parker, S. K., &Inceoglu, I. (2013). Four-quadrant investigation of jobrelated affects and behaviours. European Journal of Work and Organizational
Psychology, 23(3), 342-363.doi:10.1080/1359432X.2012.744449