Charisma in the gig economy: a controlled field experiment on the impact of communication channels on performance

The project consists of two studies
Study 1: Our intervention encompasses six treatments: In all treatments, subjects have to work for a fixed time period and fill out a brief survey afterwards. We vary the extent of charisma used in the instructions. We either employ neutral instructions or charismatic instructions. We also vary the communication channels.
Study 2: We execute 6 treatments where subjects are exposed to the instructions of one of the treatments of study 1 and then have to rate the leader. After the subjects rated the leader, they are exposed to the instructions of a different treatment and asked to predict the resulting performance difference.

2021-08-26

2021-09-30

Study 1: The main variable of interest is the output of the workers measured as the number of submitted work. We will compare this output across treatments. In addition, we are interested in the quality of the submitted work which we will also compare across treatments.
Study 2: The main variable of interest is the perception of charisma measured by a standard survey measure. We will compare this measure across treatments. We also compare the level of perceived charisma with the actual performance and predictions.

The project consists of two studies
Study 1: We vary the extent of charisma used in the instructions. We either employ neutral instructions or charismatic instructions. We also vary the communication channels.
Study 2: Subjects are exposed to the instructions of one of the treatments of study 1 and then have to rate the leader. After the subjects rated the leader, they are exposed to the instructions of a different treatment and asked to predict the resulting performance difference.
We run our experiment on Mturk. The HIT will be posted on Mturk and contain a short description of the task.

We run our experiment on Mturk. The HIT will be posted on Mturk and contain a short description of the task. If subjects accept the HIT, they will get a link to the experiment which will be hosted on Qualtrics. Workers that are located in the U.S., have at least 500 Hits, and an approval rate of 98% are eligible to sign up. As workers are only allowed to participate once, we take measures to prevent multiple participation. We impose these restrictions to ensure a high data quality. To avoid confounding effects if subjects cheat and do not read the instructions properly, we use an attention check and also record the time subjects spend working on the study. In particular, we implement one attention check question in our brief survey (the question is: Paying attention to this survey is important. Please tick “a little.”). If a subject fails the attention check, the dataset will be excluded from the analysis. In addition, we will exclude subjects that speed through the instructions at an unreasonably short time.

Randomization is done by using the randomization function build in Qualtrics.

Study 1: We will execute the treatments for each communication channel in different sessions but randomize within each session if subjects are exposed to neutral or charismatic instructions.
Study 2: We will randomize within session between all treatments.

No

1524 subjects

1524 subjects

We plan to collect 200 observations for each treatment in study 1 and 50 for each treatment in study 2. One observation means one individual that participates in one of the above mentioned treatments. In addition, we need to execute an objective manipulation check with 24 subjects. The total number of observations thus amounts to: 1524 subjects

Study 1: Our main variable of interest is the output delivered by subjects in study 1. Based in previous findings, we expect the effect size to be mild or moderate. Assuming a Cohens d of 0.3 and applying a power level of 0.8 and a significance level of 0.05, we would need 176 observations per treatment for study 1. For study 2, we expect a Cohens d of at least 0.5 to 0.6based in previous literature resulting in 64 to 45 subjects per treatment.

Remote and gig work is prevalent in today’s labor market and calls for skilled digital leaders. Signaling charisma
using charismatic‐leadership‐tactics (CLTs) to increase follower performance works in face‐to‐face communication.
However, technology‐mediated communication reduces the signaling opportunities, thereby
calling the effectiveness of charismatic signaling into question. In Study 1, I conducted a large field experiment
investigating the impact of charismatic signaling (neutral or charisma) and the chosen communication channel
(text, audio, video), on follower performance in the gig economy. Video messages led to lower output than text
or audio communication in the neutral set‐up. In contrast, the output was not significantly different between
the communication channels in the charisma set‐ups. The data revealed a positive interaction between video
communication and CLT presence. The charismatic video led to higher output than the neutral video indicating
that leaders need to deliver a coherent picture, especially when using the video channel. In Study 2, I investigated
if traditional questionnaires (MLQ) measuring perceived charisma and unrelated participants’ forecasts
predict Study 10s outcomes. Although CLT presence led to higher scores in perceived charisma, follower performance
was not predicted by the scores. Thus, the MLQ is not an appropriate instrument for predicting follower
behavior.

Nieken, P. (2022). Charisma in the gig economy: The impact of digital leadership and communication channels on performance, The Leadership Quarterly, 101631

https://doi.org/10.1016/j.leaqua.2022.101631