Charles Chung

What's the issue

Nowadays, mobile device users receive lots of notifications generated by instant-messaging (IM) apps everyday. However, users also display differing degrees of receptivity across different message types. In addition to contexts such as the presence of alerts, or whether the user has recently interacted with his/her phone, “who” has sent the message may also play a role.

There has been little systematic investigation of how the relationship characteristics such as sender’s relationship type, closeness, mobile maintenance expectation, etc, along with interruption context, affect smartphone users’ receptivity to instant messages. The four receptivity measures we consider in this work are: Attentiveness, Responsiveness, Interruptibility, and Opportuneness Moment.

How we conduct the study

In order to study the impact of sender-recipient relationship characteristics on instant-message recipients’ receptivity to mobile IM messages, we focus on a subset of characteristics of a relationship between two people that are measurable via self-reports. These includes Closeness, Dependence, Mobile Maintenance Expectation (MME), Answering Expectation (AE) and Perceived Obligation to Answering (POA).

Our general research question is: When activity context, social context, phone-interaction context, and ringer mode are taken into consideration, to what extent does each of them affect the four measures of receptivity?

We use five sources to obtain Relationship Characteristics, Interruption Context, and Receptivity: existing scales, self-report from the experience sampling method (ESM), sensor data from the phone, message logs, and post-study semistructured interviews.

We recruited 34 participants, among them, 31 participants had agreed to participate in the post-study semistructured interviews. 4,500 ESM responses are collected in this experiment. To study the impact of relationship characteristics on receptivity, we built three mixed-effect logistic-regression models – a Context Model, a Relationship Model, and a Combined Model – for each receptivity measure. As to the qualitative data analysis, we implemented Cued Retrospective Technique to conduct a semi-structured interview for participants. Then, a mixture of Grounded Theory and Evaluational Research was utilized to analyze the data.

What we find

Our results could be listed as below:

1. Interruption context generally overshadows relationship characteristics in predicting the four measures of receptivity


2. It demonstrates that a single closeness question such as Inclusion of the Other in the Self Scale (IOS) or a newly devised Simple Closeness Measurement (SCM) was as effective at predicting receptivity as the 12-item Unidimensional Relationship Closeness Scale (URCS).


3. Some variables have an effect across four Receptivity Measures.
1. Mobile Maintenance Expectation (MME) and Activity Engagement both had negative main effects on all four receptivity measures.
2. Phone Interaction Context had a positive effect on all measures except responsiveness. This is consistent with prior findings that recent phone interaction is a good indicator of the user’s recent attention on the phone.
3. Session Within had a positive effect on all measures except interruptibility.
4. Obligation to Answering (POA) and Dependence had no effect on any aspect of receptivity.
4. Each receptivity measures has its own set of predictors, showing conceptual differences among these measures.
1. Interruptibility: Uniquely predicted by whether Social Contacts are present
2. Opportune Moment: Uniquely negatively correlated with Silent Ringer Mode. 
3. Attentiveness: It was the only measure predicted by Relationship (Strong-tie).
4. Responsiveness: Answering Expectation is its unique predictor.

What's my role in the project

I was in charge of conducting the user research experiment, interviews, and qualitative data analysis. I was also highly engaged in discussions throughout the quantitative data analysis.

Our following work, which is in progress now, aims to predict receptivity variables and relationship characteristics. We also include qualitative analysis to find more insights. I'm in charge of the qualitative analysis.

One of the findings so far is that people tend to respond less instantly toward some of the most intimate partners if they consider the relationship to be steady and solid, while they tend to reply more quickly toward someone less intimate, due to the obligation to answer, or in order to maintain the relationship.