How do users really hold and interact with their mobile devices ?

A key to success, when designing anything that will be used by a human being, is understanding how it will be used. A designer can rely on psychology, physiology, logic, imagination and experience to predict how a user will. Conducting user tests of a service or product provides valuable data to make a concept user friendly. However, initial decisions are often based on assumptions and long debates with clients, in particular when mobile design is involved.

Since I believe understanding behavior implies taking in account the natural contexts in a holistic approach, while cutting up the problem in digest chunks, I decided to conduct a series of humble studies, which hopefully will provide interesting and unexpected insights into mobile user behavior. This article presents the first of those, inspired from Seven Hoober’s study on How people naturally hold  their mobile devices, published on UXmatters.

Part 1 : How do users hold and interact with their mobile devices ?

Steven’s article provided interesting insights for my mobile project at the time, yet it raised more questions. In the original study, conducted in major US cities, 780 users were observed interacting with their mobile devices.

It made me wonder if the position of the user had an impact on how they held their device? Would I have the same findings outside the US ? Why does a user hold his phone in one way or another ? What is the impact of the context, task and in use application ?

As a result, I decided to conduct a series of studies to answers these questions. As a first step of what I shall call “the metro studies” (because studies need a cool name), I started with a similar study here in Paris. This article presents the first results, and how they compare to the original study.

A word about the study design

Since I am going to compare the results of both studies, it is mandatory to compare also their design and data scale. Both studies focused on device holding in public places. This means in both cases, behaviors in private environments and interactions on tablets have not been observed. Globally, the same kind of data was gathered, except for a few, as summarized in the following table.

Steven Cornelia
User action (idle, talking, interacting Yes Interacting only
User position (sitting, standing, walking) Yes Yes
Orientation (Horizontal or vertical ?) Partial : not explicit on all entries Yes
Holding position (1 hand, 2 hands, cradle) Yes Yes
left/ right hand Yes Not recorded
Low or high position Not recorded Yes
Thumb, index Yes Yes
Location Airport, bus, train, commute, station, street, mall, voting line Subway, train, station, commute, street, mall, café
Sample size 780 825
Data date January 2013 December 2013

Also, the proportions of positions in which people were observed differ :

Steven Cornelia
Sitting 24.00% 32.00%
Standing 43.50% 49.50%
Walking 32.50% 18.50%

What this study doesn’t tell you

As Steven pointed out, it should be clear what this data does not tell you before presenting the findings in details.

The data does not take in account what the users were doing, since that would be too intrusive. As a result, positions can not be linked with a task or design principle.
It is not possible to say n% of people do this action, or hold their phone in a specific way since we did not keep track of the total number of people owning a device. In fact, this type of study might only allow you to say what you are likely to see if you look at someone interacting with their phone.

We did also not take in account which devices were used, and why people interact with them the way they do, or which mode of interaction they prefer. Lastly, it does not reflect behaviors in private contexts, touch interactions on tablets and gaming consoles : they are also not part of the data set.

Now that this is cleared, let’s have a look at the results.

Ways of holding devices

Globally, the data gathered in both studies is surprisingly similar, in terms of how many users were observed holding their phones in each way.


Ways of interating with mobile devices depends on the user’s position

Proportions of holding manners observed vary significantly in the same way based on whether the user is standing, sitting or walking. It is difficult to say whether they depend on the user’s behaviors or on the method though, at this point.

HP_cornelia HP_steven


In both cases, it seems it is much more likely to observed cradled holding when sitting, than when standing on walking. Khi² independence tests on subsets of the data shows a relation between the position and observed interactions (hands position) with p<.001. However, calculating Cramer’s V show only (very) weak relations (between .15 and .28).

Sitting Standing Walking
1 hand 34.62% 53.68% 71.05%
Craddled 48.85% 29.41% 17.11%
2 hands 16.54% 16.91% 11.84%

Interacting with fingers or thumb varies depending upon the user’s position

A similar statistical test points out that finger vs. thumb use also changes significantly depending on the user’s position when interacting with the device. Since this information was present only in a small-ish sample of Steven’s observations, I could not verify this on his data.

Sitting Standing Walking
Thumnb 60.27% 81.89% 93.04%
Finger 39.73% 18.11% 6.96%

Khi² independence tests on subsets of the data shows a relation between the position and observed interactions with p<.001. However, calculating Cramer’s V show only (very) weak relations (between .23 and .37)

Device orientation

In terms of orientation, results are similar as well in both studies. While Steven states 90% of users hold their devices vertically. In the raw data, this appears to be calculated on a small subset of data. Out of my 800+ observations, 96.5% of users were holding their devices vertically. the orientation is unrelated to the user’s position.

If anything, the data about orientation questions the importance of designing specifically for horizontal versions of apps. From open, non systematical observations, it seems at first sight that horizontal orientation is used mainly for typing more comfortably (though a majority of typing action is done vertically), and on occasion when it is the default orientation for a game, or when viewing a site that wasn’t optimized for smartphone displays.

Further inquiry about this specific topic would be needed to understand which are the specific situations or contexts in which designing in landscape mode is necessary and important.

And the metro study goes on…

We are still far from answers to all the questions we started with yet. More studies are on the way to document design decisions.

With which hand do users interact with their devices ? How often do users switch from one interaction mode to another ? How long do users stick to one particular interaction mode ? Does the position of the user affect this as well ? The first study will focus on time tracking to observe different interactions on a smaller sample of users, for short periods of time.

How does the task of the user influence the way they hold their phone, if at all ? A small scale user test will focus on providing insights into how the task influences – or not – the way people interact with with a specific smartphone, and whether personal preference is involved in their behavior.

The next batch is coming soon, so stay tuned for the next metro’s humble results. Meanwhile, feel free to discuss the study or ask any details you might find missing or check out the full data set .

2 thoughts on “How do users really hold and interact with their mobile devices ?”

Leave a Reply

Your email address will not be published. Required fields are marked *