Statistical analysis of reading experiences.

Introduction (1)

Our initial data gathering experiment revolved around the theory of the Forgetting Curve by Hermann Ebbinghaus. In 1885 he published his work Über das Gedächtnis in which he had been trying to prove the decline of memory retention over time. He did an experiment with made up words which had no prior meaning or connection for him that he tried to memorize. Even though his study was incomplete and limited he is seen as the person that discovered the forgetting curve. 06forgettingcurveLearningCurve_Formula

Experiment (1)

We decided to run an experiment where we were to see whether text, images or sound would be easier to memorized. Which could be used for design purposes in for example educational projects. So our three stimuli were the following:

  1. Text – Participants will be shown a list of 15 words that the participants can memorize for 1 They will be asked to reproduce what they have memorized at certain intervals.
  2. Image – Participants will be shown a list of 15 images that the participants can memorize for 1 They will be asked to reproduce what they have memorized at certain intervals.
  3. Sound – Participants will be requested to listen to 15 different sounds which they can memorize for 1 They will be asked to reproduce what they have memorized at certain intervals.

The intervals used for this experiment are immediately (1), 20 minutes (2), 1 hour (3), 9 hours (4) and 1 day (5). These intervals have also been mentioned by Ebbinghausen. The number of participants for this initial experiment should be at least higher than 10 (n => 10), though we realize that this number is too low as well for us to see clear trends. What it is we measure in  our experiment is the memory retention, so the amount of correctly remembered stimuli over time.

Data analysis (1)

First it was attempted to fit model and data together. Different distributions were fitted resulting eventually in a Poisson distribution which seemed the best fit and it also brought LLC and AIC closest together and on their possible lowest point. This was combined with a Rank-Order spline transformation. Resulting in the following model fit, nevertheless due to the limited amount of participants and parameters the actual fit of the model could have been better.


 Even though in the below depicted graphs the trend of the forgetting curve is somewhat visible it is not as strong as with the original experiment. It also flattens out fairly soon, indicating that people are not forgetting anything more over the given time. It is somewhat visible that Words are forgotten easier than for example the images or sounds. Nevertheless the differences are small and the overall decrease of memory retention s hardly visible.

 Thurstone_Averages_WordsThurstone_Averages_ImagesThurstone_Averages_Sounds  Words                                    Images                                  Sounds

Conclusion and Discussion (1)

The gathered data wouldn’t really allow us to properly conclude anything. We could see the desired trend but it was not that obviously present. The short time in between the intervals probably allowed our participants to rehearse the memorization of the stimuli. Also the amount of words, images and sounds had been too low possibly in comparison with the limited set of participants. All of this left us a couple of choices, we could extend the experiment and find more participants and increase the time in between the intervals. This however would take us quite some time before we would eventually see results and have our data ready for analysis. Also if the memory retention did not decrease any further no actual results would show up at all allowing us no proper practice with statistical data analysis in Illmo. Also it was hard  for us to generate a proper link between our data gathering and the use in design practice. Therefore we concluded to do another different experiment to generate new data more suitable to fit a design related cause.

Introduction (2)

Digital reading, as can be done on a tablet is a trend for the future. However currently the physical book is still preferred for reading in comparison to digital reading devices. This basically means that physical books contain certain qualities which people that read, genuinely appreciate. This is the reason why our concept named Flip was made. It is a device for digital reading which incorporates some of the physical properties of original books to enrich the reading experience.

Client & Design challenge

The (fictional) client for our design challenge is Asus. Asus creates multimedia tablets which could also be used for reading. Nevertheless they have the assumption that there still exists some sort of aversion against digital tablet reading. Therefore they want us to design a modified tablet that allows for an enriched digital reading experience. The challenge is than twofold, to see what is being especially appreciated in physical reading and how it can be translated to a digital device.


Research question:

How to design for an enriched experience of digital media using the perception of physical properties as the means of differentiating content?

Hypothesis 1:

Reading of physical books is still preferred in comparison to reading on digital devices.

Hypothesis 2:

The addition of physical properties to the interaction of digital reading will result in an enriched reading experience in comparison to existing digital reading devices.

Experiment (2)

To verify our hypothesis and get an answer to the research question, we did an experiment with 20 different participants (n = 20). They were requested to try and assess three different possible ways of reading: the original physical book, a tablet and our concept Flip. After doing so they we asked them to fill in a questionnaire on which they could value the three different stimuli on a Likert Scale.

With this experiment we first of all wanted to find out if reading physical books is still the preferred way in comparison to the currently existing digital reading devices such as the tablet. With our concept we wanted to see whether we could enrich the experience of the tablet by involving some of the physical characteristics of original books. The result of our experiment would be a recommendation for further design development and an indication whether Flip is a step in the right direction of an enriched reading experience.



Data analysis (2)

The first step was to see and explore how well our data fitted the model and if not try to make it more fitting. After trying out different distributions the Student T distribution made both LLC and AIC decrease. Then we tried both the Spline and Box-Cox transformation which had varying outcomes. The Spline transformation did not show any improvement. A Box-Cox transformation of the overall power of 2,3763 proved to decrease both AIC and LLC as well resulting in the data and model that we would work with for this analysis.

Unaltered_Data_HistoryLLC_AIC  Student_T_Data_HistoryLLC_AICStudent_T_BoxCox_Data_HistoryLLC_AIC

Cumulative_Model_Data_Unaltered Cumulative_Model_Data_Box-Cox

After we had the data fit the model as well as possible we decided to have a look at the Thurstone averages and do an ANOVA analysis of the three different stimuli based upon the participants’ reading experience. Even though a ANOVA test is for a normal distribution, which is very much alike the Student-T distribution, it was still usable to see the actual differences in averages. In our case it was not about the actual values but about seeing the differences.

Participants were requested to indicate how they rated their reading experience using the specific media. When reviewing their answers we found that the original book is still preferred for reading, especially in comparison to the tablet. Therefore we could see an indication that our initial assumption that people preferably read physical books rather than digital books was not rejected. Furthermore, we found that by means of adding the physical interaction upon digital devices, it should enrich the reading experience to as close as possible to that with original books.


It also shows that in comparison to digital reading upon a tablet the physical interaction with Flip is enriching the reading experience. It remains to be seen however what causes this increase in reading experience and whether there is a correlation between the characteristics of Flip and the reading experience.

We have used the Thurstone difference to see which attributes of Flip are most appreciated by our participants in comparison with the original book and tablet. This way we could also verify which attributes need more looking after and ought to be improved towards a next iteration. We set “book” as the reference and compared the difference between the book and the other two stimuli.

From the general interaction, we can see that Flip’s general interaction is higher than the tablet, it means Flip successfully improved the interaction of traditional digital reading. However, it still needs improvement. Specifically, we need to improve more on the interaction of opening the book to make it closer to physical reading , even though some improvement has yet been made.

Interation (General)Flipping a pageOpening the book

Interation (General)_2Flipping a page_2Opening the book_2Interaction (General)              Flipping the page                  Opening the book

When we look at how well the flipping of the page is valued it is almost halfway in between the physical book and its digital competitor. Therefore the actual interaction of flipping a digital page seems successful but should be further explored as this concept just incorporates one option. The opening of the book (digital file) did seem to be an improvement and got closer to the original book experience. On average we could say that our concept Flip is halfway in between the original physical book and the tablet.



Feel                                        Texture

There are some elements that need more looking after. We asked people how they value the feel and the texture of Flip. In this specific prototype no special attention was given to textural differentiation of digital reading content. However people did like the feel of the prototype in comparison to the more book-like physical interaction. Nevertheless this is an area in which more definitely more improvement can be made.

PracticalityWeightShift in Weight

Practicality_2Weight_2Shift in Weight_2Practicality                             Weight                                   Shift in weight

Also there are some elements which are found to already be better on the digital devices rather than with a physical book. These are for example the weight and practicality, which comes probably from the fact that on a tablet you would be able to bring along multiple books at once which is harder with actual physical books. The shift in weight (center of gravity) is also perceived equally as the physical book possibly indicating that the motion used in the interaction of Flip is perceived positively in relation to the physical books.

Conclusion and Discussion (2)

When reviewing our first hypothesis on the preference of reading physical books rather than digital books on a tablet we can say our experiment indicates this is true. Nevertheless an increased number of participants (n > 20) would provide more certainty on whether it really is.

Then on to our second hypothesis whether physical properties could be used to enrich the digital reading experience. Flip seems to be a step in the right direction in terms of enriching the digital reading experience, as could be seen from the data gathered throughout our experiments.

When reviewing the Thurstone averages of the different data attributes it was found that especially those characteristics related to interaction of original books are positively perceived. These are for example the flipping of a page and the shift in weight (center of gravity) when flipping a page through Flip’s physical interaction. These factors mainly contribute to the improvement of digital reading experience. From the Thurstone Difference, we could see that Flip still needs further improvement on its feel and the possibility to involve textural differentiation of content.

There are also a few points of discussion for what we have been doing throughout this experiment. For example the setup of our questionnaire, we  asked participants to mark how they valued each attribute on a Likert scale. What we could have done differently is to add some questions on how they themselves would describe the interaction with the different stimuli. This way it would be possible to gain both qualitative and quantitative insights but furthermore find out what our users genuinely think of the concepts.

This would have possibly also resulted in a broader range of attributes. The small number of attributes(9) made our investigation somewhat limited. Which also limited the conclusions that we could draw from the obtained data. When we tried to cluster the attributes for example,  to find correlation among the different attributes only a single cluster had been formed not indicating a genuine distinction between the different attributes. The pattern which formed from this however also indicated that our concept Flip is almost halfway in between the book and tablet.



In the experiment, we had one conceptual prototype to compare with reading a physical book and a tablet. If we would have added more prototypes, people could experience more ways of new interaction with digital devices.  They could make choices of which one they prefer and  we could get more data to analyze. Nevertheless in terms of a design process this experiment could work as reference material for future comparisons between more conceptual prototypes.

Also, a fairly obvious remark that almost any of the groups struggled with was the number of participants. With a larger number of participants we could have been able to create a better fitting model for our data and have greater clarity in our conclusions allowing us to have more certainty.

Next step in the design process (2)

Explore physical reading interactions

Our experiment indicates that physical reading currently is still preferred over digital reading. Right now we only had a single concept involving some physical characteristics of the original books. As we now have seen that involving the physical characteristics on digital reading can enrich the reading experience. The next step is to see what the true strongpoints of physical books are. Then we could multiple different explorations of conceptual interpretations and translations of these physical characteristics into a digital reading device.

Refine and explore with prototypes

As previously stated the concept used in this experiment only incorporated a few of the characteristics of physical books. We should then explore the possibilities of making the interaction even more book like in our search of enriching the digital reading experience. We could for example explore how and whether to involve the textural feel of digital reading devices. Nevertheless this first experiment could function as a reference model for future explorations. We could verify whether new concepts are a step into the right direction in comparison to Flip or if it actually makes things worse or not.



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s