Image Size vs. Scrolling in Photo Thumbnail Browsers

Discussion

Performance Time

The results supported our hypothesis that performance time would increase significantly as the collection size got larger in both the fixed and variable treatments. We also discovered that the performance time in the collection of 24 thumbnails was not significantly different for the variable treatment (small thumbnails with no scrolling) and the fixed treatment (larger thumbnails with scrolling). This supported our hypothesis that in small collections, the variable sized thumbnails would be large enough to see clearly and the fixed sized thumbnails would not require too much scrolling, so task time would be about the same. We also discovered that the performance time in the collection of 48 thumbnails was not significantly different for the variable and fixed treatments. We also predicted this result, because the extra time for scrolling in the fixed treatment was about the same as the extra time to identify smaller thumbnails in the variable treatment. We hypothesized that in the collection of size 36, subjects would be faster in the variable treatment, and we got a significant result in support of our hypothesis.

Errors

The results supported our hypothesis that the number of errors would increase significantly as the collection size got larger in both the fixed and variable treatments. However, we saw no significant difference in the number of errors between fixed and variable sized thumbnails for any of the collection sizes. This result supported our hypothesis that there probably wouldn't be much difference in errors between the fixed and variable treatments for collections of size 24 and 36. However, this result did not support our hypothesis that subjects would make significantly more errors with the variable treatment in collections of size 48. This could be a consequence of the fact that very few errors were made overall, and/or that the large amount of scrolling required for the fixed treatment with 48 photos made searching just as difficult as looking at small thumbnails in the variable treatment.

Subjective Satisfaction

We hypothesized that subjects would significantly prefer the variable treatment to the fixed treatment for the collection of size 24. However, the results showed no statistically significant preference. As mentioned in Section 2.5, some of the subjects did not realize they were going to be asked about each collection size until after the initial question about collections of size 24. As a result, the responses for this collection size are probably not as accurate as the others. For collections of size 36, we hypothesized that subjects would significantly prefer the variable treatment. This hypothesis was not supported. In fact, subjects significantly preferred larger thumbnails, even with scrolling. This result is interesting because the performance time was significantly faster in the variable treatment. It would be interesting to see if this preference changed after subjects had more experience with the browser. For collections of size 48, we hypothesized that subjects would significantly prefer the fixed treatment. Our results supported this hypothesis, likely because the thumbnails in the variable treatment were much more difficult to see than in the fixed treatment.

Department of Computer Science: Direct questions and comments to the student editorial team