Abstract
Introduction
Experiment
Results
Discussion
Conclusions

Acknowledgements
References
Appendices
Credits
Feedback

Back To Main

Logins & Passwords - What Users Want

Experiment

Introduction

The Internet is quickly becoming an indispensable tool for buying, selling and retrieval of information. With its growth comes the need to personalize information for specific users, and the ability to track their actions. This is usually done in the form of a username and password for each unique user. This study was designed to determine the relationship between simple and complex forms, simple and complex passwords, completion speed, and error rate.

The independent variables for the study are:

Simple forms with minimal help text or Complex forms with abundant help text

Simple or Complex Passwords

The dependent variables for the study are:

Form completion time

Registration and Login error rate

Testers will be asked to fill out one of the four form combinations. They will be timed to see how quickly they can fill out the form and the number of errors they make during the process (in both registration and logging in) will be recorded.

The simple form has minimal help text and all of the fields that a user must fill out are on one screen. The complex form splits up the required field over three screens and has more help text. Help text is descriptive text that instructs a user on how to fill out the form.

The simple password consisted of an easy to remember word and a two-digit number. The following letters and numbers were not used in the simple passwords: 1,l, L, 0,o, O (numeral one, lowercase "ell", uppercase "ell", numeral zero, lowercase "oh", uppercase "oh") due to possible confusion between them. The complex password uses a combination of letters from the first and last name of the user and a number. There were no restrictions as to which letters and numbers could be used in the complex password.

Hypothesis

The null hypothesis for this experiment is that there will be no significant differences in error rate and completion time for the four separate forms.

Our hypothesis is that users will be able to complete the single page registration form quicker than the multi-page registration, but the error rate on the single page registration will be higher. The simpler password will result in quicker login times and less errors than the complex password.

Pilot Study Results

We tested our experiment on four people, one person per treatment. Our materials stayed the same since being submitted, but may be altered for the actual experiment. We feel that we should be able to ask some more specific questions since the actual experiment itself only takes two to three minutes to complete (it is the filling out of a online registration form, the receiving of a username and password - which is almost instantaneous, and then attempting to log in with the username and password . There are many questions on the questionnaire that are subjective, because we would like to try to measure the "frustration factor" when dealing with the different treatments of our experiment.

We feel that the entire experiment will take around 15-20 minutes for each treatment. The breakdowns of the times are as follows:

Introduction and explanation of experiment (3-5 minutes)

Explanation of how to perform the tasks (3-5 minutes)

Filling out of registration information (2 minutes)

Receiving email and logging in (3 minutes)

Filling out of subjective questionnaire (5 minutes)

Subjects

Twenty subjects were used, five for each treatment. The subjects consisted of students of the University of Maryland and workers based in the DC Metropolitan Area. All had at least minimal computer experience. We tried to find a broad range of subjects in terms of computer experience but we found it difficult to find a large group of users that we could convince to participate in the experiment.

Materials

Potential testers were asked to fill out a questionnaire in order for us to gauge their computer experience and were also given a quick explanation of our experiment via an introduction on paper. There were four separate web pages that were created for our experiment and represented the independent variables. An additional page was created to allow users to login with their username and password. A SQL database backend was used to record the user information as well as the registration completion time and the amount of errors that occurred while filling out the registration form and logging in. Testers were then asked to fill out a subjective questionnaire after completing the experiment.

The personal computers used varied depending on the WAM computer used but all of them were Intel machines and not Macintosh computers.

Procedures and Problems

The task list formed a guideline for us to follow, and we followed it as best as possible to try to keep our results organized. It was not difficult finding users who at least had some experience with online registration forms, but the familiarity of the subjects potentially decreased the error rates to the point where it is difficult to make observations about the data.