When the Wait Isn

1
When the Wait Isnt So Bad the Interacting
Effects of Website Delay, Familiarity, and Breadth
MOIS 508 Dr. Dina Rateb
By Dina EL Touby ID 900060001
2
Introduction

• Websites became so popular that almost everyone
  uses them for research online shopping. However
  with all its popularity, it has one major
  drawback, frequent delay when moving from one
  page to the other.
• Delay is considered the most commonly experienced
  problem with the web leading many to call it the
  Worldwide Wait.
• A study was made among 160 undergraduate students
  to show the relation between delay and 2 other
  variables (breadth familiarity) and their
  effects on users performance, attitudes
  behavioral intentions. The results were
  significant.

3
Delay, Familiarity Breadth

• Familiarity site breadth are hypothesized to
  interact with delay and dampen its negative
  effects on users.
• Delay
• Effects
• Tolerable level
• Breadth
• Broad Strategy increasing the number of
  hyperlinks, decreasing the number of clicks page
  loads.
• Deep strategy decreasing the number of
  hyperlinks, increasing the number of clicks page
  loads
• Familiarity
• Familiarity Breadth dampen the effects of Delay
  on two important outcomes users performance
  intentions to return. (behavioral intentions is
  mediated by attitudes)

4
Hypothesis

• Hypothesis 1 Breadth Familiarity will dampen
  the total cost of delay on overall performance of
  a user in a set of search tasks.
• Hypothesis 2 Familiarity Breadth will dampen
  the negative effects of delay on behavioral
  intentions, but the relationship will mediated by
  attitudes

5
Research Methodology

• The study was constructed in an experimental
  setting to control delay, site depth
  familiarity and to allow measurements of outcome
  variables.
• A completely counterbalanced fully factorial
  design sites were constructed.

6
Operationalization of Independent Variables

• Delay it was manipulated with Javascript code to
  provide a constant eight-second delay per page
  for the slow-site and no such delay for the fast
  site.
• Familiarity two artificial websites were
  created.
• The familiar site contained images, process and
  descriptions of products available in everywhere
  to be familiar to the subjects.
• The unfamiliar website contained fictitious
  software products, accessories and their
  categories were meaningless.
• Breadth Websites were created with 2 hierarchal
  structures for 81 bottom-level page.
• Broad Type it was created with 2 levels to
  traverse (9 hyperlinks per page).
• Deep Type it was created with 4 levels to
  traverse (3 hyperlinks per page).

7
Operationalization of Dependent Variables

• Attitudes were measured by averaging the
  responses to a set of seven nine point likert
  type questions.
• Performance it was measured by counting how many
  of the nine search tasks were accomplished. The
  tasks required subjects to visit each third of
  the sites hierarchy the same number of times
  providing a reasonably balanced overall view of
  the site.
• Behavioral Intentions they were measured using
  the average of 2 questions addressing two related
  future behaviors.
• Question 1 How readily the subject would visit
  the site again?
• Question 2 How likely the subject would
  recommend the site for others to visit?

8
Procedure

• The websites were created and their 32
  combinations were written on CDs to control the
  browsers response time.
• Everything in the laboratory was the exact same
  to make sure there are no unwanted errors.
• Participants were asked to complete the required
  9 tasks for each website in order and to answer
  the questions given.
• Answers were assessed as correct or incorrect by
  comparing subjects answers against the facts on
  the site. Data about the actual number of clicks
  were recorded automatically.

9
Results

• After the end of the experiment, the analysis
  began with multivariate ANOVA for both attitudes
  performance and univariate ANOVA for each one
  separately. Tests of mediation of attitudes
  between the experimental factors behavioral
  intentions were also conducted.
• One test was according to Baron kenney
  another was based on PLS.

ANOVA Analysis of Variance.
10
Baron Kenney Results

• The given table shows the results of the
  Multivariate tests of Hypothesis 12. The tests
  reveal that each three way interaction was
  significant.

Overall MANOVA Attitudes Performances
11
Baron Kenney Results Contd

• Means for the dependent variables across the main
  treatments presented in the following tables.
• Each experiment factor had the intended
  expected effects on both dependent variables.

Means across the Delay Treatment
Means across the Familiarity Treatment
Means across the Breadth Treatment
12
PLS Approach

• In the PLS approach, the 3 variables were reduced
  to 1 factor that is called Cost. A code for the
  low-cost was assigned by collecting the
  following conditions together fast, familiar
  broad. Another code was assigned for the
  high-cost for the following conditions slow,
  unfamiliar deep.

PLS model without Attitudes as a Mediator
PLS model with hypothesized Paths
13
Limitations

1. The subjects were college students the results
   might not generalize to the rest of the
   population.
2. The task might not represent typical tasks that
   online shoppers undertake.
3. In real world, it is rare to find an ideally
   modeled site like the one used in the experiment
   when doing causal information search.
4. Analysis was limited by the inability to use
   structural equation modeling to test the complete
   model in one simultaneous analysis.
5. Incentives of the laboratory browsers and home
   shopper might differ. Home shoppers might switch
   their desires if they got frustrated of not
   finding what they want, its not the case with the
   subjects

14
Implications Conclusions

• The study of Delay, Familiarity Breadth factors
  have proven that these three factors should be
  studied together to get the optimum results and
  each one separately is not enough.
• The three factors were effective in predicting
  users performances in, and attitudes about
  navigating a shopping site. Further more,
  attitudes fully mediated the effect of the
  factors on behavioral intentions.

15
Thank You