CHAPTER 4 SURVEY STUDY
4.2 H ARD T RUST VS S OFT T RUST
The analysis of Feature Rankings, Website Layout, Image Size and Attribute Ratings have focused on the amount of “soft trust” associated with individual
components of a website or online store. Two items specifically addressed the specific security features of a website (Security Lock and HTTPS) that are associated with “hard trust”. This set of analysis compares “hard trust” (the specific security features) to “soft trust” (the other components of design). As noted earlier, these are largely independent, but one could expect “hard trust” to be related to “soft trust” because some effort is required to set it up (i.e., only conscientious stores would bother) or because some shoppers misattribute the meaning of HTTPS. Indeed, as noted in the analysis of the Features, some participants specifically mentioned security as a component of trust.
The first analysis examined the association of “soft” and “hard trust” for the expensive and inexpensive items separately. A simple correlation was computed between each of the prior elements (features or design) and security (see Table 4.8). In Table 4.9, the Features are grouped into Security, Design, Social Cues, Content and Navigation. In addition, a (+) or (-) has been added to each feature to indicate that higher scores imply more trust (+) or that higher scores imply less trust (-). This is important for
Table 4.9 The Correlations between “Features” and Security for Expensive Items (n = 122).
Category Feature Security Lock HTTPS
Security Security Lock 1.000 .771***a
HTTPS .771*** 1.000
Privacy Policy (+) .032 .174
Too Much Info (+) .007 .126
Graphic
design Image size (-) Professional (-) -.048 -.094 -.076 -.120
Color scheme (-) .000 .042
Familiarity with store/logo (-) -.046 .063
Familiar Logo (+) -.172 -.134
Clear images (+) -.169 -.137
Bad Color Scheme(-) -.091 -.074
Familiar store (+) .061 .004
Social Cues Website Reviews (-) .006 .089
Friends recommend (+) -.271** -.153
Friend have issues (-) -.017 -.022
Live chat (+) -.234** -.247**
Content
design Clarity of Information (-) Broken links (-) .043 .075 .009 .130
Bad Product Info (-) -.169 -.132
Bad Diction (-) -.043 -.114
Structure
design Ease of Navigation (-) Easy Navigation (+) .063 -.168 -.001-.209*
Bad Menu bar (-) -.100 -.212*
Notes: Bolded Measures are from Analysis 1. Those note bolded are from Analysis 4. a The value is slightly different from that in Table 4.9 because all correlations in Table 4.10 have n=122 (no missing data for any question). ***p < .001, **p<.01, *p<.0
To interpret this, one should first note that none of the correlations are particularly large. Hence, even the strongest relationships are not particularly strong. Secondly, one should note that although the measures of “hard trust” are highly correlated, the measures of “hard trust” with “soft trust” are not. Hence, the two are relatively independent
concepts in the minds of participants. This may be due to true independence or to the effect of random variation. Thirdly, one must be careful of the direction of the coding of
response. For example, for all of the measures in Analysis of the Features, lower scores imply that the item is more important for trust.
There is a negative association between Friends Recommend and security. This implies that those who trust the recommendations of their friends (i.e., recommendations of friend can override personal misgivings) do not look for the Security Lock. However, those who do not trust the recommendations of their friends do tend to look for the Security Lock more often. There is also a negative association between Live Chat and security. In this case, it means that those who are inclined to trust a site because of the presence of a chat room, are not inclined to look for the security features (Security Lock andHTTPS). However, those who are not impressed by the availability of a chat room do check for security. In this case, it seems that there is an either/or type of thinking. Some trust friends and chat room (i.e., Social Cues) while others trust the security features.
Table 4.10 repeats the analysis but for the inexpensive items. Note that the pattern is not quite the same. Firstly, there are more significant correlations. Secondly, those correlations are larger in a relative sense (but not large).Thirdly, some items are correlated in Table 4.10 that were not correlated in Table 4.9.
Table 4.10 The Correlations between “Features” and Security for Inexpensive Items (n = 119).
Category Feature Security Lock HTTPS
Security Security Lock 1.000 .847***
HTTPS .847*** 1.000
Privacy Policy (+) .145 .199
Too Much Info (+) -.006 .050
Design Image size (-) -.059 -.076
Professional (-) -.022 -.043
Color scheme (-) .037 .066
Familiarity with store/logo (-) .080 .121
Familiar Logo (+) -.220* -.184*
Clear images (+) -.164 -.123
Bad Color Scheme(-) -.112 -.036
Familiar store (+) -.007 -.033
Social Cues Website Reviews (-) -.043 -.066
Friends recommend (+) -.387** -.313**
Friend have issues (-) -.214* -.196*
Live chat (+) -.283** -.234**
Content Clarity of Information (-) -.031 -.013
Broken links (-) -.037 .031
Bad Product Info (-) -.309** -.207*
Bad Diction (-) -.065 -.059
Structure Ease of Navigation (-) .073 .038
Easy Navigation (+) -.166 -.118
Bad Menu bar (-) -.088 -.112
Notes: See Table 4.9
As with the previous expensive items, there is a negative correlation between Friends Recommend and security (Security Lock and HTTPS). There is a similar negative correlation for Live Chat. Note that in both cases, the correlations for the inexpensive items are much stronger than those of the expensive items (the proportion of variance explained rises from about 4% (maximum r = .271) to about 15% (maximum r = .387). In addition, there are negative correlations between Friends Have Issues and Security, but this implies that those who listen to the objections of their friends will check security. Those who do not listen to friends do not check security (generally reckless behavior). There is a negative association between Bad Product Info and security (Security Lock and
HTTPS) implying that those who check security also worry about the quality of product information. That is, those who do not check security also do not worry about product presentation (generally reckless behavior). Finally, there was a negative correlation between security and Familiar Logo (i.e., a Visa symbol). Those who check security are also more comfortable with standard logos.