3 RESEARCH DESIGN
3.1 Research Object
This study selects the Internet brand "boss direct
employment network" as the experimental material.
"Boss direct employment" is a recruitment tool for
employers and job seekers to chat and communicate
directly under the brand. Its platform covers
mainstream cultural media, Internet companies,
finance, real estate and other industries. According to
the video advertising characteristics of the "boss
direct employment" brand adopted in this study, the
following criteria are followed when selecting the
respondents: the respondents themselves or their
relatives and friends may need to apply for a job now
or in the future, pay attention to and understand the
job information, so as to ensure the accuracy and
authenticity of the collected data.
3.2 Questionnaire Design and Data
Collection
This questionnaire is divided into seven parts: the
first part is the basic information of the sample. The
second part is about the evaluation of consumers'
attitude before advertising, which is measured by
Algesheim (2005) brand familiarity scale. The third
part is the broadcast of online video ads. the
respondents need to watch a 15s video ad in the
questionnaire. The video ads embedded in volume A
are brainwashing ads, and volume B are non-
brainwashing ads. The fourth part is about the
evaluation of consumers' cognitive response in the
advertising process, namely, advertising cognitive
response and brand cognitive response. The fifth part
is about the evaluation of consumers' post advertising
attitude, namely advertising attitude and brand
attitude. The sixth part is about the evaluation of
consumer behavior intention, namely purchase
intention. The above three parts all refer to the scale
of CAI Peier (2016) which studies the brand
persuasion effect under Internet video patch
advertising. The seventh part is the evaluation of
consumer's time personality, which adopts the Likert
seven scale proposed by usunier.
225 questionnaires were collected in the pre-test,
and the Cronbach's alpha less than 0.5 was deleted. A
total of 700 questionnaires were collected and 584 of
them were valid, including 289 in volume a and 295
in volume B. the effective rate was 83.4%.
4 RESULTS & DISCUSSION
The Cronbach's α value of the overall reliability of the
questionnaire is 0.929, and the α value of each
variable of the questionnaire is greater than 0.8. The
kmo values of volume a and volume B were 0.933
and 0.922 respectively. In Bartlett spherical test, the
significance probability was 0.000, which met the
condition of factor analysis. Principal component
analysis and Kaiser standardized orthogonal rotation
method were used for factor analysis. Four common
factors were extracted by principal component
analysis. The cumulative contribution rates of
common factors were 72.567% and 68.134%
respectively. The common factors extracted could
explain the variables better. The number of factors
after orthogonal rotation is less than the number of
variables assumed in the model, but the factor load is
all greater than 0.4, almost no double load situation,
indicating that the scale has good construction
validity. In addition, the model x2 / DF = 3.114, GFI
= 0.942, RMSEA = 0.086, NFI = 0.876, CFI = 0.912,
RFI = 0.859, TLI = 0.900, PNFI= 0.765, PCFI =
0.800) shows that the model has good fitting effect
and is suitable for subsequent analysis.
4.1 Main Effect Test
The path coefficient of the overall model
standardization is shown in the figure, and H1a, H1B,
H2, H5, H6, H7 and H11 are all verified. Brand
familiarity (brand attitude before advertising) has a
significant positive impact on advertising cognitive
response and brand cognitive response; consumers'
brand cognition and advertising attitude are
dependent on the level of advertising cognitive
response. If the advertising design is complex and
consumers' advertising cognitive response is low,
then consumers' cognition and advertising attitude to
the brand will be poor, which will lead to consumers'
dissatisfaction Brand attitude and purchase intention
also decreased.
Figure 2: Results of testing hypotheses for the total sample.