24.12.2013 Views

UNDERSTANDING CONSUMER INTENTIONS TO ... - ANZMAC

UNDERSTANDING CONSUMER INTENTIONS TO ... - ANZMAC

UNDERSTANDING CONSUMER INTENTIONS TO ... - ANZMAC

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

The study was conducted in a field setting using survey methodology. A web-based<br />

questionnaire was used to collect data. Development of the scales to measure each of the<br />

constructs in the model proceeded through a series of steps, based on the procedures suggested<br />

by Ajzen and Fishbein (1980) and Ajzen (1985, 1991). The questionnaire was then subjected to<br />

two rounds of pretesting on the web prior to sending out the final instrument.<br />

Results<br />

A total of 956 e-mail letters were sent to staff at a New Zealand university and local research<br />

organisations. From this sample pool, 323 useable responses were received, for an effective<br />

response rate of 34%. Before commencing analysis of the theoretical models, all variables were<br />

examined for reliability, validity and normality, and were deemed adequate for further analysis.<br />

Each of the two models presented above was then subjected to path analysis testing, using<br />

AMOS 3.6 software for structural equation modeling. A path analysis approach was used<br />

because all variables were manifest rather than latent. Introducing latent variables in addition to<br />

the manifest variables would have required a larger sample size.<br />

The data did not prove to be a good fit for the proposed TRA model. The best model, achieved<br />

had a Chi square of 160.12, 11df. Although 41% of the variance in carpool behavioural intention<br />

was explained in the model, the root mean square error of .21, p=.00 indicated the poor fit of the<br />

model to the data.<br />

Carpooling intentions were better explained when the data were tested using the TPB model.<br />

However, the model did not fit exactly as hypothesised. The best fitting model, presented in Figure<br />

3, achieved a Chi square of 14.95, 5df, with a rmsea of .08, p=.13.<br />

SRA<br />

PRA<br />

COMPAT<br />

EXPER<br />

INT_NORM<br />

.18<br />

.23<br />

.30<br />

.22<br />

0, 1<br />

err1<br />

.78<br />

.40<br />

ATTITUDE<br />

.13<br />

.20<br />

SUBNORM<br />

.37<br />

0, 1<br />

Err4<br />

.70<br />

.50<br />

INTENT<br />

Hypothesised<br />

relationship (not<br />

significant)<br />

Hypothesised<br />

relationship<br />

(significant at p ≤ .05)<br />

Significant<br />

relationship (p ≤ .05),<br />

not hypothesised<br />

Significant covariance<br />

relationship (p ≤ .05)<br />

EXT_NORM<br />

FAC_RES<br />

SELF_EFF<br />

.40<br />

.19<br />

Err2<br />

0, 1<br />

PB_CONT<br />

.87 0, 1<br />

Err3<br />

.25<br />

.29<br />

Chi square = 14.95<br />

df = 5 p = .01<br />

cmindf = 2.99<br />

rmsea = .08<br />

p of close fit = .13<br />

Figure 3. Carpooling Behaviour as Explained by the Theory of Planned Behaviour

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!