Which analysis?

#1
Dear people,

I am currently doing a research about the purchase intentions of consumers during online grocery shopping. But I'm now stuck on the way I should analyze. You will mean the world for me right now, if you can help me

For my experiment, I divided the respondents in two groups. The experimental group and the control group, they both got the same questions, but the way how it was presented was different in both of the groups.

To determine the purchase intentions, I divided 5 different factors. I would like to do an analysis where I can see if these factors are affecting differently on the amount of products they selected for the two different groups.

I was thinking to perform a linear regression model for both and see the differences. But maybe it's better to do a form of ANOVA or a conjoint analysis. Maybe even a path analysis. Or is there another sort analysis I miss?
 

Karabiner

TS Contributor
#2
As far as I can see, this is not a problem related to the SPSS software, so I will not specifically
refer to that software in my answer (or questions).
The experimental group and the control group, they both got the same questions, but the way how it was presented was different in both of the groups.
How do the questions and the answering format look like?
What is your dependent variable, precisely?
How large is your sample size?
To determine the purchase intentions, I divided 5 different factors.
I don't know what this means. Could you describe this in more detail?
if these factors are affecting differently on the amount of products they selected for the two different groups.
Now I don't know where "amount of products" comes from. Is it the depenedent variable?
How was it measured?

Maybe you should start once more, and describe your study in a more comprehensive way (at least the research
questions, recruitment, allocation to groups, experimental manipulations, variables included and how they were
measured, main outcome, sample size).

With kind regards

Karabiner
 
#3
As far as I can see, this is not a problem related to the SPSS software, so I will not specifically
refer to that software in my answer (or questions).

How do the questions and the answering format look like?
What is your dependent variable, precisely?
How large is your sample size?

I don't know what this means. Could you describe this in more detail?

Now I don't know where "amount of products" comes from. Is it the depenedent variable?
How was it measured?

Maybe you should start once more, and describe your study in a more comprehensive way (at least the research
questions, recruitment, allocation to groups, experimental manipulations, variables included and how they were
measured, main outcome, sample size).

With kind regards

Karabiner
Thanks for your reply Karabiner. I will give an as structured answer.

I'm doing research if product bundling will influence the purchase intentions of consumers and especially their purchase intentions on healthy food. For this, I spreaded 2 different surveys to 2 different groups. Group A (N=175) can select the products and will not see the productbundles (only products) and group B (N=164) will see the products and thereby also the product bundles.

After selecting the groceries(products), both groups will receive the same questions. In total 20 7 point likert scale questions. 10 of them are meant for my first hypothesis, which is that Group B will select more products (I simplify the hypothesis).

The other 10 questions are there for the other 2 hypotheses (both 5 questions). The second hypothesis will measure the impact of product vividness on the purchase intentions of the consumers and the third hypotheses will measure the impact of the cognitive load of the consumers on their purchase intentions.

The likert scales are thus the independent variables and the dependent variables is the amount of selected products (or the total spendings during the grocery trip).

I hope this will be enough information to answer my question.

Best Regards
Sander Persons
 

Karabiner

TS Contributor
#4
I am not sure about the Likert scales here. Likert scale is the name of a measurement instrument
which is composed out of several Likert-type items. Do you intend to sum up the 10 (or 5 or
5, respectively) Likert-type items so that you will finally work with 3 Likert scale scores?

I am also not totally sure about your research questions. You had 2 different groups, but seemingly
you do not want to include "group" nto the statistical analysis, is this correct? And there
seems to be a question about the respective impact of product vividness and cognitive load,
which you want to compare?

With kind regards

Karabiner
 
#5
Again thanks for your quick response. It means a lot for me.

I will try to specify it more. To make it more simple, let's mostly focus on the first hypothesis (the other two I got an idea how to tackle them. So let's forget the other two hypotheses. The hypotheses I want to get answers on is:

"Product bundling willl positively influence consumers' purchase intention for healthy food".

In the attachment I put the two different groups (they have more products). Group A, you see the first product has products bundled in a product bundle. Group B will see only loose products. The participants were asked to select products they will choose when shopping for different shopping situations (this is for breakfast). A pretest decided which products can be assumed to be healthy or not. So I got a clear dividing of which products are healthy and which are not. This made an option that I have scales on how many healthy products, unhealthy products, percentage of it etc. are selected.

Then, I asked multiple 7-point likert scale, which tests the purchase intentions on 5 five different factors (divided in 10 different questions, all via a 7 likert scale method).

I was wondering how I should analyze it, with also a declaration of the different factors. Right now my idea was via the following steps:
1. Normal t-test --> to see a difference between selecting healthy food and the two different groups.
2. Linear regression for both groups seperately and groups combined --> To see if there is a difference in the variables which declares the variance of the purchase intentions for healthy food.
3. Analysis of variance for a complete 2 x 2 x 2 factorial design --> To measure all the variables which were relevant out of the linear regression to measure how they influence it.

I'm doubting if this is the best approach and was wondering if someone with more knowledge can guide me.

I hope this gives enough insight, let me know if you need more information.
And I say it again, but really thank you.
 

Attachments

Karabiner

TS Contributor
#6
2. Linear regression for both groups seperately and groups combined --> To see if there is a difference in the variables which declares the variance of the purchase intentions for healthy food.
Just perform 1 linear regression analysis in which you include the predictors, the group indicator, and the interaction between predictors and the group indicator. The interaction represents the different impact of a predictor in dfferent groups.

With kind regards

Karabiner
 
Last edited: