ไม่สามารถเล่นวิดีโอนี้
ขออภัยในความไม่สะดวก
Moderation and Groups Analysis in a Discrete Choice Experiment
ฝัง
- เผยแพร่เมื่อ 12 ธ.ค. 2021
- Here are some tips and tricks for the DCE analysis. I show you how to include different age groups in the analysis (you can use also different industries or countries in a similar way). Then I show the analysis of a combination of items: can it happen that some levels, if they appear jointly, have an impact on the final decision?
Here is the link to my podcast episode on moderation: mci4research.p...
Here is a more complex moderated moderation: mci4research.p...
Feel free to subscribe, like, and share. It does matter!
Thank you for your interesting video. I have the same question with the research group in your video. How can I know which is the best project should I suggest if I have 3 attributes(3 levels for each attribute) or should I make the combination for every two attribute levels then compare? Thank you!
I hope I understand your question correctly: you want two compare two groups of responses, e.g. male vs female participants in their preferences (or whatever other "external" variable you can use). In this event, you just split the file based on your grouping variable. When splitting, you can select what should be done with the groups. Click "compare groups". However, your question can be also understood differently - that you want to create groups from DCE attributes and levels. In this event, I assume that you want to investigate moderation effects of one attribute for another one. If so, don't split the file. Add as you independent variables, e.g., attribute1, attribute2, and their interaction attribute1*attribute2. Pay attention that in order to investigate interaction of two attributes you will need to look at all possible interactions of all levels. It sounds complex but it is simple - just add all possible combinations of levels to analysis: attr1lvl1*attr2lvl1, attr1lvl1*attr2lvl2, ... attrNlvlN*attrKlvlK. If I did not understand your question, you can contact me with an explanation.
Thank you so much for your reply. Sorry for the confusing question. My idea is that I give out several choice sets for the respondents to make a choice. For example, Car 1: black, auto gear, $15k; Car 2: white, auto gear, $20k; Car 3: red, normal gear, $13k and so on. Finally, which set (car) or sets (cars) that I should suggest to my boss to produce or sell? The analysis show which attribute level is preferred. However, the combination of all good may be not good (for the customer or producer). Is it right that I can make the interaction of 3 attributes: (attri1 * attr2 * attri 3) by looking at all possible interactions of all levels of these 3 levels. Thank you for your help and your time.
@@user-dv5me2vf3l There are several things that you can show: 1. Look at the most preferred levels for each attribute and pull them to one object, e.g. red car, automatic gear, 15k. This would be the optimal way to show what are most desired aspects. 2. You can have interactions (if this was your theory). E.g. if the engine is not sufficiently powerful, an air conditionair might be useless. So, there might be interaction between two attributes. In this event, you show two objects: red car, automatic gear change, 45 kWh motor AND red car, mechanical gear change, 35kWh motor. Because of the interaction between attributes, these two objects will be "best" object to present to your boss. 3. You can also show, what is the value for each attribute (look for ma video on the willingness-to-trade calculation). Here you can show what should the final price be dependent on the object properties, e.g. red car, automatic gear change, 45kWh for 12,000EUR OR red car, automatic gear change, 35 kWh for 10,000, and so on. As soon as you have price/value, the number of optimal objects to suggest will grow. If something it less optimal - just sell it with a lower price tag. I hope it helps you present your results better. :)