Van Westendorp Model: Open-ended or not?

[smarties]

Hi,

While there is a lot of info about vW here, I couldn't find the answer to this specific question. I know in theory the 4 price questions should be unaided. However, I am concerned about the quality of the data we will collect (online survey). The current price of the product is $10. Typos might be a problem (respondents typing in $100 rather than $10) but also respondents who just type in anything like $1,000. With a relatively small sample size (about n=500) we can't afford cleaning out too many respondents.

We could ask the respondent to confirm any amount over $X which might help with typos but wouldn't prevent anyone from legitimately entering $1000.

I would prefer a drop-down list of prices (e.g. $0.99 to $99) to avoid the typo issue but again we would have to set a maximum value which is wrong?

What are you thoughts on that?

[Rico]

actually i usually use it closed ended.
I put 15 price point with the one i want itn the middle, and i leave constant intervals between points.

[Statman]

The vW model is base on an unaided design and yes, outliers will exist so they must be taken into account.

Rico...

I put 15 price point with the one i want itn the middle, and i leave constant intervals between points.

This isn't the vW model it is laddering, or a variate, and applying it to the vW is inappropriate.

[Carlos Michelsen]

I agree with Statman. One of the most important outputs of the PSM is the ability to determine where psychological price barriers exist, as well as their relative size. This can only be done if ALL prices are left available.

Furthermore, if you close prices, then the "too cheap" will most times be the first price, and the too expensive the last price, which mean the optimum price points tends to appear around the center of the "scale".

[jjjjherm]

I agree with Carlos and Statman.

I would just add that you should put caps or logic in place.

e.g., If you are testing cereals, you should have a max of US$19.99 and a min of US$0.00. Consumers likely won't be paying more than that or less than that and you don't want to have outliers that are 10 Standard Deviations out.

Also by including logic, you will stop consumers from placing prices for the "too expensive" scale from being lower vs. the "too cheap" scale. Those respondents would typically get thrown out.

Just some thoughts. Good luck with your project.

[Rico]

J, ia m a bit ocnfused, u agree with Carlos and statman, but u do put caps.
although i specify middle price points, but i basically do so keeing the caps in front of my eyes for the same reason, i don't want illogical things.
i don't get the central point as the optimum price in a lot of cases.

[Sherman Whipple]

Actually, Jon, outliers are desirable, so there is no risk or need for caps, particularly if you use qualifiers such as "Do you buy ____? or "When did you last purchase _____? for respondents both as a filter as well as a context.

Rico, when you are testing price sensitivity your objective is not a specific amount but the curve which indicates range and elasticity of the market for a given product or serrvice. It is the shape that is important. Price fluctuates constantly but not the attributes or curve.

[Rico]

dears
impossible for me to argue to much in front of all u experts.
i am chaning the way i am using it as of this moment

[smarties]

I am back with data! I followed your advice, completely open-ended. The data is really revealing so far! But I have one question.

About 10% said $0 is "too expensive" which doesn't surprise me as you can get this product for free as well. So what do I do with them? Do I keep them in the analysis with $0 throughout as "too expensive" has to be the highest amount of the 4 questions?

Or I could say 10% would not pay for your product at all. The rest is willing to pay between $X and $X.

[Statman]

You could go either way but if you do exclude them from the analysis be sure to footnote accordingly. I tend to keep them in since this offers a more inclusive band. (Of course, this all depends of how many $0, particularly if much higher than 10%)

[smarties]

That's a good point Statman. I think I am going to keep them in because if I remove them it increases the lower and upper bound. While those people are saying 0$ now, I think may still buy it if the price is low enough.

The other problem I am running into is the horizontal axis. In Excel, the scale has to be the same for all four lines (they are all sharing the same horizontal axis). So I recoded my data into $0.25 increment. The problem is that it's still not entirely consistent.

There are some price points that have no data particularly the outliers between say $20 and $50. My horizontal axis should not skip those data point, right? It should 0$ to $50 in $0.25 increments even if there are no $22.75, $23.25, $24.50?

[Statman]

I usually compress the scale to focus on the point of intersection and then out a bit to show the appropriate band lines