Relevance & Research Question: Online experiments are a promising complement to traditional methods of data collection as they allow, better than surveys, to collect real behavior and, better than laboratory experiments, to reach very diverse samples. Yet, it is unsure if online and lab experiments yield equivalent results and if it is at all necessary to broaden the sample, i.e. if incentivized decisions in typical situations depend on mode, sample, and setting of experimentation.

Methods & Data: We test for a mode effect by comparing prosocial behavior in typical interactive games using thoroughly parallelized lab and online sessions within student subject pools in Leipzig and Munich. We than broaden our online sample applying the same decision interface to the general population of Germany using Forsa’s offline recruited online access panel and a convenience sample in the United States and India using Amazon’s Mechanical Turk.

Results: It turns out that the mode of data collection has no systematic effect. Behavior in lab and online experiments is very similar within both cities, but there are notable systematic differences in students’ prosocial behavior between Leipzig and Munich. Students in Leipzig reveal stronger other-regarding preferences, but the general population is even more benevolent, and participants on Mechanical Turk behave even less prosocial than students in Munich.

Added Value: Behavior, also in incentivized decision situations, strongly depends on the sample under study. This contradicts the common attempt to generalize results from empirically motivated laboratory studies to establish descriptive parameters of human behavior and asks for replications of experimental results with broader samples. Like surveys, experiments can only tell about behavior of the population sampled from. Fortunately, results of online experiments are in line with lab experiments. Therefore, we have a method at hand to examine real behavior of non-student participants.

Relevance & Research Question:

While in the scientific literature in particular market tests, auctions and indirect methods such as conjoint measurement are preferred as a suitable method for determining price-demand functions (Breidert, Hahsler & Reutterer 2006), the direct query of willingness to pay (WTP) is often characterized by a low validity. However, these direct methods are frequently used in practice (Steiner & Hendus 2012). One of the most established - but at the same time - controversial approaches is the Price Sensitivity Meter (PSM) developed by Van-Westendorp (1976) with further extensions (Roll 2010)

Methods & Data:

As part of a multi-stage research approach, the first step was to examine how well the optimum price point (OPP) derived from the classic PSM meets the actual market price (Miller et al. 2011). This is based on 10 price optimization studies (online, n = 276 to 5,000, Germany). In a second step, PSM price points are re-interpreted. The two upper price points ("expensive but acceptable" and "so expensive that I would not buy anymore") are used for estimating the individual willingness to pay (WTP). Thereafter, revenue and profit-maximizing prices are derived. In the third step (experimental design), the suggested modified PSM approach is compared with other direct methods (including incentive-compatible procedures).

Results:

Results for first step show that the OPP derived from the classic PSM is problematic for price setting: Typically, recommended prices are lower than the market price. When using the modified PSM approach, optimal prices levels can be derived that are significantly closer to the market price. Advantageously, no additional data is necessary to validate PSM results, since price points are only interpreted differently. Both, the modified PSM approach and the direct open WTP-queries lead to similar recommendations for price setting.

Added Value:

Due to a changed market environment (VUCA, volatility, uncertainty, complexity and ambiguity) there is an increasing demand for simple and short questions, which can be used to determine the consumers´ WTP as well as optimal price levels. The proposed method can overcome limitations of the original PSM approach: it meets the needs of simplicity, speed, agility and research economics.