This study by Seiter and Gass was titled The Effect of Patriotic Messages on Restaurant Tipping in it the authors compared the effectiveness of 2 types of patriotic messages with a “warmth/ingratiation” message and a control condition on restaurant tipping. The independent variable in this study was placing one of four messages on customers’ checks: “Have a Nice Day,” “God Bless America,” “United We Stand,” or no message. The servers were instructed to write one of four messages on the back of the checks, representing four experimental conditions: a “warmth/ingratiation” condition, two patriotic conditions, and a control condition.
In the “warmth/ingratiation condition”, servers placed a happy-face sticker and wrote “Have a Nice Day. ” In the next two conditions, the servers placed an American flag sticker, which was about the same size as the happy face sticker and wrote one of two patriotic messages, one or the other “God Bless America” or “United We Stand. ” In the control condition, the servers included no messages or stickers on customers’ checks. There were 25 parties in each condition. While serving customers, the servers were blind to which of the messages each party would receive until just before presenting the check.
This ensured that the particular message given did not affect the quality of the service beforehand. Instead, servers were assigned randomly a sequence of numbers, 1 through 4, corresponding to each of the experimental conditions. Upon totaling each check, servers referred to the next number in the sequence and left the corresponding message. The dependent variable of this study was the tip size as a percentage of the total bill. This was calculated for each party by dividing the amount of the tip by the amount of the total bill (before taxes) and multiplying by 100.
The scale of measurement could be ratio. A ratio scale is a scale of measurement in which there is an absolute zero point, signifying an absence of the variable being measured. It is possible that a customer left a tip of zero. It could also be considered on the interval scale of measure. The interval scale is a scale of measurement in which the intervals between numbers on the scale are all equal in size. Since the variable was measured in percentage and percentage is equally distant from each other. For example 2% is twice as much as 1% and 4% is twice of 2%.
I personally lean towards considering it on the ratio scale. As an alternative to proposing a hypothesis the authors proposed a research question. They asked “Do female food servers who use patriotic messages earn significantly higher tips than those who do not? ” They chose not to make a concrete prediction in a hypothesis form due to the fact that their data was collected during the semester immediately following the terrorist attacks, so they suspected that patriotic statements might be persuasive.
Another reason they chose not to make a concrete prediction was for the reason that of inconsistency in previous research on tipping behaviors. Since there is a question being asked there is no real prediction being made, but it seems that they are orientated towards the patriotic messages earning more based on the phrasing of the question and the time period they were in. Potential issues with this experiments design is construct validity; the degree a measurement device accurately measures the theoretical construct it is designed in order to measure.
They are fairly measuring the tipping amount but there could be other factors to why a person tips a certain amount that has nothing to do with the message left on the check. They have no type of measurement for this. This is called internal validity, the certainty with which results of an experiment can be attributed to the manipulation of the independent variable rather than a confounding variable. People may tip based on the price of the check, how much they can afford to spend overall, their perception of the servers attitude, the customers general attitude toward tipping.
There could be endless reasons to the amount a customer tips. The next issue with the design of this experiment is external validity; the degree to which the results of an experiment may be generalized. I feel that these results can only be generalized to circumstances that are after significant national events such as terrorist attacks or some type of national tragedy. Also these finding only apply to female servers because no males where used. And lastly it only applies to casual dining establishments in the medium price range. The next issue with this design is that there was no baseline taken.
Before the experiment was completed the researchers could have compared items along the lines of if a server works better with smaller or larger groups and their average of tips. Also servers could be acting differently simply because they are being watched. I would suggest that the researchers complete the experiment during a time where people are not being flooded by messages of patriotism. I and the researchers agree that the time this experiment was completed in most likely shifted the results toward the patriotic conditions.
I would also suggest a pretest for the amount the server is being tipped before the experiment and note if there is change. I would then do a posttest where the restaurant sends out an email survey about the quality of service the customer received and ask questions on how they decided to tip. Threats to internal validity are hard to prevent. While redoing this experiment you cannot prevent historic events or personal factors of the customers from affecting the results but I do think that you can account for them in some way like surveying feelings of patriotism or how they relate to a job like a server.
I feel that the reliability in the experiment was great in the follow up experiment they could keep the way the dependent variable is measured . They should then open up the experiment to male servers. Also they could test different levels of dining from casual to fine dining and something in between. I feel the results may be different between the three levels. The follow up study could continue to have a factorial design; an experiment with more than one independent variable level. The follow up experiment could also use the Pretest-Posttest design like I suggested earlier.
The first step would be to obtain four servers; 2 male and 2 female. Examine the tips they receive based on their regular serving techniques, this would be the pretest step. The next step is to introduce the independent variable. Instead of the way the random assignment was given, have the computer print one of the four messages and the corresponding picture of a flag or smiley face instead of the server writing it. Then measure the tips in the same way and compare the findings between independent variable levels and the pretest results to see if it was an increase.