Less is more?
A research on misleading unit pricing
by
Josse Wester
Less is more?
A research on misleading unit pricing
by
Josse Wester
University of Groningen
Faculty of Economics and Business
Master Thesis Msc. Marketing
Brugstraat 19a
9712AB Groningen
j.wester.9@student.rug.nl
+31(0)654994544
Student number: 2178370
1st Supervisor: Prof. Dr. Ir. Koert van Ittersum
2nd Supervisor: Dr. Y. Joye
MANAGEMENT SUMMARY
With the variety of products available in supermarkets today, efficient and yet still optimal decision-making is more difficult than ever. With the help of price information such as the unit price, the consumer is supposed to be provided with all the tools to make the most optimal decision for every choice. The ability to see very quickly which product is the most advantageous relatively speaking, has grown on the consumer over the decades. This enables the consumer to make decisions based on associations and heuristics, such as cues like packaging and the package size. Not consciously paying attention to all available pricing information and calculating the most optimal decision saves the consumer a lot of time and effort. Consumers nowadays make decision-making as efficient as possible. Though, what does this mean when actual prices do not correlate with what the cues provided by the packaging suggest? Are consumers that easily misled nowadays?
In our study we developed a two-way repeated measures design, where consumers are asked to rate the likelihood-to-buy. This is our dependent variable. Our independent variable is the package size, a small and a large package of the same product. This features as the within-subjects variable size in our repeated measures test. The amount of price information displayed and the math skills of the participant are expected to moderate the relation of package size on the likelihood-to-buy. Consumers are faced with no price information, just the absolute (overall) price, just the unit price, or all price information. Based on the price information, the smaller package is not only cheaper in absolute measures, but the unit price is cheaper as well. This is contrary to the quantity discount heuristic mentioned above, the expectation of consumers that a larger package size is relatively advantageous. Therefore, we expected to see a significant interaction of the price information that is displayed on the likelihood of our package varieties. The variables AbsolutePrice and UnitPrice featured as our between-subjects variables, together with a dummy variable MathSkills. The latter is an operationalization of our education variable, dividing the higher educated participants with the lower educated participants.
consumers preferred the relatively expensive larger bag, whereas the cheaper smaller bag was preferred when the absolute price was shown.
Ultimately, our research concludes that consumers make more optimal decisions when cues that prevent the consumer from quick-and-dirty decision-making are present. In our research, not listing the unit price highly benefited the decisions of participants, and solely focusing on the unit price lead to sub-optimal decisions. This all relates back to the use of the quantity discount heuristic, and the peripheral route of persuasion. If it seems too obvious, the consumer will not consciously invest effort and process the available information, though when a new situation arises or other cues, such as being asked to decide in a research environment, consumers will invest effort and make the most optimal decisions.
PREFACE
TABLE OF CONTENTS
MANAGEMENT SUMMARY ... 3 PREFACE ... 4 INTRODUCTION ... 6 Research Questions ... 7 Agenda ... 8 THEORETICAL FRAMEWORK ... 8 Hypothesis ... 12 RESEARCH DESIGN ... 14 Independent variables ... 14 Dependent variables ... 14 Moderators ... 15 Data Collection ... 15 RESULTS ... 16Repeated Measures: hypothesis 1 ... 17
Repeated Measures: hypothesis 2a ... 18
Repeated Measures: hypothesis 2b ... 19
Repeated Measures: hypothesis 3 ... 20
Repeated Measures: hypothesis 4 ... 20
DISCUSSION ... 20 CONCLUSION ... 22 RECOMMENDATIONS ... 23 REFERENCES ... 25 APPENDIX ... 27 1. Qualtrics research ... 28
2. Descriptive statistics, Cronbach’s Alpha, One-Way ANOVA dependent variables ... 34
3. Q-Q Plots for Normality ... 37
4. Box’s Test of Equality of Covariance Matrices, Levene’s Test of Equality of Error Variances .... 37
5. Repeated Measures output ... 38
INTRODUCTION
With the many comparison websites today, comparing unit prices across stores has become easier than ever before, resulting in more optimal decision capabilities of the consumer. In the 1970s, a similar phenomenon led to increasing optimal decision-making by consumers as well, though not across stores, but in one specific store environment (Russo et al., 1975). With the listing of unit prices alongside absolute product prices, the consumer was enabled to value a product on its relative rather than its absolute value, comparing it on its unit price with competitive products. This increase in transparency increased the mathematical ability of the consumer to make optimal decisions (Mortimer & Weeks, 2013), thus greatly benefiting consumer welfare.
This evolution in the display of pricing information led to an assumption; extra-large packages in the groceries store environment are advantageous, relatively speaking to its smaller peers. Not just because marketers say so, but also because the listed unit pricing indicates this. The absolute value of the larger product is higher, though the price per measurable substance, for instance per kilo or litre, the unit price, is lower. This principle is valid for most products that have a large and a small packaging, for example 4-packs of sodas, kilo bags of onions, and two-litre cans of milk.
As the assumption has been valid for decades, consumers have developed an association between larger packages and a lower unit price (Wansink, 1996; Sprott et al., 1998; Palla et al., 2009). This association has developed into a heuristic (Gigerenzer and Gaissmaier, 2011). As an effect, consumers have saved both financially; by making an optimal decision, and mentally; making a decision between a smaller and a larger product takes less effort when one can just follow a heuristic, instead of thoroughly processing and calculating the available price information.
The main motivation to do this research is the fact that there are companies that use the
quantity discount-heuristic to mislead, by suggesting in-existing unit price advantages. The
product is associated with a relatively advantageous larger package, such as an XL-branding on the package. These cues trigger the heuristic of the consumer, though the actual effect is lacking and the product is actually more expensive in terms of unit price. Where the consumer assumes he or she is buying the most preferred brand and size to optimize unit prices, the misleading marketing actions provide the consumer with an inexistent advantage.
The practical examples that lead to this research are prawn crackers of a Dutch A-brand, in one of the largest Dutch supermarket chains. Where the smaller package is 11.87 euros per kilo, the larger package is 13.25 euros per kilo. However, the larger package carries a large, red XL striping. With these cues, both the package size as the packaging itself, the package is associated with an inexistent advantageous unit price. The product is marketed as the value-for-money big brother of the smaller, but relatively cheaper, package. Therefore, the consumer is misled. To rule out any circumstances that might have biased this pricing for this single setting, research in a competitive supermarket chain was done. This led to the same results: the unit price of the same prawn crackers was lower for the smaller bag, instead of the anticipated larger package, which carries the same big red extra-large branding.
The assumption of the unit price is one that has been the debate of much research, yet results are differentiating. Are consumers aware of the absolute and relative price differences by consciously taking into account unit price information? And if they are misleading and marketers are making use of this heuristic, do they notice? How would preferences be affected if consumers would make an effort and become aware of the misleading marketing actions? In this study the aim is to find out to what extent the consumer is actively considering price information when making a purchase decision and how unit price information influences decision making, and ultimately brand attitudes.
Research Questions
The main question this paper aims to answer is: How does the quantity discount heuristic influence purchase intentions, and how does misleading unit and absolute price information influence this preference?
Agenda
Eventually this paper should lead to new insights into the consciousness level of unit price information, and to what extent heuristics play a role. To do this, the relevant theories developed over the past decades will be analysed in the forthcoming part of this research paper. After discussing existing theories, the research design will be discussed, followed by the results of the research. Finally, the research will be finalized with the conclusions and further recommendations.
THEORETICAL FRAMEWORK
In general, it is assumed that the consumer wants to make an optimal decision and, albeit subconsciously, attempts to achieve this by purchasing the product with the lowest unit price (Isakson and Maurizi, 1973). Therefore, when a consumer takes the displayed unit price information into consideration, he or she will opt for the relatively cheaper product, ceteris paribus. Early research on unit price information (Russo et al., 1975; Russo, 1977) found that unit pricing has a positive effect on the capability of consumers to make optimal decisions. The effect was measured by calculating the average price consumers paid for products, as a negative effect on consumer expenditures indicated more optimal consumer decision-making. However, when a consumer does not consciously consider the displayed unit price information, he or she will assume that the larger package volume is the most advantageous proposition (Raghubir and Krishna, 1999). An even larger package size will on its turn be perceived as cheaper, and thus a more optimal decision for the consumer. This would suggest that misleading the consumer by suggesting a relatively cheaper product by offering a larger package can be effective as long as the consumer does not consciously take the unit price information into consideration and is able to mathematically make the right, in this case the relatively cheapest, inferences out of the unit price information.
considering parts of the provided information such as weight and unit price, he or she is likely to be less able to make an optimal decision.
Not fully processing the available information and not paying conscious attention to make a thoroughly thought-through decision, is related to heuristics. A heuristic is the decrease in effort that is achieved by considering fewer external cues, by not processing every relevant piece of information that is present, and putting less energy into the retrieval of existing cues. Also, complexities are removed from cues, as in not to determine how important individual parts of information are for the decision, considering less informational input and lastly, considering less alternative options (Shah and Oppenheimer, 2008).
Consumers might be motivated to use heuristics and thus have a higher chance of making non-optimal decisions by not fully incorporating all available price information for two reasons (Gigerenzer and Gaissmaier, 2011). The first is that the decision, in this context the purchase of a product in a groceries store, is not of sufficient importance to the consumer to spend the time a thoroughly thought-through decision requires. When financial gains of making an optimal decision are relatively small for a product or the consumer has a specific brand preference, he or she will make a shortcut as the trade-off between the amount of effort and the potential gain that effort can result in is not worth the amount of effort one has to put into that decision. The second reason for heuristics is that the cognitive ability of the consumer is limited. We are limited in the extent to which we can make rational decisions and therefore we depend on heuristics.
Famous research by Petty et al. (1983) explained how consumer’s can make decisions based on positive cues, and makes a quick, low-effort, choice because a presented object is related to a positive association. This is better known as the peripheral route to persuasion (Petty and Cacioppo, 1986). In this context the marketing cues such as the extra large striping is one of these positive cues associated with an advantageous proposition, thus leading to the use of heuristics as the consumer assumes the presented object is favourable, without thoroughly considering the facts of that object.
Ultimately, the consumer performs a continuous trade-off between effort and pay-off, the sufficiency principle (Chen and Chaiken, 1999). The consumer minimizes the cognitive effort that is invested, whilst satisfying instead of maximizing motivational concerns. For the quantity discount heuristic, the motivational concerns of saving money are too limited to invest cognitive effort, resulting in the consumer to rely on heuristics as the decision-making instrument.
Since the consumer is not aware that he or she is being misled, the peripheral route can process the product. When the consumer would be aware of being misled, he or she would likely experience more personal relevance and thus have an increasing felt involvement. Therefore, the consumer will process the product proposition through the central route, consciously checking argument quality and the cons of the proposition. Brand attitudes are more strongly influenced when an object is processed by the central route, thus the realization of being misled when the consumer becomes aware might result in decreasing purchasing intentions, as brand attitudes are greatly influencing purchase intentions.
Thus, whereas the general goal of a consumer is to make an optimal decision, heuristics might influence the consumer. One would say that providing unit price information lowers the chance of heuristics in the decision making process, as it requires less effort for the consumer to make an optimal decision. Therefore, the chance that a non-optimal decision is made should be lower when this information is stated.
found that when the perceived unit costs are lower, the amount of consumption significantly increases for larger packages.
Early research, before unit prices were available for all groceries, found that consumers in a supermarket generally have little knowledge on the relative value of products; unit prices (Granger & Billson, 1972). When unit price information was made available, the economic gain of purchasing lower unit price products became evident and led to significant changes in purchasing behaviour, increasing the consumption of larger packages. Wansink (1996) found that this assumption led to consumer learning behaviour, developing associations between smaller packages and higher unit prices. Therefore, when consumers are subject to a larger size package, they assume the unit price is lower.
Due to the price equals quality heuristic consumers automatically infer that the larger package is of lower quality than the smaller alternative when price information is unknown to the customer, as the consumer assumes the larger packages will be relatively cheaper (Yan, Sengupta and Wyer, 2013). However, when only actual prices are shown, and the consumer does not make any effort to calculate the unit price of the products, he or she can assume that the larger package is of higher quality, as it is more expensive than its smaller equivalent. Though, when the consumer does calculate the unit prices, or more likely makes an estimation that is highly correlated with the consumer’s mathematical abilities, he or she is likely to conclude that the smaller product has a higher quality.
This ultimately concludes in a difference for consumers who focus on overall pricing and those consumers who are relying on unit prices. The latter group should be divided between those cases were the actual unit price is indeed lower for the larger packages, and those were the actual unit price is higher for the larger packages and there is a quantity surcharge. When there is a quantity surcharge and the larger package is actually relatively more expensive, the degree of consciousness is the decisive factor: is the consumer actually calculating, or basing decisions on heuristics.
Though, as aforementioned, consumers do not value the comparison of prices enough to actually do it (Stigler, 1961; Dickson and Sawyer, 1990; Yan, Sengupta and Wyer, 2013). The possible gain is relatively low, given the relative small prices of groceries. Palla, Boutsouki and Zotos (2009) found that this phenomenon of focusing on heuristics is not influenced by living conditions and living costs of areas, suggesting that purchasing power is not significantly impacting the consumer’s psychological shortcut.
Other heuristics closely related to that of the quantity discount heuristic have been subject to much research as well. Raghunatan et al. (2006) found that consumers prefer unhealthy products. The less healthy, the better it tasted. Furthermore, non-healthiness increased product preference and the amount consumed. The so-called Unhealthy=Tasty intuition is based on the same psychological principles of the aforementioned quantity discount heuristic. Even when the heuristic is proven wrong, a consumer will still confirm its personal assumption that an unhealthy product is tastier. Also, it is very hard for consumers to ‘overcome’ this heuristic, even when a product is contradicting with this association. Furthermore, even when consumers disagree unhealthy products are tastier; they still opted for the unhealthy option when they were asked to choose. This principle goes for more heuristics, such as that of the expensiveness of healthy products. Even though the general assumption of consumers is that unhealthy food is cheaper than healthy food (Golan et al., 2008), an extensive research on the price of healthy compared to unhealthy food by Carlson and Frazão (2012) found that, an average portion size of grains, vegetables, fruit and dairy foods costs less than one of foods that has high fat, sugar or sodium levels.
Hypothesis
Existing research shows that consumers have developed a heuristic that associates a larger packaging size with a lower unit price (Raghubir and Krishna, 1999). As the association has been developed over decades, the consumer is now expected to make an assumption, enabling him or her to invest less effort and time into a purchase decision and still make an optimal decision (Petty et al., 1983; Gigerenzer and Gaissmaier, 2011). This focus on a heuristic would suggest that eventhough contradicting unit price information is provided, consumers might still see a larger package as relatively advantageous over its smaller equivalent.
focus on heuristics, thus expecting a lower unit price for the larger package and thus a preference for the larger package. As the consumer gets more information of being misled, he or she is assumed to become more conscious of the decision-making. The realization that the heuristic is actually used to trap the respondent into buying the larger, more expensive, product, is expected to lead to decreased preferences for the assumed-to-be relatively cheaper, but in fact more expensive, larger package.
In general, we assume that the larger package is seen as favorable of its smaller equivalent, as consumers expect it to be the optimal decision due to its lower unit price. Therefore, the following hypothesis is developed:
H1: A larger package size positively influences purchase intentions.
The consumer is assumed to be less lenient to the heuristic when the absolute and unit price of both products is provided. Thus, the customer is expected to be more favorable to the smaller product for those conditions where more price information is shown, and the smaller package not only has a lower absolute, but also a lower unit price. Therefore, the following hypothesis are developed:
H2a: The positive effect of package size on purchase intention is smaller when misleading absolute price information is provided.
H2b: The positive effect of package size on purchase intention is smaller when misleading unit price information is provided.
Also, we expect the unit price information to be most influential for consumers. Whereas absolute price information still leaves room for calculation, and would cost the consumer more effort, unit price information is expected to be the easiest to process and thus to deliver the most optimal decisions. Therefore, we develop the following hypothesis.
H3: The positive effect of package size on purchase intention is smaller for misleading absolute price information than that of misleading unit price information.
misleading price information is provided, we expect those with higher education to be less vaulnerable to the heuristic. Therefore, the following hypothesis is developed:
H4: The positive effect of package size on purchase intention is smaller for people with higher math skills.
RESEARCH DESIGN
Independent variables
Our independent variable for this research is the package size. Package size will be devided by a smaller and a larger package. The smaller package has a weight of 125 grams, and the larger package has a weight of 250 grams. There are four conditions for this independent variable. Across all conditions the participants are shown the weight of the packages, and depending on the condition, price information. The first condition, C1, has no display of price information and is thus solely based on the package, packaging information and consumer’s assumptions. For the second condition, C2, the overall (absolute) price is shown, alongside the information provided in condition one. For the third condition, C3, the unit price of the product is shown as price in euros (€) per 100 grams, alongside the information shown in the first condition. For the last condition, all of the above information is shown. Therefore, C4 shows the weight, the package, the overall price of the product and the unit price.
Dependent variables
The dependent variable is consumer preference. This will be measured by calculating the
likelihood-to-buy, for both the smaller as the larger package size. The main thought behind the
Moderators
There are three moderator in this research. The first two are unit price information and
absolute price information, and the third moderator is math skills. These act as our
between-subject variables for our repeated measures test.
To distinguish between conditions and the effect of the absolute price and unit price, two variables are developed based on our four conditions. These variables, UnitPrice and
AbsolutePrice are either rated 0, not shown, or 1, shown.
The last moderator is consumer’s math skills, as to control for the ability of the consumer to derive correct conclusions from the provided price information. We will operationalize math skills by testing the degree of education. This is assumed to be of great importance, as they highly influence the consumer’s ability to process information and make optimal decisions based upon this information (Lennard et al., 2001). For this, the respondent will be asked to provide the highest level of enjoyed education, ranging from primary school to PhD and higher. These scores are divided in a dummy variable. Based on the data of the education of the participants, a distinction is made within a dummy variable. Participants who enjoyed an HBO education or lower are ranked 1, and those who have been in university (WO) or higher are ranked 2. With this, participants are divided exactly on a 50/50 basis.
Early research suggests that several factors might influence one’s thoroughness of information processing, such as age and gender (Carlson and Geiseke, 1983). Therefore, we include this information into the research. Furthermore, weight and length is asked as to derive the BMI score.
Ultimately, the research design described above leads to the conceptual model in Figure 1:
Figure 1: Conceptual model
Data Collection
As the data is gathered with both within-subjects as between-subjects variables, the analysis will be focused on a two-way Repeated Measures ANOVA. The within-subjects construct will be size, consisting of two variables; the likelihood one will buy the smaller package and the likelihood one will buy the larger package. These will be devided by the between-subjects variables; absoluteprice and unitprice, both consist of 0, not shown, and 1, shown, scores.
These constructs tell whether a participant has been displayed to no price information (0,0, thus C1), the absolute price (1,0, thus C2), the unit price (0,1, thus C3), or both the absolute as the unit price (1,1, thus C4).
To test the hypothesis, an online survey was conducted in which 93 respondents participated. The survey can be found in the Appendix. The data was acquired in May 2015. Participants were recruited online on social media, both in groups that can be defined as the social environment of the researcher, as in the academic environment. After analysing the data, the results of one participant were deleted, as his data formed extreme outliers for most variables. Of the 92 participants that were considered for the rest of this analysis, 52 were men and 40 were woman. Furthermore, 73 participants had a BMI that is considered to be healthy, below 25, and 19 scored above this threshold. 50 per cent of the participants enjoyed HBO or lower education, and the other half is or has been in university. Participants ranged from 18 to 60 years old, with a median of 23. Some preliminary statistics show that the reliability is sufficient, as Cronbach’s Alpha is 0.657. One-way ANOVA tests show that the dependent variables likelihood smaller bag and likelihood larger bag are significantly different across conditions, with respectively p = 0.030 and p = 0.003. Graphs of the statistics above are included in the Appendix.
To successfully perform a repeated measures analysis, several assumptions were to be met. Our dependent variables, the likelihood one purchases a small package and the likelihood one purchases a large package, are both measured on the continuous level, from 0 to 100. Furthermore, the independent variable size, for the within-subjects variable, consists of the same subjects for each construct, the small and the large package. Also, extreme outliers have been deleted from the data. Finally, the data is approximately normal, as can be found in the Q-Q plots shown in the Appendix. Sphericity was not of relevance for this research as there are only two levels of repeated measures in this analysis.
The repeated measures test was conducted with size as the within-subject variable, consisting of two different levels (the variables likelihood smaller bag and likelihood larger bag), and
AbsolutePrice, UnitPrice and MathSkills as the between-subjects variables.
RESULTS
Box’s Test of Equality of Covariance Matrices scores above 0.001 with a p = 0.497, thus
meeting the threshold. A graph of the findings above is available in the Appendix.
The repeated measures test found several significant results for the multivariate test. There was a statistically significant main effect of size with a Wilks’ Lambda value of 0.938, with
F(1,84) = 5.565, p = 0.021. Also, there was a significant interaction effect of AbsolutePrice
on size, with a Wilks’ Lambda value of 0.752, with F(1,84) = 27.769, p = 0.000. Lastly, there was a significant interaction effect of UnitPrice on size, with a Wilks’ Lambda score of 0.951, with F(1,84) = 4.326, p = 0.049. The repeated measures test found no significant results for the MathSkills variable, with a Wilks’ Lambda value of 0.971, with F(1,84) = 2.489, p = 0.118. Also, the three-way interaction of UnitPrice, AbsolutePrice and size was found to be insignificant, with a Wilks’ Lambda value of 0.995, with F(1,84) = 0.388, p = 0.535. A graph of the findings above, and the other output of the repeated measures test, is available in the Appendix.
We find that both the display of an absolute price, as that of a unit price has significant interaction effects on the likelihood-to-buy of a small or a larger sized package, when the displayed price information does not correlate with the quantity discount heuristic, and is thus misleading. Below, we will interpret these results and discover how these interactions influence one another based on both numerical and graphical representations of our findings. With the help of these representations, we can see in what way the significant differences influence the preference for the smaller or a larger package.
Repeated Measures: hypothesis 1
Figure 2: Means of purchasing likelihood with no price information
When no price information is displayed, we find that there is a preference for a larger package, as can be derived from the graphical representation in Figure 2. The mean likelihood-to-buy for the smaller package when no price information is shown is 48.7, whereas the mean likelihood-to-buy for the larger package is 53.4. This is congruent with hypothesis 1, and as we derived earlier that there is a significant main effect for the size variable, with a Wilks’ Lambda value of 0.938, F(1,84) = 5.565, p = 0.021, we can confirm hypothesis 1.
40 45 50 55
Smaller
Package Package Larger No price information visible
Mean likelihood-‐to-‐buy
Repeated Measures: hypothesis 2a
Furthermore, the significant interaction between the absolute price and size shows a clear distinction. When no absolute price is shown, the mean of the likelihood the smaller bag is chosen is 42.5, whereas the likelihood the larger bag is chosen is 50.7, clearly showing a preference for the larger bag. However, when the absolute price is shown, the likelihood the smaller bag is chosen is 52.4 compared to a likelihood of 30.8 that the larger bag is chosen. Figure 3 visualizes this clear interaction effect of the absolute price on the preference of size. As the interaction effect of AbsolutePrice on size is significant, with a Wilks’ Lambda value of 0.752, with F(1,84) = 27.769, p = 0.000, we find that congruent with hypothesis 2a, displaying the misleading absolute prices will indeed result in more optimal decisions by the consumer. Therefore, we confirm hypothesis 2a.
Figure 3: Means of interaction between absolute price information and size
Figure 4: Means of interaction between unit price information and size
0 10 20 30 40 50 60
Small package Large Package Small Package Large Package No absolute price visible Absolute price visible
Mean likelihood-‐to-‐buy
Mean likelihood-‐to-‐buy 0 10 20 30 40 50 60Small package Large Package Small Package Large Package No unit price visible Unit price visible
Mean likelihood-‐to-‐buy
Repeated Measures: hypothesis 2b
Also, the significant interaction between the unit price and size shows an obvious deviation. However, it is the exact opposite of what we expected. When no unit price is shown, the mean of the likelihood the smaller bag is chosen is 53.9, whereas the likelihood the larger bag is chosen is 41.3, clearly showing a preference for the smaller bag. However, when the unit price is shown, the preference is gone. The likelihood the smaller bag is chosen is 41.0, compared to a likelihood of 40.2 that the larger bag is chosen. Figure 4 visualizes this clear interaction effect of the unit price on the likelihood-to-buy of the two sizes. When a - for the consumer - new situation occurs (i.e. there is no unit price data available), he or she becomes more aware of what is actually cheaper, and makes more optimal decisions. Thus, the unit price, which should enable the consumer to very easily see which decision would be the optimal one, in fact leads to making a sub-optimal decision by the consumer. Albeit the significance of the interaction of the unit price on the preference, with a Wilks’ Lambda score of 0.951, with F(1,84) = 4.326, p = 0.049, we reject hypothesis 2b, as we find an opposite effect of displaying the unit price. The data of the interactions of absolute price and unit price on size is available in the Appendix.
Figure 5: Means of interaction between purchasing likelihood and price information per condition
Figure 5 is a profile plot of a repeated measures analysis, which is used to provide further proof for the results mentioned in the hypotheses above. In this analysis, the between-subject variables UnitPrice and AbsolutePrice were replaced by condition. The output of this analysis can be found in the Appendix. It shows a highly significant interaction effect of the condition on size, with a Wilk’s Lambda score of 0.726, with F(3,84) = 10.591, p = 0.000. This clearly visualizes the significant effect unit price and absolute price information has on the likelihood-to-buy for the two package sizes. This supports our findings above, on the interaction effects of absolute price and unit price on size; it shows that the availability of
0 10 20 30 40 50 60 70 No price
information Only absolute price Only unit price information All price
absolute price information resulted in highly optimal decision-making, and it shows that those participants who made a decision based only on unit price information showed a preference for the relatively expensive larger package. Furthermore, it supports the findings for hypothesis 1, with a preference for larger packages when no price information is available.
Repeated Measures: hypothesis 3
The output for the results of our first three hypotheses will act as our input for hypothesis 3. We found a reversed effect of unit price on the decision making of consumers: even when a misleading unit price is shown, consumers opt for the larger, relatively expensive, package. Where the unit price results in sub-optimal decision-making, the absolute price clearly has a positive influence on making the most optimal decision. Given that there is a reversed effect of unit price information on the purchase intentions, we reject hypothesis 3. Providing the misleading unit price proved to be stimulating purchases of the larger package, rather than consumers choosing the relatively more advantageous, and thus most optimal, smaller package.
Repeated Measures: hypothesis 4
Unfortunately, no significant effects were found for the math skills. For the interaction of math skills on size, there is a Wilk’s Lamda score of 0.971, F(1,84) = 2.489, p = 0.118. For the interaction of math skills on absolute price, unit price and size, there is a Wilk’s Lamda score of 0.985, F(1,84) = 1.312, p = 0.255. Based on these findings, we reject hypothesis 4.
DISCUSSION
What we find is that the display of the absolute price makes the smaller, relatively cheaper, package more attractive, even though this cannot be directly inferred from the price information. Contrary, the unit price, clearly showing that the smaller package is actually relatively cheaper, makes the larger package more attractive. What this could suggest, is that why early research found that the display of unit prices made consumers more aware of actual prices, is that this was contrary to existing associations, and providing consumers with a new situation. This results in more conscious attention towards the decision, resulting in the consumer to invest more effort, thus making the consumer more aware of the optimal decision.
decision based on cues (Fennis and Stroebe, 2010) such as the larger product size and the listing of a unit price, as the expected gain for consciously investing effort is none to the consumer. Thus, the consumer makes a decision without consciously calculating the actual optimal decision. When a new situation arrives, the consumer starts to invest effort into making a new, conscious, calculation, and is thus enabled to make better optimal-decisions. In old research, in the early phases of the unit price, the consumer was faced with a situation that is similar to the one in this research. Whereas just providing the absolute price is leading to a new situation for the consumer in this research, the sudden availability of the unit price was a new situation for the consumers in early research. For participants in early research, ordinary decision-making was based on absolute prices, and the related heuristics that were developed. This new situation, with the provision of the unit price, likely led to more attention and effort of the consumer. On its turn, this led to more optimal decision-making by consciously consulting the unit prices, which would explain why early research found significant positive effects for displaying the unit price.
If we relate this to our research, unit prices have been available for decades, and consumers nowadays are used to being provided with all relevant price information. For those who only saw the absolute price, and were thus provided with a new situation, the decision-making was the most optimal across conditions that based their preference on pricing information. This similarity across researches, albeit displaying a reversed effect, would explain why the unit price led to sub-optimal making in our research, but to significantly better decision-making in early research. When provided with a new situation, even though this results in a higher difficulty to calculate the most optimal option for our research, it leads to a higher ability of consumers to choose the relatively advantageous package.
With a mean of 5.26 the price equals quality heuristic scores relatively high, as can be found in the Appendix. Where this might be an explanation why participants opt for the relatively more expensive option when the unit price is shown, it does not make sense that the condition that shows the absolute price, where the larger bag is more expensive as well, consumers prefer the smaller, cheaper, package. It might be a battle between heuristics, though it is more likely that the absolute price leads to a deviation from the normal cues, and thus the normal associations and heuristics, and thus more conscious effort into calculating and determining the most optimal-decision.
the most optimal decision, or a less clear preference for the sub-optimal decision. This might indicate that there is in fact a higher ability of consumers with higher math skills, which correlates with our hypothesis, however statistically it is not highly significant in our repeated measures research.
CONCLUSION
This research proved that consumers are to a high degree lenient on heuristics for their processing of information and decision-making in a supermarket environment. In specific, the quantity discount heuristic proved to be influenced by the degree to which price information was provided. With a higher likelihood-to-buy for a larger package, when no price information was shown, and a vast amount of academic research, the quantity discount clearly is one of the main heuristics consumers focus on whilst in a supermarket. In the following part, we will wrap up this research, and report the main findings that were found.
Consumers nowadays have become used to the unit prices being listed next to absolute prices, as not displaying the unit price leads to a new situation, it is a cue that is new to todays consumer. As this new cue, no unit price, does not appeal to the heuristics of consumers, consumers consciously start investing effort into the decision making, and find that the smaller package is actually relatively cheaper and thus more attractive. This was the main reason why we found that not listing the unit price leads to better decision-making. Presenting consumers with a decision, which they are not able to make based on peripheral cues, thus not enabling the consumer to rely on heuristics, results in a shock effect. This shock effect is followed by processing the decision via the central route, opposed to the peripheral route used for heuristics.
some misleadingly priced products, does not weigh up to the extra effort a consumer would need to invest.
Furthermore, from our last condition we can conclude that when all price information is shown, albeit less clear than the condition with just the absolute price, participants in general make the optimal decision. Therefore, solely the removal of absolute price from the available information results in a high dependence on the heuristic in our research. Whereas this might be perceived as the easiest condition to derive the most optimal decision, this research shows that consumers tend to me more efficient when there actually is information in which they need to consciously invest effort in. When no effort is needed, the information is not regarded at all.
If our research underlines one principle, it is that consumers focus on satisfying their decision efficiency, rather than maximizing it. This principle is named the sufficiency principle (Chen and Chaiken, 1999). For the consumer, it is a continuous battle between pay-off and effort. As the expected gains of an incidental misleading unit price are relatively low, consumers simply opt for heuristics, decreasing their mental effort. Ultimately, even though the consumer is likely to be aware there are cases of quantity surcharging, consumers are more satisfied by disregarding price information, even when it might seem too obvious to be true.
RECOMMENDATIONS
Clearly, this research has some important limitations. The first is the small sample size, which is relatively low, negatively influencing the external validity of this research. With approximately 23 participants per condition, this research cannot be assumed to be completely valid outside the sample population. Also, the sample is gathered for a large part in the near social circle of the researcher, which might bias the results, as it limits the geographic reach of the sample.
Furthermore, the fact that this is in a research context might have led to increasing attention and more conscious processing of information. Participants are aware they are being monitored and are tested on a certain principle. In fact this might implicate that in a natural environment, the lenience on heuristics might be even larger.
Based on this data, we can expect that our sample is a relatively high-educated representation of the entire population, which suggest a higher ability to process the information and make correct inferences based on those calculations. Also, whilst income data is not available, a possible limitation of the high average education is that higher educated consumers generally have a higher average income. With the above characteristics, our sample might be less lenient towards price related heuristics due to their possibly higher financial resources.
REFERENCES
Carlson, A. and Frazão, E. 2012. Are Healthy Foods Really More Expensive? It depends on
How You Measure the Price. U.S. Department of Agriculture, Economic Research Service.
Carlson, J. and Geiseke, R. 1983. Price search in a product market. Journal of Consumer Research, Vol. 9, No. 4, pp. 357-65.
Dickson, P.R. and Sawyer, A.G. 1990. The Price Knowledge and Search of Supermarket
Shoppers. Journal of Marketing, Vol. 54, pp. 42-53.
Fennis, B. M. and Stroebe, W. 2010. The Psychology of Advertising. Hove and New York: Psychology Press.
Gigerenzer, G. and Gaissmaier, W. 2011. Heuristic Decision Making. Annual Review of Psychology, Vol. 62, pp. 451-482.
Golan, E., Stewart, H. and Kuchler, F. 2008. Can Low Income Americans Afford a Healthy
Diet? Amber Waves Vol. 6, pp. 26-33.
Granger, C. W. J., & Billson, A. 1972. Consumers' attitudes toward package size and price. Journal of Marketing Research, Vol. 9, pp. 239–248.
Isakson, H. R. and Maurizi. A. R. 1973. The Consumer Economics of Unit Pricing. Journal of Marketing Research, Vol. 10, pp. 277-85.
Lennard, D., Mitchell, V.W., McGoldrick, P. and Betts, E. 2001. Why consumers under-use
food quantity indicators. The International Review of Retailing, Distribution and Consumer
Research, Vol. 11, No. 2, pp. 177-99.
Mortimer, G. and Weeks, C. S. 2013. The effect of consumer edu- cation on unit price usage
: an early exploratory study. Conference of the European Institute of Retailing and Services
Studies.
Petty, R.E., Cacioppo J.T. and Schumann, D. 1983. Central and Peripheral Routes to
Advertising Effectiveness: The Moderating Role of Involvement. Journal of Consumer
Research, Vol. 10, pp. 135-146.
Petty, R.E. and Cacioppo, J.T. 1986. Communication and Persuasion: Central and
Peripheral Routes to Attitude Change. New York: Springer.
Raghubir, P. and Krishna, A. 1999. Vital dimensions in volume perceptions: Can the eye
Raghunathan, R., Naylor, R.W. and Hoyer, W.D. 2006. The unhealthy=tasty intuition and its
effects on taste inferences, enjoyment, and choice of food products. Journal of Marketing,
Vol. 70, pp. 170-184.
Rao, A. R. 2005. The quality of price as a quality cue. Journal of Marketing Research, Vol. 42, pp. 401–405.
Russo, E.J., Krieser, G. and Miyashita, S. 1975. An Effective Display of Unit Price
Information. Journal of Marketing, Vol. 39, No. 2. pp. 11-19. American Marketing
Association.
Russo, E.J. 1977. The Value of Unit Price Information. Journal of Marketing Research, Vol. 14, No. 2, pp. 193-201.
Shah, A.K. and Oppenheimer D.M. 2008. Heuristics made easy: an effort-reduction
framework. Psychological Bulletin, Vol. 134, pp. 207-222.
Sprott, D.E., Manning, K.C. and Miyazaki, A.D. 2003. Grocery Price Setting and Quantity
Surcharges. The Journal of Marketing, Vol. 67, No. 2, pp. 34-46.
Stigler, G.1961. The economics of information. The Journal of Political Economy, Vol. 69, No. 3, pp. 213-25.
Todd, P.M., and Gigerenzer, G. 2007. Environments that make us smart. Current Directions in Psychological Science, Vol. 16, pp. 167–171.
Wansink, B. 1996. Can package size accelerate usage volume? Journal of Marketing, Vol. 60, pp. 1–14.
Yan, D., Sengupta, J. andWyer Jr., R.S. 2013. Package size and perceived quality: The
intervening role of unit price perceptions. Journal of Consumer Psychology, Vol. 24, pp. 4–
APPENDIX
The appendix features information complementary to the research above. Albeit it is essential information for the research, it is added in the appendix to ensure a high readability of our research. In this appendix you can find information on the research design, the data, and the research output. The bibliography below shows where the relevant information can be found in this appendix.
1. Qualtrics research
2. Descriptive statistics, Cronbach’s Alpha and one-way ANOVA dependent variables
and heuristics
3. Q-Q plots for normality
4. Levene’s Test of Equality of Error Variances, Box’s Test of Equality of Covariance
Matrices
5. Repeated Measures output
6. Repeated Measures output with condition as between-subjects variable
1. Qualtrics research
2. Descriptive statistics, Cronbach’s Alpha, One-Way ANOVA dependent variables Statistics
BMIDummy Gender Age Educat
N Valid 92 92 92 92 Missing 0 0 0 0 Mean 1,2065 1,43 27,34 5,18 Median 1,0000 1,00 23,00 5,50 Std. Deviation ,40703 ,498 10,432 1,167 BMIDummy
Frequency Percent Valid Percent
Cumulative Percent Valid 1,00 73 79,3 79,3 79,3 2,00 19 20,7 20,7 100,0 Total 92 100,0 100,0 Gender
Frequency Percent Valid Percent
Cumulative Percent Valid Men 52 56,5 56,5 56,5 Woman 40 43,5 43,5 100,0 Total 92 100,0 100,0 Educat
Frequency Percent Valid Percent
Descriptives N Mean Std. Deviatio n Std. Error 95% Confidence Interval for Mean
Minim um Maxi mum Lower Bound Upper Bound Likelihood smaller bag 1 23 48,70 22,511 4,694 38,96 58,43 9 100 2 24 57,54 20,876 4,261 48,73 66,36 10 100 3 22 37,41 22,878 4,878 27,27 47,55 3 80 4 23 45,13 24,560 5,121 34,51 55,75 1 91 Total 92 47,41 23,500 2,450 42,55 52,28 1 100 Likelihood larger bag 1 23 53,39 23,461 4,892 43,25 63,54 7 86 2 24 30,79 25,267 5,158 20,12 41,46 0 89 3 22 48,36 27,604 5,885 36,12 60,60 7 100 4 23 30,87 23,744 4,951 20,60 41,14 1 84 Total 92 40,66 26,679 2,782 35,14 46,19 0 100 ANOVA
Sum of Squares df Mean Square F Sig.
2 24 2,92 1,886 ,385 2,12 3,71 1 7 3 22 3,59 2,016 ,430 2,70 4,48 1 7 4 23 3,22 1,808 ,377 2,44 4,00 1 7 Total 92 3,42 1,951 ,203 3,02 3,83 1 7 Price=Quality 1 23 5,35 ,935 ,195 4,94 5,75 4 7 2 24 5,08 1,176 ,240 4,59 5,58 3 7 3 22 5,09 1,109 ,236 4,60 5,58 3 7 4 23 5,52 ,898 ,187 5,13 5,91 4 7 Total 92 5,26 1,036 ,108 5,05 5,48 3 7 Larger=CheaperUni t 1 23 2,39 1,270 ,265 1,84 2,94 1 6 2 24 2,67 1,736 ,354 1,93 3,40 1 6 3 22 2,64 1,364 ,291 2,03 3,24 1 6 4 23 3,09 1,593 ,332 2,40 3,78 1 6 Total 92 2,70 1,503 ,157 2,38 3,01 1 6 ANOVA Sum of Squares df Mean Square F Sig.
Healthy=Expens Between Groups 8,671 3 2,890 1,892 ,137
Within Groups 134,405 88 1,527
Total 143,076 91
Large=Exp Between Groups 15,403 3 5,134 1,365 ,259
Within Groups 331,065 88 3,762
Total 346,467 91
Price=Quality Between Groups 3,131 3 1,044 ,971 ,410
3. Q-Q Plots for Normality
4. Box’s Test of Equality of Covariance Matrices, Levene’s Test of Equality of Error Variances
Box's Test of Equality of Covariance Matricesa
Box's M 22,225
F ,971
df1 21
df2 16266,033
Sig. ,497
Tests the null hypothesis that the observed covariance matrices of the dependent variables are equal across groups.
a. Design: Intercept + UnitPrice + AbsolutPrice + MathSkills + UnitPrice * AbsolutPrice + UnitPrice * MathSkills + AbsolutPrice * MathSkills + UnitPrice * AbsolutPrice * MathSkills
Within Subjects Design: size
Levene's Test of Equality of Error Variancesa
F df1 df2 Sig.
Likelihood
smaller bag 1,682 7 84 ,124
Likelihood larger
bag 1,157 7 84 ,336
Tests the null hypothesis that the error variance of the dependent variable is equal across groups.
a. Design: Intercept + UnitPrice + AbsolutPrice + MathSkills + UnitPrice * AbsolutPrice + UnitPrice * MathSkills +
AbsolutPrice * MathSkills + UnitPrice * AbsolutPrice * MathSkills
Multivariate Testsa Effect Val ue F Hypoth esis df Error df Sig. Partial Eta Square d Noncent . Paramet er
size Pillai's Trace ,06
Hotelling's Trace ,00 5 ,388 b 1,000 84,000 ,535 ,005 ,388 Roy's Largest Root ,00 5 ,388 b 1,000 84,000 ,535 ,005 ,388 size * UnitPrice * MathSkills Pillai's Trace ,01 2 1,036 b 1,000 84,000 ,312 ,012 1,036 Wilks' Lambda ,98 8 1,036 b 1,000 84,000 ,312 ,012 1,036 Hotelling's Trace ,01 2 1,036 b 1,000 84,000 ,312 ,012 1,036 Roy's Largest Root ,01 2 1,036 b 1,000 84,000 ,312 ,012 1,036 size * AbsolutPrice * MathSkills Pillai's Trace ,01 1 ,949 b 1,000 84,000 ,333 ,011 ,949 Wilks' Lambda ,98 9 ,949 b 1,000 84,000 ,333 ,011 ,949 Hotelling's Trace ,01 1 ,949 b 1,000 84,000 ,333 ,011 ,949 Roy's Largest Root ,01 1 ,949 b 1,000 84,000 ,333 ,011 ,949 size * UnitPrice * AbsolutPrice * MathSkills Pillai's Trace ,01 5 1,312 b 1,000 84,000 ,255 ,015 1,312 Wilks' Lambda ,98 5 1,312 b 1,000 84,000 ,255 ,015 1,312 Hotelling's Trace ,01 6 1,312 b 1,000 84,000 ,255 ,015 1,312 Roy's Largest Root ,01 6 1,312 b 1,000 84,000 ,255 ,015 1,312
a. Design: Intercept + UnitPrice + AbsolutPrice + MathSkills + UnitPrice * AbsolutPrice + UnitPrice * MathSkills + AbsolutPrice * MathSkills + UnitPrice * AbsolutPrice * MathSkills Within Subjects Design: size
Tests of Within-Subjects Contrasts
Measure: MEASURE_1
Source size
Type III Sum
of Squares df Mean Square F Sig. Partial Eta Square d Noncent. Parameter size Linear 1968,436 1 1968,436 5,565 ,02 1 ,062 5,565
size * UnitPrice Linear
1530,282 1 1530,282 4,326 ,04
1 ,049 4,326
size * AbsolutPrice Linear
9823,093 1 9823,093 27,769 ,00
0 ,248 27,769
size * MathSkills Linear
880,602 1 880,602 2,489 ,11 8 ,029 2,489 size * UnitPrice * AbsolutPrice Linear 137,101 1 137,101 ,388 ,53 5 ,005 ,388 size * UnitPrice * MathSkills Linear 366,420 1 366,420 1,036 ,31 2 ,012 1,036 size * AbsolutPrice * MathSkills Linear 335,579 1 335,579 ,949 ,33 3 ,011 ,949 size * UnitPrice * AbsolutPrice * MathSkills Linear 463,947 1 463,947 1,312 ,25 5 ,015 1,312 Error(size) Linear 29714,503 84 353,744
Tests of Between-Subjects Effects
Measure: MEASURE_1 Transformed Variable: Average
Source
Type III Sum
of Squares df Mean Square F Sig.
UnitPrice * AbsolutPrice * MathSkills 946,565 1 946,565 1,230 ,271 ,014 1,230 Error 64653,486 84 769,684 7. UnitPrice * size Measure: MEASURE_1
UnitPrice size Mean Std. Error
95% Confidence Interval Lower Bound Upper Bound ,00 1 53,917 3,403 47,151 60,684 2 41,336 3,743 33,893 48,779 1,00 1 41,013 3,393 34,266 47,759 2 40,222 3,732 32,800 47,643 9. AbsolutPrice * size Measure: MEASURE_1
AbsolutPrice size Mean Std. Error
95% Confidence Interval Lower Bound Upper Bound ,00 1 42,495 3,405 35,723 49,266 2 50,744 3,746 43,296 58,193 1,00 1 52,435 3,390 45,694 59,177 2 30,813 3,729 23,398 38,229
11. UnitPrice * AbsolutPrice * Math skills
Measure: MEASURE_1
UnitPrice AbsolutPrice Math skills Mean Std. Error
95% Confidence Interval Lower Bound Upper Bound
12. UnitPrice * AbsolutPrice * size
Measure: MEASURE_1
UnitPrice AbsolutPrice size Mean Std. Error
95% Confidence Interval Lower Bound Upper Bound
,00 ,00 1 47,865 4,742 38,436 57,295 2 51,985 5,216 41,612 62,357 1,00 1 59,969 4,881 50,262 69,676 2 30,688 5,370 20,009 41,366 1,00 ,00 1 37,124 4,888 27,403 46,845 2 49,504 5,377 38,811 60,198 1,00 1 44,902 4,706 35,544 54,259 2 30,939 5,176 20,646 41,233
15. UnitPrice * AbsolutPrice * Math skills * size
Measure: MEASURE_1 UnitPrice
Absolut
Price Math skills size Mean Std. Error
95% Confidence Interval
Lower Bound Upper Bound
6. Repeated Measures output with condition as the between-subjects variable Within-Subjects Factors Measure: MEASURE_1 size Dependent Variable 1 Likelihoodsm allerbag 2 Likelihoodlarg erbag Descriptive Statistics
Math skills Condition Mean Std. Deviation N
Total 1 53,39 23,461 23
2 30,79 25,267 24
3 48,36 27,604 22
4 30,87 23,744 23
Total 40,66 26,679 92
Box's Test of Equality of Covariance Matricesa Box's M 22,225 F ,971 df1 21 df2 16266,033 Sig. ,497
Tests the null hypothesis that the observed
covariance matrices of the dependent variables are equal across groups. a. Design: Intercept + MathSkills + Condition + MathSkills * Condition Within Subjects Design: size
Levene's Test of Equality of Error Variancesa
F df1 df2 Sig.
Likelihood smaller
bag 1,682 7 84 ,124
Likelihood larger
bag 1,157 7 84 ,336
Tests the null hypothesis that the error variance of the dependent variable is equal across groups.
a. Design: Intercept + MathSkills + Condition + MathSkills * Condition
Multivariate Testsa Effect Value F Hypothes is df Error df Sig. Partial Eta Squared Nonce nt. Parame ter
size Pillai's Trace ,062 5,565b 1,000 84,000 ,021 ,062 5,565
Wilks' Lambda ,938 5,565b 1,000 84,000 ,021 ,062 5,565
Hotelling's Trace ,066 5,565b 1,000 84,000 ,021 ,062 5,565
Roy's Largest Root ,066 5,565b 1,000 84,000 ,021 ,062 5,565
size * MathSkills
Pillai's Trace ,029 2,489b 1,000 84,000 ,118 ,029 2,489
Wilks' Lambda ,971 2,489b 1,000 84,000 ,118 ,029 2,489
Hotelling's Trace ,030 2,489b 1,000 84,000 ,118 ,029 2,489
Roy's Largest Root ,030 2,489b 1,000 84,000 ,118 ,029 2,489
size * Condition
Pillai's Trace ,274 10,591b 3,000 84,000 ,000 ,274 31,774
Wilks' Lambda ,726 10,591b 3,000 84,000 ,000 ,274 31,774
Hotelling's Trace ,378 10,591b 3,000 84,000 ,000 ,274 31,774
Roy's Largest Root ,378 10,591b 3,000 84,000 ,000 ,274 31,774
size * MathSkills * Condition Pillai's Trace ,038 1,116b 3,000 84,000 ,347 ,038 3,347 Wilks' Lambda ,962 1,116b 3,000 84,000 ,347 ,038 3,347 Hotelling's Trace ,040 1,116b 3,000 84,000 ,347 ,038 3,347
Roy's Largest Root ,040 1,116b 3,000 84,000 ,347 ,038 3,347
a. Design: Intercept + MathSkills + Condition + MathSkills * Condition Within Subjects Design: size
b. Exact statistic
Tests of Within-Subjects Effects
Measure: MEASURE_1
Source
Type III Sum of
Squares df Mean Square F Sig.
size * Condition Sphericity Assumed 11239,899 3 3746,633 10,591 ,000 ,274 31,774 Greenhouse-Geisser 11239,899 3,000 3746,633 10,591 ,000 ,274 31,774 Huynh-Feldt 11239,899 3,000 3746,633 10,591 ,000 ,274 31,774 Lower-bound 11239,899 3,000 3746,633 10,591 ,000 ,274 31,774 size * MathSkills * Condition Sphericity Assumed 1184,130 3 394,710 1,116 ,347 ,038 3,347 Greenhouse-Geisser 1184,130 3,000 394,710 1,116 ,347 ,038 3,347 Huynh-Feldt 1184,130 3,000 394,710 1,116 ,347 ,038 3,347 Lower-bound 1184,130 3,000 394,710 1,116 ,347 ,038 3,347 Error(size) Sphericity Assumed 29714,503 84 353,744 Greenhouse-Geisser 29714,503 84,00 0 353,744 Huynh-Feldt 29714,503 84,00 0 353,744 Lower-bound 29714,503 84,00 0 353,744
Tests of Within-Subjects Contrasts
Measure: MEASURE_1
Source size
Type III Sum
Tests of Between-Subjects Effects
Measure: MEASURE_1 Transformed Variable: Average
Source
Type III Sum
of Squares df Mean Square F Sig.
Partial Eta Squared Noncent . Paramet er Intercept 342897,099 1 342897,099 445,504 ,000 ,841 445,504 MathSkills 1465,117 1 1465,117 1,904 ,171 ,022 1,904 Condition 3378,146 3 1126,049 1,463 ,231 ,050 4,389 MathSkills * Condition 2696,789 3 898,930 1,168 ,327 ,040 3,504 Error 64653,486 84 769,684 6. Condition * size Measure: MEASURE_1
Condition size Mean Std. Error
95% Confidence Interval Lower Bound Upper Bound
