Quantitative research methods PSYCHOLOGY Course
Note: Your lab report has a strict limit of max 2000 words. This is just an example of lab report from previous years to help you understand what you need to do, but SHOULD NOT be used as guideline on word count.
The Impact of Framing Effects: How Do Risky Choice and Attribution Frames Influence Responses and Decision-Making? Comment by Janet Bultitude: This person obtained a mark of 72 Comment by Meike Scheller: Okay title - could have been kept a bit shorter/precise ("Responses" is an empty word here as this is implied in "decision making")
Abstract Comment by Janet Bultitude: This is quite a good abstract: most important sections are covered, including introduction and conclusion. There is enough information for the reader to understand what was done and what was found. One way this could be improved is if there was a clear rationale. Comment by Karin Petrini: Also the introduction part of the abstract could have been much shorter thus reducing its length…
The differences in individual decision making based upon the way in which information is framed is a broadly researched field. The various ways in which content can be phrased is referred to as framing effects, and although there are various forms, this study focussed on risky choice and attribute framing specifically. While risky choice frames are supported by more theoretical research, attribute frames have also been shown to have an influence although they are much simpler. In the study, participants were exposed to both a risky choice and attribute frame, and their responses to the two hypothetical scenarios presented was measured. One scenario presented participants with a sure gain or loss versus a risky choice. The other framed information either positively or negatively, and required participants to evaluate their preferences. The differences between responses for both risky choice and attribute framing was found to be significant. These results along with a range of additional literature suggests that framing effects are significant in influencing decision-making. However, the distinct lack of research specifically into the simplicity of attribute framing in comparison to risky choice is noted, and this is recommended as an area for future research. Comment by Meike Scheller: A bit vague and not very informative here Comment by Meike Scheller: an influence on what? Again, a bit unclear this sentence. Comment by Karin Petrini: This is rather vague, what was found significant? Comment by Meike Scheller: watch out for typos Comment by Meike Scheller: Does this actually add anything to this study? Make the distinction clear between this study's results and your idea here.
Introduction
Individuals have been found to differ in their responses to a presented issue when given options which are equal descriptions of the same problem, yet varied in their phrasing (Levin, Schneider & Gaeth, 1998). This difference in judgement is referred to as ‘framing effects’, and an individual’s preferences for one ‘frame’ over another is the main focus of study (Kuhberger, 1998). Within this broad area of research, there are two primary theories which have been developed in order to attempt to explain why individuals have such a preference. Comment by Meike Scheller: Good that you provide your definition of framing and framing effects Comment by Janet Bultitude: Good opening. Defines broad area then brings the focus down to the two specific theories that they will cover. Comment by Karin Petrini: The use of an approachable example here would have further improved this already good opening
Firstly, framing effects have roots within economic theory as there is a focus on the differences between ‘risky choices’ associated with probability and gambling (Mongin, 1997). This is best examined within the framework of utility theory, which suggests that individuals are more likely to seek risk where there is a sure loss, yet avoid risk where there is a sure gain (Kahneman & Tversky, 1979). A situation in which this can be exemplified is in a classic experiment by Tversky and Kahneman (1981) where they presented participants with a hypothetical scenario relating to disease, and asked them to choose between a sure gain or loss, or a gamble. As expected based on the theory, participants opted for risk-aversion when exposed to potential gains, and were risk-seeking when there was a potential for loss. However, Kahneman and Tversky (1979) proposed a revised theory which complements some of the aspects of utility theory, yet focusses on the value of gains or losses. This is called prospect theory, and suggests that individuals edit and subsequently evaluate information based on the frame provided. These two theories provide the basic background for the majority of research on framing effects, yet both focus mainly on risky choice frames and do not consider or account for other frame variations. Comment by Janet Bultitude: Good, succinct definition Comment by Janet Bultitude: Good use of literature. Comment by Meike Scheller: If this is a revised version of the previous theory, I would not provide a reference that is older than the first, non-revised theory for this. Comment by Janet Bultitude: It wasn’t absolutely necessary to explain both utility theory and prospect theory, but this person has done it succinctly and in a way that indicates the breadth of theory in this field. Comment by Janet Bultitude: Good concluding sentence to this paragraph – nice flow to the next paragraph.
In a meta-analysis of framing effects, Levin, Schneider & Gaeth (1998) focus on goal and attribute framing in addition to the standard risky choice frames studied in most of the previous literature. In this paper, goal frames are defined as when a consequence or goal of an action is framed. Attribute framing, on the other hand, is described as the most basic type of frame where something is simply framed either positively or negatively. A notable example of attribute framing can be viewed in a study where participants rated beef as better tasting and less greasy when it was labelled as ‘75% lean’ rather than ‘25% fat’ (Levin & Gaeth, 1988). Levin, Schneider & Gaeth (1998) argue that these types of framing effects are slightly varied from the standard risky choice frames and demonstrate how the theoretical concepts initiated by Kahneman and Tversky (1979) could be extended to wider methods of framing. Comment by Meike Scheller: It is nice that you provide an example here so the (naive) reader can relate better Comment by Janet Bultitude: Here is a rationale, although this idea could have been explained a little more.
Consequently, the purposes of this study was to investigate how both a risky choice and attribute frame would impact participant’s responses to hypothetical scenarios. This is in order to demonstrate that attribute frames, although more basic than risky choice frames (Levin, Schneider & Gaeth, 1998), can still influence participant responses. The risky choice scenario used in this study is identical to the disease scenario previously discussed (Tversky & Kahneman, 1981), whereas the attribute frame scenario is similar to the product study (Levin & Gaeth, 1988), yet focussed on the likelihood of investment. Comment by Janet Bultitude: Narrowing once again. Continuing the ‘upside down triangle’ structure
More specifically, the aim of this study was to investigate whether responses would differ between participants when provided with equivalent outcomes framed differently either using risky choices or attributes. Also, as much of the literature within this field is fairly out-dated, there exists the need to conduct research which investigates framing effects in a more current context. A recent study by Kamoen, Mos & Dekker (2015) has demonstrated the relevance of framing effects in modern situations, as they found hotel reviews are perceived to be more persuasive if they are framed as ‘good’ (positively), rather than ‘not bad’ (negatively). This suggests that it can be predicted that the results of our study will be consistent with both older and more recent findings. Therefore, it can be hypothesised that within the risky choice scenario, more participants will choose to avoid risk when presented with a sure gain, and more participants will choose to seek risk when presented with a sure loss. Similarly, in the attribute scenario, more participants will choose the option framed positively rather than negatively. These hypotheses are based on a broad range of research discussed which ultimately demonstrate the well-supported influence that framing effects have. Comment by Janet Bultitude: Clear aim. Refers to both types of framing. Comment by Janet Bultitude: Further rationale Comment by Janet Bultitude: Recent evidence upon which they can base their hypotheses. Comment by Janet Bultitude: Hypothesis 1 refers to both conditions Comment by Janet Bultitude: Not quite accurate, since it suggests the participants could choose between ‘yes’ and ‘no’ when in fact there were 6 choices. Comment by Janet Bultitude: This sentence is perhaps unnecessary, although it’s nice to have a clear concluding sentence Comment by Meike Scheller [2]: Yes this paragraph could have ended with the hypothesis and the argument about testing framing in the modern or current time could have been introduced earlier…this would be best as a clear and straightforward final/hypothesis paragraph…
Method
Design
An independent measures experiment was used, whereby participants took part in one of two conditions. The independent variable manipulated in this experiment was the phrasing of the situation given to the participant. In the investment scenario, this was framed as “outperformed by 20%” or “outperformed 80%”, whereas in the disease scenario the conditions were either “people die” or “people will be saved”. The dependent variable for the investment scenario was the rating participants gave on a Likert scale between one and six. For the disease scenario, the dependent variable was the choice of either programme A or B. Comment by Janet Bultitude: Brief and clear description of the IVs Comment by Meike Scheller: Good description of variables but it would be good to put them into context a bit better. At this stage, the reader has no clue what the "outperformed..." and "people die/will be saved" relate to. The only becomes clearer in the Materials. Comment by Karin Petrini: Here the type of measurement for the dependent variable should have been specified, the use of likert scale or two programmes do not explain clearly what the DV was…i.e., what was measured exactly. Sometime simplicity is best…
Participants
In group one there were a total of 51 participants, comprised of 41 females and 10 males with a mean age of 19.22 ( SD = 1.67). In the second group, there was a total of 76 participants, including 69 females and 7 males with a mean age of 19.22 ( SD = 3.45). All participants were selected through an opportunity sampling method as all were Psychology undergraduate students present in a lecture theatre.
Materials
The transcripts for each scenario (investment and disease) were obtained from Reisberg, Gleitman & Gleitman (2004). The investment scenario described an opportunity to invest in a new company that has shown promising growth. The element of the scenario that was changed was whether the participants read that the company “outperformed 80% of its competitors” or that it was “outperformed by 20% of competitors”. In the second scenario relating to the spread of a disease expected to kill 600 people, participants were asked to choose a programme to help prevent the disease. Participants were either presented with programmes framed as “people will die” or “people will be saved”. The programme choices entitled A and B were presented with A being a sure loss or gain of lives, and B being a risk for potentially saving or losing lives based on probability. For the full transcripts of the scenarios, see Appendix 1. Comment by Meike Scheller [2]: Good that you indicate the original source of the material (unless you have created it). Comment by Janet Bultitude: Clear, brief description of the scenarios. Should have indicated what the choices of the Likert scale corresponded to.
Procedure
The participants were divided randomly in half by the experimenter depending on where they were sat in the lecture theatre. This formed the two experimental groups. Group one were instructed to read the first scenario displayed on the projector screens, while group two looked away from the screens so they were unable to read any instructions. Once group one had completed and recorded each of their responses to both of the scenarios, they were instructed to then look away from the screens. Group two then repeated the same procedure, only with different instructions featured in the scenarios. There participants also recorded their answers and then both groups were debriefed on the purposes of the experiment and shown the differentiation between scenarios. Comment by Janet Bultitude: Probably should have written ‘both scenarios, one at a time’ Comment by Janet Bultitude: Clear, succinct procedure. No repetition between the methods and procedure section.
Results
In the investment scenario, the “outperformed 80%” condition provided higher ratings (mean = 4.49) than the “outperformed by 20%” condition (mean = 3.96), as shown in Figure 1. Through the use of a Mann-Whitney U statistical test, the higher ratings from the “outperformed by 80%” condition and overall difference between the two groups was found to be significant; U(51, 76) = 1265.0, Z = -3.591, p<0.001. Comment by Janet Bultitude: Good use of descriptive names Comment by Janet Bultitude: Described direction of the effect Comment by Meike Scheller: Description of what these ratings are exactly would have been good (in the Method section) Comment by Janet Bultitude: Gave means within text Comment by Meike Scheller [2]: Needs to also provide some dispersion measure (e.g. standard deviation) Comment by Janet Bultitude: Referred to figure in text. Comment by Janet Bultitude: Accurate reporting of stats, including correct APA format (letters in italics).
Figure 1 . Mean rating on Likert scale given by participants in reponse to the investment scenario. Error bars denote one standard deviation from the mean in each condition. Comment by Janet Bultitude: This is a very good figure. It is helpful that the conditions are named according to their descriptive label rather than ‘group 1’ and ‘group 2’. Both axes are labelled and the Y axis include 0. There are error bars. The Figure caption is below the figure and includes reference to the DV and the scenario that the data refer to, as well as a description of what the error bars represent. Comment by Janet Bultitude: Could have written ‘by participants in the two groups\
Similarly, in the disease scenario, in the “people will be saved condition” more participants (68.6%) chose programme A, while in the “people will die condition” more participants (72.4%) chose programme B. This is further explained in Figure 2. Using a Chi-square (χ2) test for of independence, a significant association was found between the framing condition and choice of programme; χ2(1) = 20.81, p<0.001.
Figure 2. Percentage of responses to either Programme A or B given by participants in each condition of the disease scenario. Comment by Janet Bultitude: This is also a good figure. For this scenario it was better to present the percentages rather than the actual numbers since there were different people in each group.
Discussion
The results show that, in each of the conditions, framing effects created a significant difference between the responses participants gave to the presented hypothetical scenarios. That is, in the investment scenario, participants in the “outperformed 80%” condition rated that they were more likely to invest than those in the “outperformed by 20%” condition. Similarly, in the disease scenario more participants sought the risk shown in programme B when presented with a potential loss in the “people will die” condition. In the “people will be saved condition”, more participants chose risk aversion when presented with a potential gain by choosing programme A. These results are consistent with the hypotheses made based on prospect theory (Kahneman & Tversky, 1979), as well as research supporting the effects of attribute framing (Levin & Gaeth, 1988). Comment by Janet Bultitude: General summary of findings Comment by Janet Bultitude: Specific description of findings Comment by Janet Bultitude: Linking to both the hypotheses and previous literature.
Although much of the literature discussed focussesfocuses widely on prospect theory and standard risky choices frames, this research demonstrates that much simpler frames can still impact participant responses. This highlights one of the main weaknesses within Kahneman & Tversky’s (1979) theory, as it lacks generalisability to other frames. This is significant especially in a modern context, as people are more frequently exposed to attribute frames in their daily lives than the risky choices exemplified in the disease scenario. This is supported by the fact that positive and negative frames have been found to be the most common types used in speech as reference points (Tribushinina, 2008). As such, this resultsthe results of this study are highly relevant in providing support for the significance of attribute framing despite the majority of literature focus being on risky choice frames. Within this study, it is also significant that all participants were exposed to both a risky choice frame and an attribute frame, as the results demonstrate that participants were susceptible to influence from both types. Comment by Meike Scheller: Great to put the study back into context – i.e. what does this mean for the general "bigger picture"? Comment by Janet Bultitude: This idea has originality, and is supported by literature in the next sentence. Comment by Janet Bultitude: This paragraph is a good example of discussing the relevance of the findings. This person could have gone a bit further by stating that if this pattern of similar results for the two framing scenarios were replicated in future studies, then researchers could perhaps apply what has been discovered about risky choice framing to attribution framing, too.
However, the limitations of the study should be discussed. Firstly, the study lacks applicability in relation to how frames might be used in genuine situations, as the scenarios were hypothetical and participants were unlikely to be able to relate to them. While Kamoen, Mos & Dekker (2015) demonstrate the use of scenario that is common and familiar to participants, the investment and disease scenario used in this study are not situations participants would be accustomed to. This should have been considered within this study, as previous research has shown that participant’s familiarity with the content of a scenario can change the impact of the frame (Nelson, Oxley & Clawson, 1997). Also, the overall sample comprised entirely of students used in this study can be criticised. In relation to framing effects, motivational and cognitive differences may exist between students and general samples (Kuhberger, 1998), which limits the generalisability of results from laboratory settings to genuine populations. This should be taken into consideration if, as previously mentioned, the recent application of framing effects to modern contexts is to be fully understood. Comment by Meike Scheller [2]: There is the tendency in students to spend too much text to discuss the limitations of the study rather than the importance and implications of the findings. Though discussing and being aware of the limitations is important, the discussion serves as a means of putting the results back into the bigger picture. This student did a good job in balancing between the two (taking into account the word limit). Comment by Janet Bultitude: This is a good point, and many people raised it. This person has described this point fairly well, in that they have backed up their point with previous literature. They could have gone further by suggesting what the impact could potentially be on the results (i.e. would they expect framing effects to be weaker, or stronger, if the person was instead placed in a genuine situation such as investing their own real money?)
Overall, this study contributes to an extensive range of literature which confirms how framing effects can impact an individual’s decisions and responses. Additionally, although attribute framing is not as widely researched as the standard risky choice frames, this study demonstrates how attribute frames also produce significant results. Along with other recent research (Kamoen, Mos & Dekker, 2015), this study suggests the need for future research into specifically how the simplicity of an attribute frame influences decision-making. While prospect theory and a range of associated literature has comfortably accounted for risky choice frames, there is a distinct lack of research and theory into the reasons why attribute framing works so well despite its simplicity. This would ensure that framing effects could fully be applied and understood in wider contexts more familiar to individuals to assess the specific influence that they have. Comment by Janet Bultitude: This is a very common grammatical error that people make. The correct phrase in this case is ‘literature that’. Microsoft Word’s grammatical autocorrect tool is almost always right about when to use ‘which’ or ‘that’ Comment by Karin Petrini: This could have been clearer, what is meant by ‘how simplicity of an attribute frame’? Comment by Karin Petrini: Extensively? Comfortably is an odd term here Comment by Janet Bultitude: Strong conclusion Comment by Karin Petrini: Mentioning also the ‘modern context and situation’ here to link to what was outlined in the introduction would have make it even stronger.
Word Count: 2125
References
Kahneman, D. & Tversky, A. (1979). Prospect theory: An analysis of decisions under risk. Econometrica, 47(2), 263–291. Comment by Karin Petrini: Attention to use the correct APA style, this reference section is pretty good! However, there are still some inconsistency, e.g., lack of comma after D. in the first reference. Comment by Karin Petrini: Also would be best to start the references in a new page not just after the discussion
Kamoen, N., Mos, M.B.J, & Dekker (Robbin), W.F.S. (2015). A hotel that is not bad isn’t good. The effects of valence framing and expectation in online reviews on text, reviewer and product appreciation. Journal of Pragmatics, 75, 28-43.
Kuhberger, A. (1998). The Influence of Framing on Risky Decisions: A Meta-analysis. Organisational Behaviour and Human Decision Processes, 75(1), 23-55.
Levin, I.P., Schneider, S.L., & Gaeth, G.J. (1998). All Frames Are Not Created Equal: A Typology and Critical Analysis of Framing Effects. Organisational Behaviour and Human Decision Processes, 76(2), 149-188.
Levin, I.P., & Gaeth, G.J. (1988). Framing of attribute information before and after consuming the product. Journal of Consumer Research, 15, 374-378.
Mongin, P. (1997). Expected utility theory. In: Davis, J., Hands, W., Maki, U. (Eds.), Handbook of Economic Methodology. London: Edward Elgar.
Nelson, T.E., Oxley, Z.M., & Clawson, R.A. (1997). Toward a Psychology of Framing Effects. Political Behaviour, 19(3), 221-246.
Reisberg, D., Gleitman, H., & Gleitman, L. (2004). Instructor’s resource manual with classroom demonstrations (6th ed.). London: W.W. Norton & Company.
Tribushinina, E. (2008). Cognitive reference points: Semantics beyond the prototypes in adjectives of space and colour. Netherlands: LOT.
Tversky, A., & Kahneman, D. (1981). The framing of decision and the psychology of choice. Science, 211, 453-458.
Mean 0.96 0.83 0.96 0.83 "Outperformed by 20%" "Outperformed 80%" 3.96 4.49Condition
Mean Responses
Programme A "People will be saved" "People will die" 68.599999999999994 27.6 Programme B "People will be saved" "People will die" 31.4 72.400000000000006
Condition
Programme choice (%)