Review of causal research Phil 12: Logic and Decision Making Fall 2010 UC San Diego 11/22/2010
Announcements Trying to answer paper topics for other sections a good way to study Sample final exam questions will be posted tomorrow Must register your clickers if you didn t already Today CAPE should send you an email at your UCSD email address - link for evaluating each of your courses - Deadline: midnight before Finals Remember no lecture or section on Wednesday My office hours for this week are today rather than Wed
Clicker Question 1 Which of the following involves questioning the internal validity of an experiment? A. Questioning whether the causal relation between these variables would hold when using a different task B. Raising doubts about whether the results are applicable to a different context C. Questioning whether your operational definitions really measure the theoretical constructs of interest D. Questioning whether there might be an alternative explanation for the effect produced
Clicker Question 2 Which of the following concerns are about external validity? A. Your study did not achieve statistical significance B. A non-representative group of participants was studied and one can t anticipate what sorts of effects the independent variable will have on other participants C. Participants might have responded to the novelty of the experiment more than the specific effect of the independent variable that was manipulated D. The participants figured out which treatment they were on
Review Internal validity: are the effects on the dependent variable due solely to the manipulation of the independent variable? - Was there a confounding subject variable? - Was there a confounding procedural variable? Demand characteristics did subjects behave as they did because of knowing they were in an experiment? Experimenter bias in measuring variables External validity: do the results of the study generalize to the population, setting, and manipulation of interest?
Clicker question 3 Which type(s) of causal research involve identifying groups in terms of the independent variable? A. Experiments B. Prospective studies C. Retrospective studies D. Two of the above E. All of the above
Clicker question 4 Which type(s) of causal research can make use of randomization to control for subject confounds? A. Experiments B. Prospective studies C. Retrospective studies D. Experiments and prospective studies E. Experiments and retrospective studies
Review - 1 Experiments provide the best evidence of a causal relation, but sometimes they are not possible - Because it is impossible physically or morally to manipulate the independent variable Two strategies - Prospective studies Divide groups according to the independent variable and investigate correlation with the dependent variable - Retrospective studies Divide group according to the dependent variable and investigate correlation with the independent variable
Review - 2 All studies of causation are beset by confounds - Factors correlated with the independent variable that may themselves be the cause of the change in the dependent variable By manipulating the independent variable in an experiment, researchers reduce the risk of confounds - Researchers can randomize or match subjects or lock (control) procedural variables to minimize confounds Prospective and retrospective experiments do not allow manipulation - Greater risk of confounds. Try to reduce the risk by: matching subjects measuring possible confounds
Science in the news Paper 2 asked you to come up with a causal study to test a given causal hypothesis Today we ll practice understanding and evaluating causal research reported on in the news
Alcoholism: Clinical and Experimental Research Vol. 32, No. 10 October 2008 Sound Level of Environmental Music and Drinking Behavior: A Field Experiment With Beer Drinkers Nicolas Guéguen, Céline Jacob, Hélène Le Guellec, Thierry Morineau, and Marcel Lourel Objective: It had been found that environmental music was associated with an increase in alcohol consumption. The presence versus absence of music, high versus slow tempo and the different styles of environmental music is associated with different level of alcohol consumption. However, the effect of the level of the environmental music played in a bar still remained in question. Methods: Forty male beer drinkers were observed in a bar. According to a random distribution, patrons were exposed to the usual level of environmental music played in 2 bars where the experiment was carried out or were exposed to a high level. Results: The results show that high level volume led to increase alcohol consumption and reduced the average amount of time spent by the patrons to drink their glass. Conclusions: The impact of environmental music on consumption was discussed and the arousal hypothesis and the negative effect of loud music on social interaction were used to explain our results. Key Words: Young Adults, Environmental Music, Music Influence, Alcohol Consumption. E NVIRONMENTAL MUSIC IS known to affect behavior and particularly the consumer s behavior. Several experimental studies carried out in natural settings have shown that different environmental music and structural components of the music (e.g., sound level, tempo, tonality) affect the consumer s behavior such as in-store traffic flow (Milliman, 1982), sales volumes (Areni and Kim, 1993), product choices (North et al., 1999) or time elapsed in a commercial area (Milliman, 1986). It was shown in social psychology literature that drinking behavior is affected in multiple ways by environmental music. Drews et al. (1992) shown that male beer drinkers unobtrusively observed in 2 bars, drank significantly more beer when environmental music was played than when the music was off. Milliman (1986), when examining the effect of music on the behavior of restaurant patrons, shown that aslowermusic tempoledtoan increase in the average dollar amount of bar charges per customer. These findings were the fact that customers selection of French and German wines was strongly affected by stereotypic French and German environmental music played in the store. French music increased the sales of French wines compared to German ones whereas German music led to the reverse effect. Jacob (2006) had conducted an experiment in a bar to test the influence of 3 different styles of music on patrons. According to a random assignment, patrons were exposed to top 40 music, which was usually played in the bar, cartoon music or drinking songs. Results showed that drinking songs appeared to increase the length of time customers stayed in the bar and the average amount spent. An aspect of the structural components of music that has been studied less as far as drinking behavior is concerned is the sound level. A recent experiment of Guéguen et al. (2004) had tested the effect of sound level on drinking behavior in a bar. An experiment was carried out in 2 bars to test patrons response to music loudness. Using random assignment,
Music and alcohol consumption Previous research found that environmental music is associated with an increase in alcohol consumption - - - The presence versus absence of music high versus slow tempo different styles of environmental music Arousal hypothesis: music enhances arousal level, which enhances drinking behavior What about the effect of the sound level of the environmental music played in a bar?
Music and alcohol consumption Subjects: 40 males (aged 18-25) drinking in 2 bars in a medium-sized city in France On 3 Saturday nights, observers manipulated the sound level of the music in the bar - control condition: normal sound level of 72 db - high sound level condition: 88 db Sound level randomly manipulated, then observers looked for first people entering bar to observe - Participants only selected for observation if: male, in a pair, and at least one person ordered a glass of draft beer
Music and alcohol consumption The observers used several measures of the DV, alcohol consumption: - number of drinks ordered - time spent to drink each glass (in minutes) - number of gulps used to drink each glass
Music and alcohol consumption Level of environmental music Number of drinks ordered Time spent to drink a glass (in minutes) Number of gulps per drink Usual level 2.6 (1.14) 14.51 (4.88) 7.02 (1.26) High level 3.4 (0.99) 11.45 (2.89) 7.18 (1.29) Is the difference between groups significant? p<0.03 p<0.03 n.s.
Music and alcohol consumption Internal validity - how were potential subject and procedural confounds controlled for? - any potential confounds not controlled for? External validity - Would results generalize to other populations, settings, manipulations?
2010 Fe bru ary M arria g e a nd C oh a bitation in the Unite d States: A Statistic al Portrait Base d on C ycle 6 (2002) of the N ation al Survey of Fa mily Growth Seri es 28 Nu m b er23,
Cohabitation and marriage success Data from the 2002 National Survey of Family Growth Subjects: 12,571 Americans (7,643 females and 4,928 males) aged 15-44 Survey questions either asked by a female interviewer or a computer
Cohabitation and marriage success Figure 6. Percentages of women and men aged 25 44 who have been married two or more times, by current age: United States, 2002
Cohabitation and marriage success 70 Non-Hispanic white Non-Hispanic black Hispanic 60 Non-Hispanic white Non-Hispanic black Hispanic 60 50 63 58 51 51 49 50 40 53 42 50 49 53 47 Percent 40 30 39 Percent 30 20 20 10 10 0 Ever married Ever cohabited 0 Ever married Ever cohabited Women Men
Cohabitation and marriage success 30 25 28 26 20 21 Percent 15 10 11 5 0 No high school diploma or GED High school diploma or GED Some college Bachelor s degree or higher NOTE: GED is General Educational Development high school equivalency diploma. SOURCE: CDC/NCHS, National Survey of Family Growth, Cycle 6 (2002). Table 11 in this report. Figure 9. Percentage of women aged 22 44 who have had two or more cohabiting relationships, by her level of education: United States, 2002
Cohabitation and marriage success Probability that a first marriage will remain intact (survive) For women: 1 year 3 years 5 years 10 years Probability Probability Probability Probability Characteristic of survival SE of survival SE of survival SE of survival SE Ever cohabited before first marriage Yes... 0.94 0.005 0.84 0.008 0.75 0.009 0.60 0.012 No... 0.95 0.005 0.87 0.008 0.80 0.011 0.66 0.020 Engaged to first husband when premarital cohabitation began Yes... 0.94 0.005 0.86 0.008 0.79 0.009 0.65 0.013 No... 0.94 0.005 0.80 0.011 0.71 0.014 0.55 0.015 Probability that a woman s marriage would survive for 10 years: - Did not cohabit at all: 66% - Cohabited before marriage: 60% p = 0.06 Difference not significant at 5% level
Cohabitation and marriage success Probability that a first marriage will remain intact (survive) For women: 1 year 3 years 5 years 10 years Probability Probability Probability Probability Characteristic of survival SE of survival SE of survival SE of survival SE Ever cohabited before first marriage Yes... 0.94 0.005 0.84 0.008 0.75 0.009 0.60 0.012 No... 0.95 0.005 0.87 0.008 0.80 0.011 0.66 0.020 Engaged to first husband when premarital cohabitation began Yes... 0.94 0.005 0.86 0.008 0.79 0.009 0.65 0.013 No... 0.94 0.005 0.80 0.011 0.71 0.014 0.55 0.015 Probability that a woman s marriage would survive for 10 years: - Did not cohabit at all: 66% - Was engaged when they began cohabiting: 65% - Not engaged when premarital cohabitation began: 55% n.s.
Cohabitation and marriage success Probability that a first marriage will remain intact (survive) For men: 1 year 3 years 5 years 10 years Probability Probability Probability Probability Characteristic of survival SE of survival SE of survival SE of survival SE Ever cohabited before first marriage Yes... 0.94 0.008 0.84 0.013 0.77 0.013 0.62 0.015 No... 0.94 0.008 0.84 0.015 0.78 0.016 0.69 0.017 Engaged to first wife when premarital cohabitation began Yes... 0.96 0.004 0.90 0.006 0.82 0.010 0.71 0.015 No... 0.94 0.014 0.78 0.021 0.72 0.020 0.53 0.017 Probability that a man s marriage would survive for 10 years: - Did not cohabit at all: 69% - Cohabited before marriage: 62% p <0.05
Cohabitation and marriage success Probability that a first marriage will remain intact (survive) For men: 1 year 3 years 5 years 10 years Probability Probability Probability Probability Characteristic of survival SE of survival SE of survival SE of survival SE Ever cohabited before first marriage Yes... 0.94 0.008 0.84 0.013 0.77 0.013 0.62 0.015 No... 0.94 0.008 0.84 0.015 0.78 0.016 0.69 0.017 Engaged to first wife when premarital cohabitation began Yes... 0.96 0.004 0.90 0.006 0.82 0.010 0.71 0.015 No... 0.94 0.014 0.78 0.021 0.72 0.020 0.53 0.017 Probability that a man s marriage would survive for 10 years: - Did not cohabit at all: 69% - Cohabited before marriage: 62% - Was engaged when they began cohabiting: 71% - Not engaged when premarital cohabitation began: 53% n.s.
Cohabitation and marriage success What kind of study was this? Independent and dependent variables? Internal validity - were any potential confounds controlled for? - any potential confounds not controlled for? External validity - Would results generalize to other populations, settings, manipulations?