Background - Does music help infants identify ingroup members?
Background - Does music help infants identify ingroup members?
527050351790
-634993162300-634993162300
Original article
Mehr, S. A., Song. L. A., & Spelke, E. S. (2016). For 5-month-old infants, melodies are social.Psychological Science, 27,486 501. https://doi.org/10.1177/0956797615626691Listening to music as an adult has been shown to provide a multitude of benefits, including improvements in memory and learning, mood regulation, increased motivation in activities involving high levels of physical exertion, as well as the formation and maintenance of social bonds. Studies have also demonstrated many of these benefits in children as young as a few months of age. A study byTrehub and colleagues (2009), for example, provided evidence that infants rely on both visual and auditory (musical) cues to identify familiar individuals and thus form the necessary social bonds they need to meet the necessary developmental milestones. The study by Mehr and colleagues (2016) tried to build off this idea and wanted to explore the role that music plays in forming the appropriate social connections. Specifically, the authors wanted to test whether infants who hear familiar melodies sung by strangers use this information as a signal that the stranger is part of their social in-group.
To test this, the study recruited 32 5-month-old infants (and their parents) to take part in the experiment. In the first session, the parents (and their infants were randomly allocated to one of two conditions, each with a different lullaby. The lullabies used were novel to both the parents and the toddler. The parents were then asked to sing this lullaby to their infants every day for the next one to two weeks and then return to the lab for the second part of the experiment. First, the infants were shown a TV screen with videos of two strangers side by side, looking at the infant with friendly smiles. This was the baseline measure, and researchers measured the proportion of time the infant gazed back at each of the faces. After this baseline, each unfamiliar person was presented alone on the screen singing one of the two songs. The second unfamiliar person followed straight after and sang the second song. Of course, the toddler was familiar with only one of these lullabies. The final part of the trial once again involved the presentation of both strangers smiling back at the toddler and the researchers once again measured the proportion of time the toddler gazed back at each stranger. The order of the experiment is summarised in the figure below. For more details, please see the method section of "Experiment 1" (there are actually 4 experiments in total in this study, but our focus is going to be on experiment 1 only).
Preparing the Data File
If you had a chance to engage with the Bonus Material in the practice exercises in weeks 3 and 4, you by now will be aware that there is a grassroots movement that has been building in the scientific community to improve the core values of scholarship, and to increase the integrity of research findings being published. One key way this is being promoted is through theCentre for Open Science (COF), whose mission is toincrease the openness, integrity, and reproducibility of research. This will take time - cultural norms and practices that have been ingrained over decades will not shift overnight. But we are, as a scientific community, moving in the right direction.
One important recent shift has been the researchers are now advised (and sometimes expected to by some journals) to make their data freely available online so that others can access it and corroborate their findings if need be. To their credit, the authors of this study have done just that.The data for their experiments is available on theOpen Science Framework (OSF)website, throughthis direct link(which you can also find in their article). Go ahead and download both theLUL_pss.csvfile (opens in Microsoft Excel) and theLUL_pss_codebook.txtfile (opens in any text editor like Notepad). A "codebook" is a file that contains information on all the variables in the data file, as well as the structure, layout and any other relevant contents included. Creating codebooks is an excellent habit to pick up, especially when you start working on projects with large teams.
Open both the codebook and Excel file. You might initially be surprised by how much information it contains and the sheer number of variables. But this is how data is collected and stored in research, and with the aid of the codebook and a little bit of time and effort, it will all start to make sense. Remember also that the file contains data on 4 separate experiments, and we are only looking at the first one. In general, another excellent skill to master ishow to filter the dataso that only the relevant parts remain visible.
But perhaps just this once to make things as easy as possible, let's go ahead and delete everything we don't need.
Columns C to G provide information on which experiment the participant was in (1 for yes, 0 for no). You can see that the first 32participants(labelled 101 to 132 in Column A) have a "1" in column C, so this is the sample we are after.
You can therefore go ahead and delete all the participants from 133 onwards.
You can also delete columns B to G now as we don't need an experiment label anymore, as well as columns H to K because data on the date of birth and the date of the experiment are not needed for our appraisal.
Finally, all the columns for variables "etotal" onwards can also be deleted.
NB. For all the analyses that follow, you have the choice of using Jamovi or SPSS. You will need to provide all relevant output for all tests and analyses that you conduct in an appendix at the end of your submission file. The Appendix will NOT count towards the word count limit.
Reproducibility, Analysis and Appraisal (Q1 to 8)
Reproducibility is a fundamental principle that underpins the scientific method. Scientific findings are reproducible when others are able to use the original researcher's data and code/methods of analysis and obtain the same result.
Question 1. (Practice question no mark allocation)
In the results section for experiment 1, the authors report that "the infants showed no preference for either individual at baseline. The proportion of time that they looked toward the person who sang the familiar song did not differ from chance (.5;M= .521,SD= .177,t(31) = 0.67,p= .51. (p. 489).Use the relevant data in the data file and attempt to reproduce this reported result. Were you successful? If not, why not?
Remember to include all relevant/necessary output in the appendix.
Note: an APA style write-up and assumption testing is NOT required to fully address this question
Question 2.
Why did the authors feel the need to run a baseline comparison analysis? Why was it not sufficient to simply compare the attention given by the toddler to each stranger as they sang? As they explain in the method section, one of the songs was familiar to each toddler, and one was not. So why bother with the baseline measure at all? (See p. 489, Experiment 1, Results section, paragraph 1 sentence starting with The proportion of time that they looked toward the person who sang the familiar song did not differ from chance)
Question 3.
Later in the results section, the authors claim that the "infants attended highly and equally to the two singers during the familiarization trials, as each singer appeared by herself and sang a song (duration of looking toward the singer of the familiar song:M= 15.6 s,SD= 5.07; duration of looking toward the singer of the unfamiliar song:M= 15.3 s,SD= 5.10),t(31) = 0.28,p= .78". Once again, use the appropriate variables from the data file and attempt to reproduce this result.Hint: the videos used in the experiment use 30 frames per second.
Remember to include all relevant/necessary output in the appendix.
Note: an APA style write-up and assumption testing is NOT required to fully address this question
Question 4.
In the method section for experiment 1, under the subheading "Statistical power", the researchers claim that "a sample of 32 had .84 power to detectan effect" with a magnitude of 0.54.
a) Is this claim accurate? Use G*Power (or other suitable software) to confirm or disprove this, and provide evidence in the form of a screenshot in the appendix of your submission.
b) The authors claim that a "similar experiment testing effects of language rather than music (Kinzler et al., 2007) obtained an effect size (d) of 0.54", which is the justification they rely on to use the same size for their power analysis. Are you able to corroborate this by searching through this previous study (link here)? Use the "control-F" function to simply search for key terms such as effect size, 0.54, etc. What do you notice?
c) Putting this current example aside and merely speaking broadly and hypothetically, what incentive do researchers have to misreport the size of the effect found in previous research? What (if any) benefits do they gain?
Question 5.
Why do you think the authors selected a repeated measures design for this experiment? Do you agree with this choice, or would you have gone with a different design? Make sure you discuss both the strengths and limitations of this approach in your response. Do not simply provide a list of generic pros and cons from the lectures and readings. Rather use and extend what you have learned and apply it to the context of this specific experiment.
Question 6.
As you by now know, one of the assumptions of a repeated measures ANOVA is sphericity. Why is this not also an assumption for a repeated measures t-test?
Question 7.
What attempts have the authors made to control for the extraneous variables that could bias the results? Are there any that they have missed?
Question 8.
In an interview after the publication of these results, one of the authors was quoted as saying that "even toddlers as young as 5 months of age start to develop preferences for in-group members, and that experiment 1 was able to prove that familiar melodies are one of the tools that they rely on to make the in-group/out-group distinction". Do you think thatis a justifiable claim and summary of the research findings? Why or why not?
Extending the Analysis (Q9 & 10)
Question 9.
Is there a gender difference regarding preference for either singer at baseline (i.e., before the toddlers hear the melody)? Note. The gender is referring to the toddlers gender, not the singers gender.
Provide the null hypothesis for the appropriate analysis to answer this question.
Provide the alternative hypothesis for the appropriate analysis to answer this question.
Run the appropriate analysis to answer this question. Remember to make sure togenerate the relevant descriptive statistics, test any required assumptions and select the appropriate test. Remember to include all relevant/necessary output in the appendix.
Generate a full write-up of the results according to APA requirements.
Question 10.
Although this is not reported in the published article, the toddlers were actually tested one more time, 10 minutes after the experiment concluded, for a total of three timepoints. This third time period trial only involved the two singers once again simply directly gazing at the infant, for 16 seconds. Once again, the researchers measured the proportion of this time that the infants selectively attended to the now-silent singer who had sung the song that was familiar to them and were interested in any differences that might exist across the trials.
The data is presented below, labelled pref3 (double click on the table and an excel file will pop up):
Using this new data in combination with the data file you have been provided previously, select the most appropriate analysis of the research question under investigation.
Run the appropriate analysis to answer this question. Remember to make sure togenerate the relevant descriptive statistics, test any required assumptions and select the appropriate test. Remember to include all relevant/necessary output in the appendix.
Generate a full write-up of the results according to APA requirements.
