by Lina Guerrero
What is
your first memory of your childhood? How many toys and books did you have
access to? Did your parents provide good care for you? The reason I want to ask
this question is to make you aware of the conditions that surrounded you when
you were growing up.
I am
interested in testing if socio-economic status (SES) affects children’s
cognitive development. In order to measure SES, I used the
proxy wealth index which is composed of the average of three different indexes: Housing
Quality Index, Consumer Durables Index and Services Index. SES is an important
factor when it comes to understanding children’s cognitive development as some
studies suggest that low-income families provide poor stimulation in the home
affecting cognitive development. Parents in a higher income family might
provide a better home environment such as having more leisure time to spend
with their children and might be more prone to provide books, toys and other
types of stimulating materials that might make them better off than poor
children. The environment for children is of great importance and there are
some frameworks attempting to measure key conditions. For instance, The Home Observation for Measurement of the
Environment (HOME) Inventory measures different dimensions that include the
quality of the house materials, the quality of the stimulation on the home-based
by learning materials, parental involvement, play activities with parents, etc.
In order to understand the effect of cognitive development and SES I used a common measure of child cognitive development: the results on the
widely-used Test de Vocabulario en Imágenes Peabody (TVIP), the Spanish version
of the Peabody Picture Vocabulary Test (PPVT) . The
PPVT assesses receptive vocabulary for children using color pictures as
response options on a page that requires the child to point to the picture
which demonstrates the word they hear. For example, the examiner says “Show me
cat” and the child needs to look at the four pictures on the page and point to
the picture of the cat. This is a popular proxy of testing cognitive
development. The dataset I used was collected in the Young Lives Study and it is
composed of data from 5-year-old children in Peru. (Click
here to watch a real test)
In order
to ensure I am predicting the best summary line through the cloud of data I
began the analysis of my two variables by creating a bi-variate scatter plot to
visualize if there was a relationship between my two variables. In order to
improve it, I transformed my dependent variable using logarithmic data to
better fit it to a straight line.According to the bivariate regression model, the wealth index has a statistically significant effect on the score of the PPVT test.
Although this finding is significant, it is important to acknowledge that SES and Cognitive Development are also affected by other factors that are important to include in this model. These control variables help to come to a closer understanding of the true effect of the wealth index variable on the results of the PPVT Test. Although there are many factors that cannot be captured in one model, I selected key factors related to the parent’s characteristics: education of the mother and the father; child characteristics: sex, relative health status, and attendance of preschool and Home environment: living in rural or urban area, and access to sanitation. Summary statistics for all of the independent variables can be found in Table 1 at the end of this post.
After analyzing the results of my estimates I can conclude that wealth
index does affect the scores of the PPVT Test. There is a positive correlation between theses two variables. On average,1-point increase in the wealth index would result in a 85% increase in the score of PPVT test, while all other variables in the model are held constant. This result was highly statistically significant as wells as preschool attendance, parents education level, and living in urban or rural areas. In contrast, sex of the child, relative health level and access to sanitation services do not appear to be statistically significant.Therefore, we can conclude that there is no relationship between the Results in the PPVT Test and these variables and any relationship is due to chance. Moreover, the adjusted R squared statistic tells us how well the regression model predict the distribution of the data overall. In this case, 46% of the variation in the PPVT Test Results is explained by a variation of 46% in the independent variables.Details of the regression analysis can be found in the Annex section, Table 2.
So why
should we care about the effect of SES in children’s cognitive development? The
first 5 years of a child’s life are critical. It is a time of dramatic brain
growth and development. However, this important time of development can be
badly affected by living in an unstimulating home environment. During early
childhood the brain forms and refines a complex network of neural connections
that are established by the interaction with the environment. Connections used frequently become stronger
and more complex. Connections not used are eventually removed. By understanding
the importance of these concepts, governments can design programs that can
intervene early in the lives of poor children to ensure a more productive
society.
In
conclusion, this analysis showed me that that SES plays an important role in
brain development in early years of life. Interventions in early childhood can
have a significant effect on children’s development. This association is
a very important determinant of everything that happens in the development of a
person’s life such as school performance, getting a job and amount of income. In Peru, there is a program sponsored by the current President
and entitled Cuna Mas that aims to improve development of children less
than 3 years of age in areas of poverty and extreme poverty, to overcome gaps
in their cognitive, social, physical and emotional development. I believe that
investing early in the life of children will significantly help to reduce the
gaps between rich and poor and improve the SES of Latin America that is the most unequal region in the world. These results confirm the magnitude of the challenges faced by
policy-makers seeking to close the gaps in early childhood development in Latin
America.
Annex:
Your topic is very interesting, and you provide a great deal of supporting evidence for the importance of the study. However, the post is very heavy on literature review and additional information about the topic. I'd like to see more about the regression analysis itself.
ReplyDeleteYou might consider posing the research question much earlier in the blog. When it appears, it's very clear, but up until that point, I wasn't sure what the analysis would cover exactly. Your description of variables and confounds is good; you clearly demonstrate how you're measuring and what else may affect your results.
The visuals aren't appearing. The issue may be that copying from Word doesn't paste them; you have to insert the images manually. So I don't know how the graphs/tables look, but I'd be interested to see the results in greater detail.
I'd like to hear more about what exactly you did in your analysis. Were the variables skewed? Was there multicollinearity? I think it would be helpful to just walk your audience through each step you took to arrive at the results. Also, when you conclude that the results are statistically significant, you might talk about the p-value; there's still a (potentially extremely, but I can't see the table) small probability that the relationship occurred due to chance.
Also, it might be helpful to ask someone to proofread. Your post is really interesting, but there are a few grammatical issues that detract from full understanding.
Overall, a very good post/topic. The main issue is the ratio of information about the field to the analysis; the analysis needs more focus. Nice job!
Hola Lina! :)
ReplyDeleteThe topic you chose is really interesting! Both your title and your beginning paragraph grab the reader's attention by relating the question to their own experience and upbringing (I think that's really good). I really liked that you clearly stated the significance of the study, particularly in Latin America, and heavily supported with a literature review why the topic is important and studies and findings performed in the past. Although, you clearly stated both the IV and the DV that are being tested in the study towards the later paragraphs,I would have liked to learn about what you were testing before the literature review. Maybe add just the intentions of your study and what you are measuring in a sentence at the beginning and then proceed with the literature review?
Once you get to the study, you clearly state the confounding variables, how you are measuring all your variables, and any weaknesses in the data. I would love to see more information on the study, perhaps going more in depth into each variable, how you got to that point, maybe even the changes that you made to the variables and why they were relevant. I would suggest adding to the conclusion whether or not the relationship is statistically significant, and whether or not you are taking into account other factors that might affect the causality of the study. There are a few grammatical and syntax errors at the beginning, so it'd be a good idea to proofread it multiple times or have someone else look over it. Also, I would like to see the years in which the studies where conducted, and which year you are getting your data/variables from. Also, I can't see your graphs, so I can't really comment on them :(
Good job!!! Really good topic!!!
Great opening paragraph, you make a strong case for the reader to pay attention to the problem that you’re addressing. You clearly identify both the holes in previous research and your research question, but the research question seems to be addressing something that the research has answered already. If you can elaborate on why your study is different/contributes new material to the field, I think it would make your blogpost stronger.
ReplyDeleteYou do a very good job of pointing out other variables that could have been included in the study (and non-measurable conditions that could confound it), but you don’t tell the reader why you didn’t include the measurable variables. It could be as simple as the data not being available, but spell that out a little more clearly.
I like that you include an endorsement for a specific program in the last paragraph. However, you might want to mention it earlier in the blogpost (possibly using it as the reason you want to look at Peru specifically). Mentioning it in passing at the end is a little confusing for the reader.
Just a stylistic note, you might want to reformat the regression table to look like the ones we’ve been making for the problem sets. I don’t think anyone without a statistics background would be able to understand the current table. You also should include a summary table so the reader knows how many observations the dataset includes, etc. And you should add some labels on your scatterplot (y-axis, title).
It’s a good start looking at a subject that you clearly care about and explain well why others should care about.