Who has the risk to become a low performer?
Does social status or immigrant background matter?
Further analysis, based on multivariate logistic regression, considers how
different risk factors change (increase or decrease) the likelihood of low
performance in Math.
Risk factors were picked out manually according to common
practice and stereotypes. As a result, in some cases a risk factor can have
actually positive effect and decrease the likelihood of becoming a low performer.
Note that influence of each risk factor is measured in comparison with alternative.
For example, influence of “being a female” is measure in comparison with “being a male”,
“studying in public school” is compared with “studying in private school” and etc.
Having disadvantaged social and economical status and having no preschool
education are two most significant risk factors, which have negative effect
in all considered countries and economies.
In most country gender has very low influence. However, in Brazil, Chile
and Columbia “being a female” can have negative effect and increase likelihood
of low performance. On the contrary, in Thailand “being a female” has positive effect,
which means that male students have higher chances to become low performers.
In some countries school type is quite important and it is interesting to note
that public schools do not always have negative effect. For example, in Canada
and Japan being in public school decrease the likelihood of low performance
(comparing with private schools). In Brazil, Columbia and Peru situation is
the opposite, studying in public school has negative effect.
Values in the heatmap correspond to coefficients of multivariate logistic
regression. Low performance status (binary variable, 1 – student is low
performer, 0 – otherwise) is regressed one a set of risk factors (categorical
variables with two levels).
Positive coefficient value means that a factor has negative effect and
increase the likelihood of low performance in Math; negative value means
positive effect and decrease of likelihood of low performance in Math.
Social and economical status is based on the PISA index of economic,
social and cultural status. Disadvantaged status is defined as a status
with the index below 25% quintile, threshold is defined separately for
each country.
Note that value of logistic coefficients are not equivalent to the value of
change in likelihood, but show change in logarithm of odds ratio (odds ratio equals
to probability of becoming low performer divided by the probability of not
becoming low performer).
Used Abbreviations:
UAE - United Arab Emirates
USA - United States of America
Russia - Russian Federation
Perm(RF) - Perm(Russian Federation)
Connecticut - Connecticut (USA)
Massachusetts - Massachusetts (USA)
Hong Kong - Hong Kong-China
Who has the risk to become a low performer?
Does social status or immigrant background matter?
Further analysis, based on multivariate logistic regression, considers how different risk factors change (increase or decrease) the likelihood of low performance in Math.
Risk factors were picked out manually according to common practice and stereotypes. As a result, in some cases a risk factor can have actually positive effect and decrease the likelihood of becoming a low performer.
Note that influence of each risk factor is measured in comparison with alternative. For example, influence of “being a female” is measure in comparison with “being a male”, “studying in public school” is compared with “studying in private school” and etc.
Having disadvantaged social and economical status and having no preschool education are two most significant risk factors, which have negative effect in all considered countries and economies.
In most country gender has very low influence. However, in Brazil, Chile and Columbia “being a female” can have negative effect and increase likelihood of low performance. On the contrary, in Thailand “being a female” has positive effect, which means that male students have higher chances to become low performers.
In some countries school type is quite important and it is interesting to note that public schools do not always have negative effect. For example, in Canada and Japan being in public school decrease the likelihood of low performance (comparing with private schools). In Brazil, Columbia and Peru situation is the opposite, studying in public school has negative effect.