Using chi-square analyses, explore the relationship between children’s race (RACE5) and poverty status (whether the child’s family lives below the federal poverty line; WKPOVRTY); full-day kindergarten attendance (FULLDAY); daycare participation (P1PRIMNW - be sure to use the categorical variable, not the dummy variable originally listed); kindergarten repetition (P1FIRKDG); and whether the child has an identified disability (P1DISABL). What patterns did you find? Your findings may be placed into a single table.
*I'll be hanging out in the 2nd floor of library Thursday afternoon if you need extra help - email me. -Megan.
Subscribe to:
Post Comments (Atom)
4 comments:
Hi Megan,
I have two questions if you don't mind,
Sorry I got a bit confused, can we mention practical significance in our findings for this week?
Also, I received your e-mail regarding the weighting issue, I'm wondering for our weights section, could we still write stuff like "I weighted all of our analyses using the student-level weight BYCW0 (normalized to preserve the child sample size for statistical testing)..." or should we be writing "...using child_wt..."? What is the difference between the two anyway?
Thank you so much!
Best,
Esther
hi
Left comment the other day but I don't see it here.
with the chi sq table. What do we include in it? Everything from output? only the the topic of interest ie repeaters and do not inlcude non repeaters? I am confused.
Esther -
Yes, you should always mention practical significance when talking about your findings. That doesn't mean use the word "practical significance," it just means to be sure to address whether or not the difference is meaningful - discuss the magnitude of the difference. i.e. if the relationship is significantly different at the the p<.001 level, but the difference is .003 percentage points, you would mention the statistically significant relationship, but further note that the difference is negligible.
As for the weight, child_wt IS bycw0 normalized to preserve the sample size - so use the first example you gave. (The difference between the two is that bycw0 weights the sample up to an n of the total kindergarten children in the US. The child_wt is bycw0/mean(bycw0) so that you keep the same sample size.)
Martie -
Re:the table, take a look at what Doug did in his explaining girls' kindergarten lit paper, p 29. When you have a dummy variable, only include one indicator, the indicator for whichever is most interesting (%female; %dropouts; %repeater; etc.). When you have a categorical variable, include all the indicators (for example, Race/ethinicity, and then indent and include %asian, %black, %hispanic, %white, %other).
Post a Comment