Reading output 2.3 in Fixed Effects Methods for Linear Regression (Paul.Allison)

[Phil_NZ]

Hi all SAS Experts,

I am new to SAS regression, I am working in finance research. After asking some questions in this forum, some experts ask me to read the book "Fixed Effects Methods for Linear Regression" (Paul Allison, 2005), then search lexjansen for answers then come back to this forum to ask if there are any questions (e.g: @StatDave  and @Reeza ). I am wondering if the questions about that book are allowed to be posted here, if not, please let me know and I am happy to receive any suggestion to align with the forum's rules.

I am quite a bit confused about that 

1. The example in this book (I am standing on chapter 2) is straightforward with simple examples, I am wondering why many suggest me to go through lexjansen to search in any case, can I ask how to use lexjansen website and this book efficiently?

2. Regarding Output 2.2 in this book (page 13), he wrote something that I did not understand, and I do not know how to search from lexjansen website to go through it

I do not understand the part "Output 2.3 shows that although the majority of children did not change in status, about 24% did change in one direction or another", what does "24%" means and where it is on the Table?

Apart from that, the Table output 2.3 also really unfamiliar to me, because I do not know how to read this Table. For example, I do not know pov90 standing there for what, and what do 0 and 1 mean?

Warmest regards.

[Rick_SAS · Accepted solution]

I think the suggestions "read the book and search lexjansen.com" are meant as generic recommendations. For example, if someone is trying to solve a logistic regression problem, they need to understand the basics before they can start solving problems. The forum doesn't have explicit "rules," but it is not a replacement for studying an area of statistics such as logistic regression. I think the suggestion is that you should try learning the basics on your own.

To answer your question, I would check the book for the definitions of the pov90 and pov94 variables. (Is there an appendix that lists the data sets?) From the variable names, I assume that the table shows the poverty status of 581 children. A standard notation for a binary variable is that 1 means "has the property" whereas 0 means "does not have the property. The 2x2 table shows the counts of students who were or were not in poverty in 1990 and in 1994. Of the children,

65 were not impoverished in 1990 but were in 1994 (the upper right cell). Additionally, 

73 were impoverished in 1990 but were not impoverished in 1994 (the lower lefts cell). Those 65+73=138 students "changed status". Because there are 581 students, 138/581=0.237 ~ 24% of the students changed status between 1990 and 1994.

The 321 in the upper left cell were classified as living in poverty in 

pov90 indicates the number of students who were in poverty in 1990. There were 73 in poverty and 321 not in poverty. I assume that pov94 indicates the number of students who were in poverty in 1994. There were 73 in poverty and 321 not in poverty.

[Reeza]

The reason LexJansen is mentioned is that it has detailed user papers on almost every topic in SAS that are free and easily accessible. It goes back very far and is considered the best resource for learning about any topic in SAS and is better than a general Google search.

A textbook is great, but many people balk at paying for books or resources, so LexJansen is an alternative resource.