Dear Mr. Stanley:

I appreciate the publication of my letter to the editor, and I was interested to see the varied reactions of your readers to the series of stories on segregation.

Contrary to your conclusion in your note to me, I see two areas in which Mr. Murphy's stories require correction.

1. My view of the validity of the UWM methodology

I remain disturbed by what I consider to be a flagrant misrepresentation of my opinion of the approach taken by the UWM study. Your readers were led to believe that I approved of their methodology. One letter yesterday recited this clearly: "The new UWM research has concluded - and the leading professionals in the measurement of integration have accepted as valid - that Milwaukee now ranks in the middle of integration in the nation." This is not surprising because Mr. Murphy's original text stated as follows: "Vigdor, Glaeser and Logan are three of the best known of the many academic proponents of the dissimilarity index. All three said an approach measuring blocks with at least 20% white and 20% black population would be a valid way to measure integration."

Here is what one might assume Mr. Murphy had done. He might have called me, told me that a study was using the percent of the population in blocks that are 20-80% black as a measure of integration, and asked me my opinion of that measure.

Here is what I remember instead. He asked me whether it made a difference that in Salt Lake City there were very few blacks even in the most integrated neighborhoods, while in Milwaukee there were more people in neighborhoods with a mixture of blacks and whites. And I said that the racial composition of the neighborhoods where people live is an important aspect of the residential pattern, and one that is not directly reflected in the Dissimilarity Index. I told him that is one reason why the Mumford Center also includes various exposure indices in our reports on segregation, such as the percent black in the neighborhood where the average black person lives. And I told him that when using the Dissimilarity Index I consider it more revealing to compare metro areas with similar overall racial composition.

If he had asked me about the 20%-80% approach as a measure of integration, I would have expressed great reservations. First, a measure of this type inevitably reflects the overall racial composition of the region. Areas with larger black populations are more likely to have neighborhoods that are as much as 20% black. Second, in a metropolis that is 2% black like Salt Lake City, or 16% black like Milwaukee, or even 29% black like Atlanta, including neighborhoods that are as much as 70% or 80% black in the definition of "integrated" defies common sense understandings of what is meant by integration. Such neighborhoods are at the very high end of extreme racial concentration. So, whether at the lower end or the upper end of the definition of what is integrated, this is a flawed measure. A measure like this could be used in a limited way to assess integration in a single metropolitan area over time, if its overall composition didn't change much, or to compare two metropolitan regions with roughly the same racial composition. But even then, the lower and upper bounds would have to be set in a reasonable way. In Milwaukee, where 16% of residents are black, one might very well think of a neighborhood that is 10% black as "integrated," but not a neighborhood that is 80% black. The definition would need to fit the circumstances of the cases that are being compared. But I can't think of any metropolis where a 20%-80% criterion seems reasonable, and I wouldn't use a measure like this at all if my purpose were to compare metros with very different racial makeup.

How did Mr. Murphy conclude that I believe "an approach measuring blocks with at least 20% white and 20% black population would be a valid way to measure integration" -- well, it certainly wasn't by asking me. I wonder if he asked Professors Glaeser or Vigdor?

2. My reporting of segregation indices for areas with smaller black populations

Here is what Mr. Murphy thought was worth a story in itself. "Logan simply drops metro areas with low numbers of African-Americans, such as Salt Lake City, from his index." Mr. Stanley, I invite you to look for yourself at our webpage (brownS4.dyndns.org/cen2000_s4/WholePop/WPsegdata.htm). You will see that indices are presented for every metropolitan region in the country, and for every city with more than 10,000 people. We do advise caution in interpreting Dissimilarity Index values for places with smaller black populations, and for good reason. In the explanatory text for the sorted list where a person can see values for all 331 metro areas in a single table, we offer this advice:

"Remember that values of the index of dissimilarity in areas with small group sizes are very sensitive to change in population size. In such areas, even a small increase or decrease in the size of the group can cause a large increase or decrease in the dissimilarity index. This change is not necessarily indicative of a real change in the residential experience of the average group member. As a reminder, we have displayed in red the names of areas where one of the groups in the sorted column has a small population total (under 50,000 for Hispanics and blacks, under 20,000 for Asians). While these cutoffs are somewhat arbitrary, they are provided as an aide in identifying areas that should be interpreted with caution."

This is certainly not the same as "simply dropping" some metro areas, and we do not advise people to disregard these statistics or or that they "shouldn't be taken seriously," as Mr. Murphy's story suggests.

But then Mr. Murphy complains that "Logan simply drops any metro area with less than 50,000 African-Americans from his rating of segregation in the top 50 metro areas." Well, the cutting point was actually 147,000, the black population of Louisville. Given the number of cases that we thought could fit in a one-page table, we had to decide which ones to include. Any choice would have meant omitting 280 metros. Here's what we said about the "top 50" list in the text of the report where this table is found: "The next tables in this sequence list the 50 metropolitan regions in the country that had the largest black populations in 2000." I think that's clear. I'm distressed that anyone read it differently. But my point is that we included every case in every way that we had the opportunity to do so, and we provided appropriate advice to viewers about how to use these numbers. In fact, I am very proud that we developed a system that made it possible for people everywhere to learn what Census 2000 revealed about their city or metropolitan region.

I have been reassured by the many contacts that I have had in the past year with the press. Many journalists, on your staff and elsewhere, have spent a lot of time to get the story right. Personally I think it is more an insult to them than to me when one writer gets it so wrong.

I honestly don't have in mind a particular remedy. I doubt that you will retract any part of the stories that you published, or that you will inquire of other experts whether their views were correctly reflected. And I found a degree of balance in related coverage -- in the variety of letters to the editor, in the content of the editorial "A new look at segregation," and in a related column by another member of your staff. Let me just express the hope that your newspaper will continue to promote informed public discussion of this topic that is so very important to your hometown and to the nation.