Pets and Peeves
    
Educational Attainment in the United States 2005 [link] (11/3/06)
According to A. N. Kolmogorov, maximum complexity (“randomness”) occurs when the shortest formula that specifies the sequence is the one that names it, element by element.
Read more [link] 10/21/06
See the most recent issue of the Evaluation Exchange newsletter [link] (3/8/06)
Winter 2005-2006
Are journals doing enough to prevent fraudulent publication? [Link] (2/15/06)
Canadian Medical Association Journal
Editiorial
February 14, 2006; 174 (4)
When Good Data Goes Bad [link] (2/10/06)
Society Report Describes Historic Drop in Cancer Deaths [link] (2/10/06)
Various Websites that Report Opinion Polls (1/29/06)
The Pew Charitable Trusts: Public Opinion and Polls
Zogby International
PollingReport.com
Public Agenda
The Roper Center
Google Search of Polling-related Websites
PULLING APART: A State-by-State Analysis of Income Trends (1/28/06)
View image of state by state income ratios
Abstract
In most states, the gap between the highest-income families and poor and middle-income families grew significantly between the early 1980s and the early 2000s, according to a new study by the Center on Budget and Policy Priorities and the Economic Policy Institute. The study is one of the few to examine income inequality at the state as well as national level. The incomes of the country’s richest families have climbed substantially over the past two decades, while middle- and lower-income families have seen only modest increases. This trend is in marked contrast to the broadly shared increases in prosperity between World War II and the 1970s.
All About Sicily (1/19/06)
Strategically using General Purpose Statistics Packages: A Look at Stata, SAS and SPSSMitchell MN, Statistical Consulting Group UCLA Academic Technology Services Technical Report Series, December 15, 2005, Report Number 1, Version Number 1. Accessed on 2006-01-10.
(1/19/06)
Abstract
This report describes my experiences using general purpose statistical software over 20 years and for over 11 years as a statistical consultant helping thousands of UCLA researchers. I hope that this information will help you make strategic decisions about statistical software { the software you choose to learn, and the software you choose to use for analyzing your research data. www.ats.ucla.edu/stat/technicalreports/Number1/ucla_ATSstat_tr1_1.0.pdf
Researchers' Misunderstand Confidence Intervals and Standard Error Bars (1/19/06)
Psychological Methods
2005, Vol. 10, No. 4, 389–396 2005: 10(4); 389-396
Abstract
Little is known about researchers’ understanding of confidence intervals (CIs) and standard error (SE) bars. Authors of journal articles in psychology, behavioral neuroscience, and medicine were invited to visit a Web site where they adjusted a figure until they judged 2 means, with error bars, to be just statistically significantly different (p <.05). Results from 473 respondents suggest that many leading researchers have severe misconceptions about how error bars relate to statistical significance, do not adequately distinguish CIs and SE bars, and do not appreciate the importance of whether the 2 means are independent or come from a repeated measures design. Better guidelines for researchers and less ambiguous graphical conventions are needed before the advantages of CIs for research communication can be realized.
Use of the Extreme Groups Approach: A Critical Reexamination and New Recommendations
Psychological Methods
2005, Vol. 10, No. 2, 389–396 2005: 10(4); 178-192 Abstract
Analysis of continuous variables sometimes proceeds by selecting individuals on the basis of extreme scores of a sample distribution and submitting only those extreme scores to further analysis. This sampling method is known as the extreme groups approach (EGA). EGA is often used to achieve greater statistical power in subsequent hypothesis tests. However, there are several largely unrecognized costs associated with EGA that must be considered. The authors illustrate the effects EGA can have on power, standardized effect size, reliability, model specification, and the interpretability of results. Finally, the authors discuss alternative procedures, as well as possible legitimate uses of EGA. The authors urge researchers, editors, reviewers, and consumers to carefully assess the extent to which EGA is an appropriate tool in their own research and in that of others.
An Abductive Theory of Scientific Method (1/19/06)
Psychological Methods
2005, Vol. 10, No. 4, 371–388
Abstract
A broad theory of scientific method is sketched that has particular relevance for the behavioral sciences. This theory of method assembles a complex of specific strategies and methods that are used in the detection of empirical phenomena and the subsequent construction of explanatory theories. A characterization of the nature of phenomena is given, and the process of their detection is briefly described in terms of a multistage model of data analysis. The construction of explanatory theories is shown to involve their generation through abductive, or explanatory, reasoning, their development through analogical modeling, and their fuller appraisal in terms of judgments of the best of competing explanations. The nature and limits of this theory of method are discussed in the light of relevant developments in scientific methodology.
* Age-adjusted to the 2000 U.S. standard population using four age groups: <18 years, 18--44 years, 45--64 years, and >65 years.
† 95% confidence interval.
During 1999--2002, approximately 50% of females and 40% of males reported using at least one prescription drug during the preceding month, with non-Hispanic whites more likely to do so than non-Hispanic blacks and Mexican-Americans. In each racial/ethnic population, females were more likely than males to have used at least one prescription drug during the preceding month.
Sources: National Center for Health Statistics. Health, United States, 2005. Table 91. Hyattsville, MD: National Center for Health Statistics; 2005. Available at http://www.cdc.gov/nchs/hus.htm; National Health and Nutrition Examination Survey, 1999--2002. Available at http://www.cdc.gov/nchs/nhanes.htm.
Full article is here
(1/19/06)
The Use of Stories in Clinical Research and Health Policy (12/24/05)
JAMA
2005, Volume 294, 22,
2901-2904
Abstract
Physicians are immersed in stories. They hear stories from patients, tell them to other physicians, and recall them in quiet moments. Literary scholars, folklorists, and historians have long emphasized the importance of stories. In recent years, physicians trained in these disciplines have considered the role of stories in clinical practice. The physician-anthropologist Kleinman suggests that physicians need to move beyond "clinical interrogation" to listen attentively to their patients’ narratives of illness. Charon draws on her background in literary studies to suggest that the practice of medicine requires "narrative competence," which she defines as "the set of skills required to recognize, absorb, interpret, and be moved by the stories one hears or reads. She further proposes that physicians can enhance their clinical and emotional development through retelling clinical stories.
(12/24/05)
See the concrrent issue of Preventing Chronic Disease for several program evaluation-related papers, including:
-The Common Threads in Program Evaluation
-Using Concept Mapping to Develop a Logic Model for the Prevention Research Centers Program
-Race, Ethnicity, and Linguistic Isolation as Determinants of Participation in Public Health Surveillance Surveys
Einstein, Ethics and Science (12/21/05)
Journal of Academic Ethics
2005;
Volume 2, Number 4 Abstract
In celebration of Einstein's remarkable achievements in 1905, this essay examines some of his views on the role of “intellectuals” in developing and advocating socio-economic and political positions and policies, the historical roots of his ethical views and certain aspects of his philosophy of science. As an outstanding academic and public citizen, his life and ideas continue to provide good examples of a life well-used and worth remembering.
Reaching the Poor (12/19/05)
BMJ
2005, 331:1417 (17 December) Editorial
Need initiatives that reduce rather than exacerbate health inequities Globally around 1.2 billion people live on less than $1 a day. The erosive impact of poverty on their health and the link between ill health and poverty is well known. 1 But most interventions aimed at alleviating poverty and improving health in poor countries help the better off more than the most disadvantaged. 2 Such inequity of impact is often conveniently masked by expressing outcomes of evaluations as population averages, a flaw inherent even in the three health related millennium development goals. Now a report from the World Bank looks at the evidence and suggests how to reach the poor more effectively. 3 The data in the report, from 78 programmes on health, nutrition, and population conducted in 56 poor and medium income countries between 1990 and 2001, are sobering. Under-5 mortality was more than twice as high among the poorest fifth of the population than the richest fifth . . . [Full text of this article]
Socioeconomic Status in Health Research - One Size Does Not Fit All (12/14/05)
JAMA
2005; 294:2879-2888
Abstract
Problems with measuring socioeconomic status (SES)—frequently included in clinical and public health studies as a control variable and less frequently as the variable(s) of main interest—could affect research findings and conclusions, with implications for practice and policy. We critically examine standard SES measurement approaches, illustrating problems with examples from new analyses and the literature. For example, marked racial/ethnic differences in income at a given educational level and in wealth at a given income level raise questions about the socioeconomic comparability of individuals who are similar on education or income alone. Evidence also shows that conclusions about nonsocioeconomic causes of racial/ethnic differences in health may depend on the measure—eg, income, wealth, education, occupation, neighborhood socioeconomic characteristics, or past socioeconomic experiences—used to "control for SES," suggesting that findings from studies that have measured limited aspects of SES should be reassessed. We recommend an outcome- and social group–specific approach to SES measurement that involves (1) considering plausible explanatory pathways and mechanisms, (2) measuring as much relevant socioeconomic information as possible, (3) specifying the particular socioeconomic factors measured (rather than SES overall), and (4) systematically considering how potentially important unmeasured socioeconomic factors may affect conclusions. Better SES measures are needed in data sources, but improvements could be made by using existing information more thoughtfully and acknowledging its limitations.
Prescription Drugs: Out-of-Pocket Expenses and Unmet Need Relative to Family Income, 2002 (12/13/05)
Beth Levin Crimmel , MS and Marie N. Stagnitti, MPA Link: http://www.meps.ahrq.gov/papers/st102/stat102.pdf
Highlights
-Families without an elderly member were more than three times as likely not to have an out-of-pocket expense for prescription drugs (19.2 percent) than families with an elderly member (5.4 percent) during 2002.
-Families with an elderly member had prescribed drugs out-of-pocket expenses that accounted for greater than 5 percent of family income at a much greater rate (30.8 percent) than families without an elderly member (6.8 percent).
-Families who spent greater than 5 percent of their income on out-of-pocket expenses for prescribed drugs were more likely to report an unmet need (15.4 percent) than families who spent less than or equal to 5 percent of their income (6.3 percent).
-The majority of families with an unmet need during 2002 reported a financial cause as the sole or partial reason for the problem (66.3 percent).
Cost-Effectiveness in Individual Development Accounts (12/10/05)
Research on Social Work Practice
2006, Vol. 16, No. 1, 28-37
Abstract
Because resources are limited, the benefits and costs of social-work interventions—like all interventions—must be compared with the benefits and costs of alternatives. Evidence-based practice should ask, What works? How well does it work? And what does it cost? This article analyzes the provision of Individual Development Accounts (IDAs) with a new cost-effectiveness framework meant to help make assumptions and judgments explicit. In the specific IDA program examined, 1 month of services for 1 participant costs about $64. The mere existence of a cost figure— regardless of whether it is seen as high or low—has sparked many questions in the IDA community: How can costs be reduced without sacrificing quality? Which features of IDAs are essential? Are IDAs worth it? This sort of healthy questioning is precisely the purpose of cost-effectiveness analysis in social-work practice.
Explaining Educational-Related Inequalities in Health: Mediation and Moderator Models (12/3/05)
Social Science & Medicine
2006, Volume 62, Issue 2 , 467-478
Abstract
This paper studies how education and certain lifestyle factors affect people's self-reported health. In addition to the assessment of the effects of education and lifestyle, the study contrasts two models of explaining educational-related health inequalities: the mediation model and the moderator model. The mediation model posits that well-educated people's better health, as compared to the poorly educated, is caused by their more ‘healthy’ lifestyles. The moderator model suggests, by contrast, that the effects of the lifestyle variables on health are dependent upon educational level. Several analyses are carried out on two large data sets comprising of middle-aged men and women in two Norwegian counties, Rogaland and Nordland. Two main findings are presented: (1) Both education and lifestyle factors have the expected effects on health. (2) The results do not permit a clear-cut conclusion as to which of the two models of educational-related health inequalities should be preferred: whereas the results support the mediation model in the data from Rogaland, the moderator model is partially supported in the Nordland data.
Cross-Cultural Perspectives on Research Participation and Informed Consent (12/2/05)
Social Science & Medicine
2006, Volume 62, Issue 2, 479-490
Abstract
This study examined Portuguese Canadian and Caribbean Canadian immigrants’ perceptions of health research and informed consent procedures. Six focus groups (three in each cultural group) involving 42 participants and two individual interviews were conducted. The focus groups began with a general question about health research. This was followed by three short role-plays between the moderator and the assistant. The role-plays involved a fictional health research study in which a patient is approached for recruitment, is read a consent form, and is asked to sign. The role-plays stopped at key moments at which time focus group participants were asked questions about their understanding and their perceptions. Focus group transcripts were coded in QSR NUDIST software using open coding and then compared across cultural groups. Six overriding themes emerged: two were common in both the Portuguese and Caribbean transcripts, one emphasized the importance of trust and mistrust, and the other highlighted the need and desire for more information about health research. However, these themes were expressed somewhat differently in the two groups. In addition, there were four overriding themes that were specific to only one cultural group. In the Portuguese groups, there was an overwhelming positive regard for the research process and an emphasis on verbal as opposed to written information. The Caribbean participants qualified their participation in research studies and repeatedly raised images of invasive research.
From: http://www.word-detective.com/060704.html (11/27/05)
A "peeve" is something that annoys or irritates you, and since irritation is a highly individual emotion, one's "peeve" mileage may vary from one's neighbor's. I am "peeved," for instance, by people who assume that my license plates (which refer rather cryptically to books) mean that I spend every waking hour rooting for the Ohio State Buckeyes. Buckophiles, conversely, are probably peeved at the cool disdain with which I disclaim any pro-Buckeye sentiments.
For a word that expresses a universal (one presumes) human emotion, "peeve" is a remarkably recent coinage, first appearing in print as a verb only in 1908 and a noun (the thing that peeves) in 1911. Both "peeves," however, arose as what linguists call "back-formations" of the much older term "peevish," meaning "ill-tempered," that first appeared in the late 14th century. Back-formations, the derivation of a "root" word from a more complex form, are common in English -- the verb "to sculpt," for instance, was formed from the much older word "sculptor."
The precise derivation of "peevish" is uncertain, but it may be related to the Latin "perversus," meaning "reversed, perverse." The original meaning of "peevish" was simply "silly or foolish," but by about 1530 it had acquired the sense of "irritable, ill-tempered or fretful." Surprisingly, it then took several hundred years to develop "peeve" as the word for the irritating agent or action. "Pet peeve," meaning the one thing that annoys you more than anything else, first appeared around 1919. The "pet" (in the sense of "favorite") formulation probably owes its popularity and longevity to its mild perversity ("favorite annoyance" is a bit oxymoronic) as well as its snappy alliteration.
Stepwise Regression to Screen for Covariates (from: http://www.childrens-mercy.org/stats/weblog.asp) (11/27/05)
Someone wrote asking about how best to use stepwise regression in a research problem where there were a lot of potential covariates. A covariate is a variable which may affect your outcome but which is not of direct interest. You are interested in the covariate only to assure that it does not interfere with your ability to discern a relationship between your outcome and your primary independent variable (usually your treatment or exposure variable).
The writer offered up a couple of approaches. First, include all the covariates (but not the primary independent variable) in a stepwise regression model and then adjust your primary independent variable for those covariates which survive the stepwise regression. Second, include all the covariates and the primary independent variable in a stepwise regression model and then report the final model. If the final model fails to include your primary independent variable, that is just evidence that your primary hypothesis is negative.
The person who wrote in was well aware of the weaknesses of stepwise regression, but for those of you who are not familiar with those weaknesses, please read
What are some of the problems with stepwise regression?
which is a summary I made of comments about stepwise regression by Ira Bernstein, Ronan Conroy and Frank Harrell that were published on the email discussion list, stat-l.
The research community is gradually moving away from stepwise regression to other more sophisticated methods, but for now you can probably get a stepwise regression model published in most medical journals. Furthermore, there is no established method for how to use stepwise regression, so you are free to use any approach that is not totally outrageous. Here are some general comments, though.
First, if your goal is to assure that no confounding variables produce an incorrect relation between exercise and breast cancer, then the safest thing to do is to include all the potential covariates in the model and not worry about which ones pass some threshold for inclusion in the model. The drawback to this approach, of course, is that you lose a lot of degrees of freedom.
Second, never let a stepwise regression model violate your notion of common sense. If a particular covariate is known to be important (e.g., cigarette smoking in a cancer study) then exclusion of this covariate on the basis of a stepwise regression approach is a mistake. I like to think of stepwise regression as an intelligent assistant. It offers some help and guidance, but don't let it dictate the form of your final statistical model.
Third, never let stepwise regression bypass your primary research hypothesis. If a stepwise approach tosses out your primary independent variable, force it back into the equation anyway at the end, because you need to see the confidence interval and p-value associated with this variable.
Finally, as noted above, there are some new approaches that compete very well against stepwise regression in this particular situation. You should examine the use of propensity scores (which I hope to write an example for soon), as these offer all the advantages of including all possible covariates and none of the disadvantages. There is also a book by Frank Harrell on regression modeling approaches that is well worth reading.
Regression Modeling Strategies: With Applications to Linear Models, Logistic Regression, and Survival Analysis. Harrell FE (2001) New York, NY: Springer. ISBN: 0387952322. [BookFinder4U link]
Why Most Published Research Findings are False. Ioannidis JP. PLoS Med 2005: 2(8); e124. [Medline] [Full text] [PDF]
(A, worth-reading, discussion of power analysis and sample size.) (11/27/05)
Still Going Strong: Head Start Children, Families, Staff, and Programs in 2004 [link] (11/27/05)
Over the past 40 years, the Head Start program has delivered early education and support services to 23 million low-income preschool children and their families. In 1995, Head Start expanded to serve children from birth to age 3 and pregnant women through the Early Head Start program. In addition to early education, Head Start programs must provide children and families with access to a range of comprehensive services, including parenting resources, health screenings and follow-up, and social services.
Search Engine Use Shoots Up in the Past Year and Edges towards Email as the Primary Internet Application [link] (11/27/05)
Search engines have become an increasingly important part of the online experience of American internet users. The most recent findings from Pew Internet & American Life tracking surveys and consumer behavior trends from the comScore Media Metrix consumer panel show that about 60 million American adults are using search engines on a typical day. These results from September 2005 represent a sharp increase from mid-2004.
AIDS Epidemic Update: December 2005 [link] (11/27/05)
The annual AIDS epidemic update reports on the latest developments in the global AIDS epidemic. With maps and regional summaries, the 2005 edition provides the most recent estimates of the epidemic’s scope and human toll, explores new trends in the epidemic’s evolution, and features a special section on IV prevention.
Using Covariates to Improve Precision [link] (11/27/05)
The best way to measure the impacts of many important educational interventions is to randomly divide schools into a treatment group, which receives the intervention, and a control group, which does not — and then to compare future student achievement outcomes for the two groups. This paper examines how controlling statistically for baseline covariates (especially pretests) improves the precision of studies that randomize schools.
I'm Thankful to Social Security for Still Being There [link] (11/27/05)
Understanding the Basics of Managed Care [link] (11/27/05)
|
