Reply: Making the Most of Ideas, Facts, and Data About Age and Depression

Abstract

Normally, a critical comment begins by summarizing the results or conclusions being challenged. Then, each challenged finding or inference is addressed, citing counter-arguments and their supporting principles and evidence. Professor Angel's comments are not this direct. What, specifically, does he regard as implausible and unsupported? That depression is highest among the old and young, and lowest among the middle aged? That life-cycle differences in marital, employment, and economic statuses explain the association between age and depression? He clearly doubts that these findings demonstrate anything about the effects of age on depression, or that they are relevant to views of age as maturity, decline, life-cycle stage, survival, and historical trend. We disagree with his view. To the extent we can extract them, these seem to be Professor Angel's main points: (1) the article adds little to preexisting knowledge; (2) the model and analysis are too clever and complicated; (3) the measures are unreliable, imprecise, and invalid; (4) the cross-sectional data confound age with cohort, cause with selection, and outcome with process; (5) the model should include interaction terms and additional variables; and (6) quantitative models do not represent people's lives and feelings, so they do not mean much. Our responses follow. 1. Our analysis integrates and advances the ideas and observations of earlier studies. Prior studies took a variety of views, with apparently contradictory results. Feinson documented those contradictions in 1985. Newmann found a clue to an explanation of the contradictions in her meta-analysis published in 1989, which pieces together a U-shaped association from a number of age-limited samples. At a minimum, our results make the following worthwhile contributions to the literature: (1) depression among the elderly is not a scientific myth; (2) the U-shaped age curve, hypothesized based on Newmann's meta-analysis, is confirmed in two samples (one statewide and one national) that represent the whole range of adult ages (from 18 to 90); (3) the U-shaped curve does not disappear with adjustment for the fixed attributes of sex, minority status, and education; and (4) the U-shaped curve flattens with adjustment for social statuses that rise and fall over the lifetime-employment, marital and economic status. To our knowledge, no previous survey demonstrates the pattern of depression falling and then rising in successive age groups over the whole range of adult ages, and no previous analysis explains the pattern by adjusting for the links between age and depression. We also found clear results consistent with the proposition that some of the depression among the old reflects more difficult lifetime conditions associated with lower average education. We found only inconsistent support for the proposition that survival differences across status groups produce some of the age curve, possibly because survival effects are small until old age. To our knowledge, all of this is new. Professor Angel cites no evidence that it is not. 2. Our model and analysis are as clever as they need to be and as simple as they can be. When thinking about the different views and results of earlier studies, we realized that an analysis of depression by age must do several things. It must compare adults across the whole range of ages. It must adjust progressively for ascribed status, the stable trait of education, the main social statuses contingent on age, and the spiraling dysfunction and demoralization that might plague the very old. It must consider carefully the form of the age curve and the changes with specific adjustments predicted by each view of age. To do that, it must distinguish the falling from the rising component, and specify an