Goal
This guide will examine trends in attitudes regarding the rights of homosexuals in the United States from the early 1990s through 2007. Frequencies, crosstabs, and comparison of means tests will be used for the analyses.
Concept
The issue of same-sex marriage is just one of several that help to illustrate the ongoing struggle of homosexuals to claim equality in civil rights. Part of the problem faced by homosexuals rests in levels of public support for certain legal rights for homosexuals. We will consider attitudes toward three different rights: same-sex marriage; legal adoption of a child by homosexuals; and homosexuals being allowed to teach in public schools.
If attitudes toward homosexual rights vary according to age, especially if it appears that there is a linear relationship, then it is possible that there is either a cohort effect (people born about the same year having similar attitudes) or a learning/socialization effect (attitudinal change over the life course). Because this dataset includes many successive samples (surveys taken annually from 1982-2007), we may be able to distinguish between a cohort effect and a learning effect. Furthermore, by including year of survey as a variable in our analyses, we will be able to see if there are any general trends in attitudes toward the rights of homosexuals.
Examples of possible research questions about attitudes toward the rights of homosexuals:
This exercise will use the Houston Area Survey, 1982-2007 (Cumulative File). The survey is a longitudinal study that began in 1982 and has been administered annually since. The interviews conducted for this research were designed to measure perspectives on the local and national economy, on poverty programs, and on inter-ethnic relationships. Also captured were respondents' beliefs about discrimination and affirmative action, education, crime, health care, taxation, community service, and their attitudes toward abortion, homosexuality, and other aspects of the social agenda. The researchers were also concerned with issues that are specific to Houston and other urban areas, such as respondents' assessments of downtown development, mobility and transit, land-use controls, and environmental concerns.
This exercise will use the following variables:
To begin this exercise, examine the attitudes toward the three measures of homosexual rights.
First, look at the frequency tables for
Age and Attitudes about Gay Rights
Next, think about the relationship between the three measures of attitudes toward rights and
age. To aid in the analyses, we have recoded the age variable into a new variable called "AGE4"
which groups the respondents into four age categories: 18-34; 35-49; 50-64; and 65-95. Now look
at the crosstab of
When you look at the crosstabs for
Attitudes about Gay Rights 1991 - 2006
Also consider whether or not people have become more supportive of the rights of homosexuals
over the past 15 years or so. Examine the crosstab of
Next look at the crosstab of
Finally, consider support for homosexuals teaching in public schools. You can conduct both a
crosstab and a comparison of means test. First, in the crosstab of
Attitudinal Change among Age Cohorts?
Because this dataset contains a series of successive samples, it is possible to test whether any observed differences among the age groups and across the years are cohort effects (people born about the same year having similar attitudes) or a learning/experience effect (attitudinal change over the life course). A cohort is a group of people born around the same time. Thus, a cohort born in 1942 would have been 49 years old in 1991, and 64 years old in 2006.
To create the cohort variable, we computed the respondent's birth year by subtracting the
respondent's age from the year the survey was taken. Next, we grouped respondents into age
cohorts that roughly corresponded to the age groups that we have been using (18-34, 35-49, 50-64,
and 65+) in 1991 and 2006. By doing this we are hoping to look at how these cohorts have changed
over time with regard to their attitudes toward gay rights. The age cohorts do fit our age groups
perfectly in 2006, but because different questions were asked in different years, the overall fit
of age group to the year the questions were asked is not perfect. Still, it is a reasonably close
approximation. The cohort variable is called
Consider the results of a
To be consistent, look at two additional comparison of means tests. First, a test of
Think about your answers to the application questions before you click through the interpretation guide for help in answering them.
Gay Rights
Gay Rights and Age
Attitudes about Gay Rights 1991 - 2007
Attitudinal Change among Age Cohorts
Interpretation
Things to think about in interpreting the results:
It is always important to consider how missing data might affect the ability to make inferential statements about the population from which the sample was drawn. Large amounts of missing data can limit our ability to generalize sample results. Even small amounts of missing data can affect the results, especially if missing responses are not randomly distributed across categories of the independent (grouping) variable.
Weights (mathematical formulas) are often used to adjust the sample proportions, usually by race, sex, or age, to more closely match those of the general population. The analyses used in this guide did not use any weights, which may reduce the generalizability of the findings, but the resulting tables are accurate descriptions of the relationships found between these variables among these respondents.
This set of analyses is based on a set of successive samples taken annually from the residents in and around the city of Houston. As such, any generalizations should be limited to the Houston area except in situations where we can be certain that the Houston area is reflective of some larger population. In order to make general statements about a larger population, some form of probability sampling must be used such that all members of the population must have some known chance of being included in the sample. Because the sample is drawn from Houston area residents, people living in and near other cities, towns, and counties in the U.S. are not part of the sampling frame, therefore, we should not generalize the findings from these samples to a population larger than the Houston area.
Reading the results:
Briefly, the comparison of means calculates the mean of a dependent variable within categories of the "independent" variable(s). The t-statistic measures whether the mean in a cell is larger or smaller than the overall mean. You can also think about it as how far the observed value in the cell is from the expected value in the cell (the sample mean) with the distance being measured in number of standard errors. The t-statistic also takes into account the total number of cases (n) in a cell. If there are only a few cases in the cell, then the deviation from the overall mean is not as significant as if there are many cases in that cell. Remember, when the absolute value of the t-statistic is large, the probability that you would obtain that value if the null hypothesis were true is small. That is, the probability that you would get the large t-statistic due to sampling error alone is small.
Frequency Distribution tables (crosstabs) display percentages in each cell of the people who fall into the overlapping categories (if two variables are used), followed by the 95% confidence interval (in parentheses, where used) for the expected percentage in the population, followed by the actual number of respondents who fall into the overlapping categories (cells). The coloring in the tables demonstrates how the observed numbers in each cell compares to the expected number if there were no association between the two variables. Paler colors represent lesser significance of the difference between observed values and expected values. The accompanying bar charts display the patterns visually as well.
The analyses show the following:
The frequency tables show that there is far greater acceptance of the right of homosexuals to teach in public schools (63.2%) than of the right to marry (38.5%) or of the right to adopt (34.6%).
From the lower right cell of the table of GAYMARRY by AGE4, in the "agree" row and the "1" (18-34 year old) column, we see that 48.8% of the youngest respondents believe that same-sex marriages should be afforded the same legal status as heterosexual marriages.
In general, older respondents are less likely to agree that homosexuals should have any of these three rights. When we look at the three tables, we see that the lowest percentage of support for these rights is always in the column associated with the oldest respondents. Still, we cannot be sure that this observed difference between 50-64 year olds and 65-95 year olds is a difference that is likely to be found in the population of interest. In other words, we do not know if we can try to generalize to the larger population. Fortunately, we can use confidence intervals to help us decide. A 95% confidence interval can be constructed around each observed value (percentage) in the table using knowledge of the normal distribution (or t-distribution), standard error, and the sample size. For these tables, the width of the confidence interval will be the same for both cells in the same column. In order to state that people 65 and older are the least supportive of gay rights in the Houston area, we want to be at least 95% confident that this statement is true and is not merely a description of the sample results. We can achieve this level of confidence if our confidence intervals do not overlap.
Using the GAYTEACH by AGE4 table as an example, the confidence interval in the "For" row and the "50-64 year old" column suggests that we are 95% confident that the true population percentage of support among 50-64 year olds in Houston during the years the question was asked is between 57.5% and 66.1%. The confidence interval in the same row, but the "65+" column suggests that we are 95% confident that the true population percentage of support among persons 65 and older during the years the question was asked is between 46% and 56.5%. We can see in this table, as well as the other two tables that the confidence intervals in columns 3 and 4 do not overlap, suggesting that we are at least 95% confident that persons 65 and older are the least supportive of any age group of these three right of homosexuals in Houston between 1991 and 2006.
Respondents show the greatest support for gay marriage in 1999 (43.9%) and the lowest support in 1997 (32.1%). We do not have to consider the confidence intervals in order to make these statements because the question is directed at the sample results and not attempting to make a claim about the larger population of interest. Ultimately, we do not see any pattern of increased or decreased acceptance of gay marriage over these years.
In reviewing the table of GAYADOPT by YEAR, there is a reasonably clear pattern of increasing support for the right of homosexuals to adopt children despite fact that the majority of respondents oppose this right. Support increased from 20.4% in 1991 to 46.2% in 2006. Interestingly, this support does not seem to be a gradual shift. Between 1991 and 1996, we see a 12% increase in support followed by a slight decrease in the two successive years the question was asked. Then we note the 12.1% increase in support between 2000 and 2002 followed by slight increases in support in the following years.
Support for gay teachers in public school increased from 53% in 1992 to 73% in 2006. The t-statistics of -5.2 associated with the 53% and 5.6 associated with 73% suggest that this observed 20% difference does appear to be statistically significant, thus allowing us to make the more general statement about the population. It does appear that Houston area residents have become more tolerant of homosexuals teaching in public schools.
We cannot definitively say anything about changes in support of same-sex marriage in Houston between 1993 and 2007 because the question is asking us to make an inference about the population (Houston) and is not just a description of the sample results. The sample results show that there are some rather slight differences among the cohorts from 1993 to 2007, with the younger cohorts showing decreases in percentage of support and the older cohorts showing increases in percentage of support. However, when we scroll down the output page and look at the confidence intervals, we see that the confidence intervals across each row exhibit some overlap. The overlap suggests that the observed differences across the age cohorts between 1993 and 2007 are not large enough to reject the null hypothesis that there has been no change.
The observed changes in support for gay adoption and gay teachers in public schools seems to be a combination of actual increased support and younger (born later) respondents. From both tables, we note that each age cohort exhibits increases in support for both of these rights and looking at the confidence intervals, there is no overlap across the rows for the middle three age cohorts. The oldest and youngest cohorts do show an overlap, but this is related to the low n (number of respondents) in at least one of the cells across those rows. The lack of overlap does suggest increased support, but we cannot ignore the fact that the youngest age cohort (the cohort that is clearly most supportive of gay rights) comprises a far greater proportion of the sample in 2006 than in 1991-92.
Summary
The goal of the guide was to examine trends in attitudes regarding the rights of homosexuals in United States from the early 1990s through 2007. We looked specifically at attitudes toward three rights: same-sex marriage; adoption by homosexual parents; and a homosexual's right to teach in a public school. Our primary concern was whether attitudes vary by age. We also hoped to understand if any observed differences in attitudes across age groups were a function of chronological age or age group cohort. In other words, are age differences in attitudes something that is learned over the life course (i.e. "people become more conservative in their thinking as they get older") or are attitudes more or less fixed at an early age (i.e. "people are a product of their time").
The results in this guide are mixed. If we consider support for gay adoption and gays teaching in public schools, we see that there is about a 15%-20% increase in support from the early 1990's through 2006 across all age cohorts suggesting that there are changes as people get older. However, there are still marked differences between age cohorts, suggesting that older age cohorts are less tolerant of gay rights and this is likely to be a product of their upbringing.
The real conundrum here is same-sex marriage. The results demonstrate that this is a very different issue than either of the other two rights. There seems to be no appreciable change in the level of support between 1993 and 2007 despite the presence of more respondents born later in the 20th century in the later sample. Future research may be able to address if this pattern is reflected in other parts of the United States and if so, why is the issue of same-sex marriage different.